ThesisPDF Available

Role of Automation in PPC Management: Google Ads Smart Bidding Strategies and Opportunities for third-party solutions


Abstract and Figures

This study focused on the role of automation and prediction about the future of a specific type of online advertising - pay-per-click advertising (PPC). The PPC advertising, also known as cost-per-click advertising is a competition-based charging method, which is a click-based purchase model for advertisers and charges them for only clicks that reflect the visits to the links provided by the advertiser. An amount that advertisers pay is calculated by predetermined factors by search engines, which are related to the bidding of competitors and quality of content published by the advertiser. After the rise of online advertising, companies attempt to target specific audiences with PPC advertising solutions on search engines by using detailed targeting options. PPC did not help companies only on search engines, but also served as a billboard on various websites that are Display Network’s partners like Google Display Network, Yandex Ad Network etc. Advertisers got the benefit of using contextual advertising and getting rare audiences with specific targeting features. Additionally, businesses were able to make data-driven marketing decisions by analysing historical data and switched from old school intuition-based decisions to data-driven decisions in this new digital marketing era. From the beginning of the rise of internet or online marketing, Pay-Per-Click become an actor in main marketing activities, merging 4 sides - PPC vendors (Google, Bing etc.) advertisers (businesses promoting their services and products), publisher websites who act as placements for display ads, internet users who are potential buyers and seek for information about products or services in different intent cycles like in sales funnel. In non-display ads, search engines themselves act as publishers of sponsored ads. Therefore, advertisers focus on focus keywords of internet users and other variables that affect the conversion probability of them, such as time, device category, household income, gender information provided by the advertising platform. As the advertising effectiveness depends on several factors, growing large PPC accounts become hard to manage and optimize. In that stage, automated optimization and management solutions become economically and practically available. Increasing the efficiency requires human intuition and routine optimization tasks for PPC managers in manual bidding. Although manual bidding gives more control on the PPC campaigns, larger accounts create larger challenges when managing thousands of keywords and ad groups. Additionally, human intelligence lacks to find all correlation among different variables for several campaigns at the same time. In order to decrease this guesswork many PPC management tools offered by third parties and Google Ads introduced its solutions to increase the advertiser experience. Both PPC agencies, which serve companies in the optimization, reporting, management of the advertising campaigns, third-party software companies and PPC platforms (Google, Bing etc.) introduce their unique solutions to manage PPC accounts. This study investigates the effect of automation solutions introduced by Google Ads, analyzes the results of using Google Ads Smart Bidding Strategies and the marketplace of PPC Automation, and gives a future prediction about the possible changes in the way of automation with non-empirical research.
Content may be subject to copyright.
Kamal Allazov
Corvinus University of Budapest
Corvinus Business School
International Study Programs
Role of Automation in PPC Management:
Google Ads Smart Bidding Strategies and
Opportunities for third-party solutions
Kamal Allazov
MSc. in Marketing
Thesis Supervisor: Sárvári Balázs
Table of Contents:
1. Introduction ..................................................................................................................... 1
1.1 Background .............................................................................................................. 1
1.2 Research Objective ................................................................................................... 2
1.3 Methodology ............................................................................................................ 3
1.4 Thesis Structure ........................................................................................................ 3
2. Literature Review ........................................................................................................... 5
3. Theoretical Framework ................................................................................................. 12
3.1 Key Performance Indicators in PPC Advertising ................................................... 12
3.2 The AIDA Funnel approach for PPC advertising .................................................. 21
4. PPC Automation ........................................................................................................... 31
4.1 Importance of Automation in Digital Marketing ................................................... 31
4.2 Management of PPC Advertising: ......................................................................... 34
4.3 Automation of PPC management ........................................................................... 35
4.4 Automation of PPC Management on Google Ads ................................................. 37
4.5 Smart Bidding Strategies: ...................................................................................... 44
5. Third-party Automation Tools ...................................................................................... 60
5.1 Current situation and opportunities ........................................................................ 60
5.2 Impacts of Automation for different players in the industry: ................................ 65
6. Conclusion .................................................................................................................... 68
7. Limitations and Further Research ................................................................................. 69
8. References: .................................................................................................................... 71
Table of Figures:
Figure 1:Influence diagram of the keyword bidding model. (Küçükaydin et al., 2019) ...... 16
Figure 2:EKB model of sales funnel ..................................................................................... 28
Figure 3:Contextual signals on Google Ads ......................................................................... 44
Figure 4:Top ad positions on Google Ads search campaigns ............................................... 45
Figure 5:Screenshot by Lyubomir Popov to prove that Search Impression Share, Search Top
Impression Share and other related metrics are not updated real-time. ................................ 48
Figure 6: Case study results from ............................................... 54
Figure 7: Case study results from ......................................................... 54
Figure 8:Case study results from ................................................ 56
Figure 9: Relevant KPIs for each Smart Bidding Strategy ................................................... 58
Figure 10: Lifecycle of Smart Bidding Strategies ................................................................ 58
Figure 11:Players and roles in the PPC industry by Li et al. (2016) .................................... 65
1. Introduction
1.1 Background
This study focused on the role of automation and prediction about the future of a specific
type of online advertising - pay-per-click advertising (PPC). The PPC advertising, also
known as cost-per-click advertising is a competition-based charging method, which is a
click-based purchase model for advertisers and charges them for only clicks that reflect the
visits to the links provided by the advertiser. An amount that advertisers pay is calculated by
predetermined factors by search engines, which are related to the bidding of competitors and
quality of content published by the advertiser.
After the rise of online advertising, companies attempt to target specific audiences with PPC
advertising solutions on search engines by using detailed targeting options. PPC did not help
companies only on search engines, but also served as a billboard on various websites that are
Display Network's partners like Google Display Network, Yandex Ad Network etc.
Advertisers got the benefit of using contextual advertising and getting rare audiences with
specific targeting features. Additionally, businesses were able to make data-driven marketing
decisions by analysing historical data and switched from old school intuition-based decisions
to data-driven decisions in this new digital marketing era.
From the beginning of the rise of the Internet or online marketing, Pay-Per-Click becomes
an actor in main marketing activities, merging four sides - PPC vendors (Google, Bing etc.)
advertisers (businesses promoting their services and products), publisher websites who act
as placements for display ads, internet users who are potential buyers and seek for
information about products or services in different intent cycles like in sales funnel. In non-
display ads, search engines themselves act as publishers of sponsored ads. Therefore,
advertisers focus on focus keywords of internet users and other variables that affect the
conversion probability of them, such as time, device category, household income, gender
information provided by the advertising platform. As the advertising effectiveness depends
on several factors, growing large PPC accounts become hard to manage and optimise. In that
stage, automated optimisation and management solutions become economically and
practically available.
Increasing efficiency requires human intuition and routine optimisation tasks for PPC
managers in manual bidding. Although manual bidding gives more control of the PPC
campaigns, larger accounts create larger challenges when managing thousands of keywords
and ad groups. Additionally, human intelligence lacks to find all correlation among different
variables for several campaigns at the same time. In order to decrease this guesswork, many
PPC management tools offered by third parties and Google Ads introduced its solutions to
increase the advertiser experience. Both PPC agencies, which serve companies in the
optimisation, reporting, management of the advertising campaigns, third-party software
companies and PPC platforms (Google, Bing etc.) introduce their unique solutions to manage
PPC accounts. This study investigates the effect of automation solutions introduced by
Google Ads, analyses the results of using Google Ads Smart Bidding Strategies and the
marketplace of PPC Automation, and gives a future prediction about the possible changes in
the way of automation with non-empirical research.
1.2 Research Objective
As the online advertising market changes and is influenced by automation, the usage of its
sub-sectors, such as PPC is also impacted by contemporary business solutions.
The research objectives are as follows:
1. Understanding the main role of PPC advertising as an online advertising channel in the
customer journey.
2. Understanding the automated solutions for PPC advertisers to maximise their return on
investments, investigating how automated bidding strategies work and are there any other
alternatives provided by third-parties by applying non-technical theoretical analysis.
3. Predicting the possible future outcomes of implementing automation solutions for industry
players - PPC managers or agencies, businesses or advertiser companies, ad platforms (in
case of Google Ads), online users (as an affected audience).
1.3 Methodology
The research is mainly focused on the description and prediction about the central theme by
mainly using non-empirical analysis with theoretical aspects of the topic. Collected data
includes the case studies as secondary data to examine the current implementations in the
market, and the analysis of the data has been performed according to the related literature
review for deriving strong and relevant conclusions. The content analysis method is used to
identify, analyse and report patterns in the theme. The fact-finding method is used to analyse
current circumstances. The primary object of descriptive study is to describe events,
phenomenon and circumstances based on observation and other sources.
The main weakness of the method is that it does not include primary empirical data collected
from advertisers who use this advertising channel. Collected secondary data and existing
industry use cases were provided by advertising platforms and agencies which aim to
promote and attract advertisers, which means that overall performance evaluation is biased.
Additionally, further research results with primary data retrieved from different advertisers
may argue against the effectiveness of using Smart Bidding Strategies by Google Ads.
Limitations occur because of the data security and confidentiality concerns by businesses that
are unwilling to share information. However, the general overview provides insights into the
system and the shaping industry.
1.4 Thesis Structure
After the introduction chapter, Chapter 2 introduce literature review and theoretical studies
regarding PPC advertising and optimisation. Chapter 3 continues with the theoretical
framework, listing main KPIs which were discussed in previous studies. Chapter 3 introduce
AIDA sales-funnel approach to allocate KPIs to different phases of purchase behaviour of
online users while including frameworks applied to optimise the results of PPC campaigns.
Chapter 4 briefly discusses contemporary Machine Learning and Artificial Intelligence in
Digital marketing and introduce PPC management processes to automate. Then Chapter 4
continues with Automation of PPC management by discussing the main concerns and
problems, and then introduce solutions by Google Ads PPC platform - the Smart Bidding
Strategies (also known as Automated Bidding Strategies with its previous name). This part
includes secondary data both from blogs (case studies, experiments by PPC managers and
agencies) and from authorised sources, such as Google marketing research blog - Chapter 5 discusses opportunities for third-party services by giving
examples from current solutions in the industry and also propose suggestions.
2. Literature Review
This chapter includes a review of the literature and theoretical studies regarding PPC
advertising. The PPC was defined as advertising which is based on competitive bidding
among advertisers through search engine advertising platforms (Nunan and Knox, 2011). As
described as keyword advertising by Liu et al. (2010), one of the main advantages of PPC
makes advertisers able to target audiences based on their search intent on search platforms
via leveraging sponsored ads. Therefore, it's also known as search engine marketing (Jansen
and Schuster (2011) because of searching and finding relevant information about products or
services. Although there are click fraud incidents in PPC history, also known as a "hit
inflation, search engines are still trustworthy for companies to publish ads and acquire
potential customers. Advertisers pay when their ad gets clicks rather than impressions. (Farris
et al., 2010) This charging method was first introduced by Overture, which was previously, and this conception first was offered at a conference in 1998 in California (Ellam,
2004). The competitor brands, such as Google, Yahoo! also implemented the same model
and got billions of dollars from this advertising concept (Dellarocas, 2012)
Liu et al. (2010) classified two categories in keyword advertising contextual advertising
and sponsored links. Sponsored links are text ads on the search engine results which trigger
certain targeted search terms used by search engine users when they look for specific
information and the sponsored text ad including website landing page URL, as stated by (Das
Sharma et al., 2012), appears on the particular position of the results absolute top, top and
bottom of the page. The auction system determines the position of each ad based on several
factors, such as ad content quality, relevancy, bid (Chen et al., 2007, Katona & Sarvary 2010).
If one of the Quality score determinant - the expected CTR is less than or equal to the ratio
of the ad price to the cost per click on an advertising network, the advertiser is always
recommended to prefer pay-per-view bidding strategy for advertisement campaigns. (Kwon,
2011). Matz, Kosinski, Nave, and Stillwell (2017) found that for the effective outcomes from
psychological targeting in the contextual advertising convincing contents matched to the
psychological personas of a large segment of online users produced up to 50% more online
sales and 40% more clicks than their non-customized or mismatched equivalents. Overall,
PPC campaigns were found effective to increase sales and traffic through search engines and
display advertising networks.
Additionally, some researches showed the higher importance of paid search ads in the
increase of conversion rates and customer lifetime value in comparison to other advertising
types. (Chan, et al., 2011; Rutz & Bucklin, 2011). Practical results from the research by Chan
et al. (2011) showed that customers obtained from Google search advertisements had higher
transactions rather than those obtained from other advertising channels. Research also
included future sales data and the value of the new purchasers. This practical framework also
supported the notion that paid search advertising is a productive, effective and powerful tool
to increase long-term profit with certain investment in search advertising.
Paid search advertising was determined as having a more positive impact on increasing sales
in comparison to offline advertising. It has an influential role in the purchase decision of
online users in the stage that is close to purchasing and provides enhanced targeting features
to get purchases. In addition to paid search advertising, display advertising has a long-term
effect on the market value of the firm. (Bayer et al., 2020).
As search advertising targets users who have previously reflected their interest in the certain
product or services by searching for a related search keyword on a search engine (Abou et
al., 2012; Sayedi et al., 2014), it was focused to analyse the effects of PPC advertising in the
sales funnel concept by several types of research.
The effectiveness of PPC in buying funnel models was evaluated in order to analyse
consumer interaction and experience with search advertising campaigns on search engines
during the purchase decision and purchase process. It revealed the effectiveness of using a
funnel model to classify search keywords and target different query types based on search
intent in buying funnel stages (Jansen et al., 2011). This research also found that online
purchase behaviour by using search engines does not exactly fit and represent the exact
buying funnel stages in real life, and consumer actions did not follow the pattern as predicted.
However, it was still determined as having an impact on the final purchase decision by
capturing a certain segment of users and educating them in the process of the purchase
decision. Goldfarb & Tucker (2011) stated that paid search advertising could show ads at the
time, which is closer to the purchase decision and is prone to impact it. Paid search
advertisements generate clicks and user engagement with the website of the promoted brand,
which is uncommon and unavailable in offline advertising. Another research (Kim et al.,
2019) also analysed the same buying funnel model in the example of shoe brands Nike and
Adidas in Korea and found the different patterns in user behaviour for leader and follower
brand in terms of following funnel stages. This difference points out the difference in brand
loyalty, product awareness and other factors that affect how to act differently during the
purchase decision. In that case, manual optimisation and bidding become ineffective in case
of lacking expertise and data analysis when finding unique patterns and correlations for
certain stages of buying funnel while several contextual signals changing the results and
affecting outcome.
The effective use of PPC depends on the right bidding and sophisticated targeting in order to
capture more relevant audiences on the Internet-based on the behaviour and interests of
certain products and services. Paid search campaign management includes some tasks that
must be performed regularly: (1) selecting relevant keywords, (2) setting and changing the
bids for keywords, and (3) creating the advertising content (Rutz et al., 2012).
Selecting relevant keywords can be classified as keyword research, structuring keyword
categories and match types. Du et al., (2017) proposed that it is necessary to distinguish
among the different bidding strategies for keyword categories and match types as the
financial performance, such as the number of sales, sales profit and return on investment
differs for the keywords. This research classified keywords as generic-relevant (keywords
that are associated with product or service), focal-brand (keywords that are associated with
the name of the advertiser firm), and competing-brand (keywords that are associated with
trademarks of competitors in the industry) and the results suggest creating a strategy for each
of these keyword groups when working on the PPC management strategy.
The literature admitted the fact of a positive impact of advertising on the ability of consumers
to use branded search terms associated with adding content after being displayed to display
ads (Fulgoni & Mörn, 2009; Nguyen et al., 2010; Papadimitriou et al., 2011), social media
ads (Mukherjee & Jansen, 2017; Pashkevich et al., 2012), sponsored search ads (Rutz &
Bucklin, 2011; Rutz et al., 2011), and TV ads (Joo et al., 2013; Joo et al., 2016; Zigmond &
Stipp, 2010).
The matching option of keywords helps advertisers to target specific search terms and filter
them based on the function introduced by the search engine marketing platform. Match
options are broad, phrase and exact match. (Bartz et al., 2006; Gupta et al., 2009). Broad
match offers general matching of keywords with their original ones and synonyms. For
example, if advertisers target those who search for "car dealers" the broad match keyword
"car sales" (without quotation marks on the system) will be triggered by both "car dealers"
and "car sales" and an advertisement will appear on search results. When using phrase match
with quotation mark the words and its close variants, such as misspellings, plural forms,
acronyms etc. should be in the search terms with before or after any additional search term.
For example, if advertisers use "BMW price" users who typed "new BMW prices" will see
the advertisement on the results because the phrase "BMW prices" were used in the search
term and the plural form of the word "price" is a close variation of the given keyword. When
it comes to exact matching, this option does not let any additional words before and after the
targeted keyword. For example, if advertisers use "car prices", the search term "new car
prices" will not trigger this advertisement content. This restriction helps advertisers to target
the most specific words without showing an advertisement to similar searches.
Using broad match and phrase match keywords may negatively affect the Clicks/impression
ratio, which is called Click-through rate, and eventually, the quality of traffic can be low
because of irrelevance. Using exact match keywords may improve click-through rate and
show ads more specific searches rather than general search terms. (Klapdor et al., 2014)
In another study, it is stated that in comparison to broad match, a phrase or an exact match
more relevant match with the consumer's search intention. (Agarwal et al. 2011).
(n.d.). About keyword matching options - Google Ads Help. Retrieved May 10, 2020, from
Since Anderson (2006) brought the popularity to the term "long-tail", agencies, blogs and
advertisers began to promote this notion claiming that search campaigns are driven by the
long-tail keywords which refer to three or more search keywords typed on the search engine.
Advertisers had strong motives to argue that long-tail keyword mining is crucial in PPC
management to find new conversion opportunities, and this created additional tasks for PPC
Jerath et al. (2014) found that click activity of online users is low and heavily focused on the
organic results. Additionally, a larger portion of clicks is associated with less popular
keywords (i.e. keywords having lower search volume), indicating that in comparison to more
popular ones, less popular keywords drive more attention on information search and
consumers are closer to purchase with these search keywords. This study did not focus on
long-tail keywords but mentioned the importance of choosing less popular keywords which
requires additional effort and time in PPC management. However. Skiera et al. (2010) found
that the top 100 keywords used in each campaign brought on average, 81.40% of all clicks
and 88.57% of all searches. Results showed that advertisers should focus on finding the best
100 keywords and do not need to concentrate more on long-tail keywords. (Skiera et al.,
Keywords advertising requires to understand the interest of a "searcher" from the search
terms they are using when searching for information or products or services via search
engines. Many keywords may have several meanings. Therefore, additional phrases are
required to include or exclude within the keyword list to specify. In this case, negative
keywords act as an exclusion role to ignore irrelevant search terms. Using negative keywords
reduces the risk of lower purchase rates and loss of ranking in the auction. (Scaiano et al.,
2011, September).
As more data is available in digital marketing and collected on search engine advertising
platforms and web analytics software, marketers become able to analyse historical data to
optimise their future campaigns with more data-driven strategic decisions. By processing
large qualitative and quantitative data, companies switched to a user-centred approach by
building personas (Schäfer et al., 2014). Even some researchers criticised creating personas
in order to avoid stereotypes (Turner & Turner, 2011).
Especially in PPC advertising using contextual signals and identifying behavioural patterns
of online users contribute getting better results after sophisticated targeting and data-driven
strategic decisions in terms of budget allocation and determining bids for different keywords,
audiences and other contextual signals. The patterns represent personas and create a fictional
character whose background information help advertisers to build a plan and understand the
audience by developing empathy. Lee et al. (2020) proposed that using statistical analysis
helps to derive profiles of customers, and this can be used to market products and services to
analysed persona segments. In this research this segment was millennials and findings
showed the importance of deriving personas on a deep understanding of target users, their
behaviours and needs, which can support improvements on the customer experience of
products or services offered by the company. The advantages of using buyer personas have
been stated in some studies (Goodwin, 2008; Miaskiewicz & Kozar, 2011)
The Google patent on "Targeted Advertising based on user profiles and page profile"
(Haveliwala et al., 2012) explains (at a high level) the way Google provides personalised
Personalised advertising is available on Google Ads because Google analyses the networks,
especially website content and user activity on Google Display Network (GDN). The content
of each website is evaluated, and the Content Analysis module categorises topics and makes
advertisers able to target based on content topics.
Google also builds online user profiles by analysing the behaviour through the web and
learning the search activity. This monitoring enabled Google Ads to target audiences of users.
DoubleClick cookie tracking collects user data, such as visited websites, geo-location of the
users, the content category (topics) of these websites, time duration spent etc. and all this
information creates a user's "interest" representing the user profile. (Haveliwala et al., 2012)
(n.d.). About topic targeting - Google Ads Help - Google Support. Retrieved May 10, 2020, from
Google Ads uses cookies to track user behaviour after clicking an ad and follow further
interaction with other ad contents through internet and website activities of this user.
user refers to the browser because if this person switches to a different device, the same
person will be considered as a different user technically. However, if a person signed-in with
a Gmail account, Google identifies and matches this user ID. This enables advertisers to
target the same audience across multiple browsers in different devices and expand the reach.
Advertisers should comply with personalised advertising policies when implementing cross-
device linking, and it is forbidden to share any Personal Identifiable Information via Google
remarketing tag and any data feed related to ads.
As the contextual signals and variety of parameters help advertisers determine the
behavioural patterns of the targeted users, it makes it available to analyse large datasets and
find the relationships, optimise campaigns based on the historical data and modify
advertising messages for a better outcome. Yang et al. (2019, July) investigated the
automated bidding strategies deployed in the Baidu search engine and admitted that the
optimisation or manual selection of best performing is difficult for advertisers. Additionally,
several studies were conducted on Real-time bidding to optimise revenue (Li et al., 2018;
Qin et al., 2017; Qin et al., 2019), bid optimisation (Broder et al., 2011, February; Even et
al., 2009) and bidding strategies (Zhang et al., 2012) for better budget allocation between
keywords and for a higher return on investment. Zhu et al. (2017) introduced optimised CPC
to adjust bids automatically for better results with high-quality traffic in the Taobao platform.
However, none of the researches discussed the effectiveness of Smart Bidding Strategies of
Google Ads and opportunities for third-party PPC management tools, which are used by
agencies to optimise PPC campaigns through multiple advertising accounts. This study aims
(n.d.). Cookie: Definition - Google Ads Help - Google Support. Retrieved May 10, 2020, from
(n.d.). About Advertiser cross-device linking - Google Ads Help. Retrieved May 10, 2020, from
(n.d.). Personalized advertising - Advertising Policies Help - Google .... Retrieved May 10, 2020, from
to fill the gap with theoretical and secondary data sources by using the descriptive analysis
to find opportunities in the industry.
3. Theoretical Framework
This chapter continues with the applied theoretical frameworks and related works in order to
understand the role of PPC in online advertising and to create a detailed understanding of the
business KPIs with the sales-funnel approach.
Search Engine Advertising was described as the main revenue source for search engine
platforms (Jafarzadeh et al., 2015) and is also known as keyword advertising, search
advertising, search engine marketing or PPC. It is determined as one of the essential
marketing elements for many companies (Quinton and Khan, 2009).In 2019, Google
advertising revenues was $134.8 billion, with YouTube contributing $15 billion for the year,
which shows the main revenue stream on search engines. (Alphabet, 2019)
Although digital or online marketing differs from traditional marketing, the way of attracting
customers and the journey of buyers almost follow the same patterns in both channels. In
digital marketing concept companies attempt to adapt their business objectives to digital
metrics and Key Performance Indicators (KPIs) derived from a short-term and long-term
marketing strategy.
3.1 Key Performance Indicators in PPC Advertising
The main objective of each company is to maximise profits while reducing costs. Each brand
wants to get recognition in the customer mind and create loyalty to drive more customers
through word of mouth as people are convinced by other opinions about products or services
(Banerjee, 1992). In the case of online advertising (i.e. e-marketing or digital advertising),
the main performance metrics are monitored in order to evaluate the effectiveness and
efficiency of certain campaigns. Based on the advertisers' goals, the KPIs can differ. Main
KPIs for PPC advertising are as follows:
Brand Awareness and Reach KPIs (Visibility)
Advertisers do not always focus on website actions and achieve direct results, such as sales
revenue from advertising campaigns. Companies use brand awareness goals, usually when
introducing new products or services, showing updated features of their products and
announcing upcoming innovations. In such cases, they need more engagement with their
content or just viewability on relevant placements, such as websites and YouTube videos.
Display and Video network is popular for brand awareness campaigns. Many campaigns
focus on building brand awareness, and this objective has its own KPIs. Brands attempt to
reach as many internet users as possible with their advertisement message to create brand
The research published by Google and Ipsos MediaCT in 2014 revealed that search ads have
a notable impact on brand awareness and should be included in marketing plans (Bao &
Koppel, 2014). It found that search advertising influences brand perception and provide cost-
effective solutions. Additionally, unlike traditional advertising channels, it can create brand
awareness in the process of the search when users gather information about products or
services. In other words, users see search ads at the moment that they are interested in related
In the case of Google Ads, main KPIs of these campaigns are as follows:
"Impressions" as a metric represents how often advertising content is shown on Display and
Video Network in case of Google Ads. The different metrics "viewable impressions" is used
to calculate how many times internet users see at least 50% of the ad content. From the
perspective of advertisers, impressions, and the audience reach of ad content is an indicator
of brand awareness and brand recognition. Mangani (2004) analysed the difference between
pay-per-view and PPC methods from the perspective of a web publisher and found that the
revenue depends on the attitude of consumers toward advertising.
In the case of brand awareness campaigns, advertisers are charged for thousands of viewable
impressions which is called target vCPM bidding on Google Ads. This is different from the
PPC charging method, in which advertisers pay only for actual clicks. When advertisers are
looking for actual clicks and user engagement with their ad content, they focus on
engagement metrics, such as click-through-rate and conversion rate.
Advertisers not only promote their products or services via banner and video contents but
also create special introductory landing pages to bring internet users to the website for several
reasons. Some of the practical implementations are:
- Targeting them later by using remarketing features
- Making them take certain actions called "conversions."
- Promoting brand-new website and testing user behaviour on it
- Analysing user experience results based on coming visitors
Any keyword with a higher bid gets more clicks and higher positions among other ads while
requiring more budget spend. Several studies, such as Brooks (2004), Feng et al. (2007),
Regelson & Fain (2006) showed the relationship between ad position and clicks.
Additionally, there is a positive relationship between clicks and conversions. (Jerath et al.,
2014; Park & Park, 2014)
Companies focus on as many clicks as possible by relevant users from targeted audiences
based on product or service category. For example, if a company is a restaurant chain, it tries
to drive attention and generate clicks from those who regularly visit other restaurants and
frequently eat out. To capture this audience, advertisers can use contextual targeting to find
these people based on their online activity (audience targeting) and also can target specific
search keywords to show relevant text ads when people look for restaurants or special dishes
to eat out. In this example, this advertising strategy focuses on clicks and generating traffic
to the website to introduce the menu of restaurants, price of certain dishes, locations of
restaurants in different parts of the city, contact information, events and other additional
information. Additionally, when new innovative products enter the market company focuses
on educating users and using different contents, such as explanatory videos, charts, graphs
etc. for the value proposition.
Katona & Sarvary (2010) assumed that the users' clicking behaviour is impacted by four
following factors:
1) The order of the website links on the search results
2) The difference in probability of clicking on the search result list (organic) and
sponsored list
3) Individual differences between websites intrinsic attractiveness
4) Whether the website is shown in both sponsored list and organic results or only one
of them.
Eventually, the study found that the interaction between the list of organic and sponsored
lists and the inherent differences in attractiveness between sites have a major effect on
websites' bidding behaviour. All these factors affect the optimisation and management of
PPC campaigns while encouraging advertisers to make a deeper analysis.
Click-Through Rate
Click Through Rate is a percentage metric representing the ratio between clicks and
impressions. The formula is as follows:
CTR = Clicks/impressions
CTR metric is used with keywords, ads, audiences, topics, placements etc. and evaluates the
relevance and quality of ads. Higher CTR is a positive sign for engaged contents and affects
the performance of ad campaigns. When ad content is highly engaged and has high quality,
an advertiser pays less because higher CTR increases Quality score, therefore also increases
Ad Rank and gets competitiveness between ad Auction.
Olivier et al. (2016) and Rutz & Trusov (2011) confirms the notion that the click-through-
rate can be used to evaluate advertising success. The research found the relationship between
click-through rate and conversion rate, which is one of the main determinants of advertising
effectiveness on search campaigns. (Ghose & Yang, 2009)
Lower Average cost per click
Each advertiser wants to achieve a lower average cost per click when getting traffic to their
website via user engagement with their ad content. However, several factors determine the
cost per click of certain ad content on the advertising networks beginning from ad quality to
auction bid. Real-time bidding is a programmatic auction, having a competitive method
which aims to maximise the profit of mediums on search advertising. (Donnellan et al., 2015)
In Google Ads, the ad position is determined based on Ad Rank, which refers to the
calculation including a bid, auction-time ad quality (expected CTR, landing page experience
by a user, ad relevance), the Ad Rank thresholds (the reserve price), auction competitiveness,
contextual signals (user device, device type, time, search term characteristics, other ads on
the search result etc.) and the expected impact of ad extensions and other ad formats.
Figure 1:Influence diagram of the keyword bidding model. (Küçükaydin et al., 2019)
Once an online user clicks on the ad (sponsored link on the search result) search engines
direct this user to the website while charging the advertiser for this click. Advertisers consider
two factors - 1) a group of keywords to include in ad campaign 2) the bidding price for chosen
keywords. (Küçükaydin et al., 2019)
The quality measures include CTR, the relevancy of ad content, landing page experience of
website visitors. These variables determine the Quality Score of certain keywords on search
advertising. Quality Score with bid determines the position of ad content on the auction
system. Mostly, search engines, such as Bing, Yahoo and Google rank the ad position of
keywords based on the bid, quality score (which is a relevancy measure) of an individual
keyword. (Li et al., 2016) Bids are a main influencing factor which determines whether to
"About ad position and Ad Rank - Google Ads Help."
ads/answer/1722122?hl=en. Accessed 10 May. 2020.
display search ads on the search engine ranking results. Bids impact overall search
advertising costs of the advertiser and ad rank determine the position of the ad calculated by
using bid, quality of the ad, competitive metrics, thresholds, contextual signals etc. The final
results are evaluated by sales, which is an essential objective for many advertisers to achieve.
(Jansen & Clarke, 2017)
One study found a notable variation in the role of keyword competition for multi-channel
versus web-only retail businesses. The study found that keyword competition has a
substantial moderating impact solely for multichannel retailers. The empirical examination
also shows that the position of ads for web-only retailers is reliant on determinants of ad
relevance and bids, whereas multi-channel retailers are more dependent on their bids.
(Ayanso & Karimi, 2015) In addition to ad quality variables, ad extensions have an impact
determining the Ad rank. The extensions to search ad content which improves the visibility
and enriches the content by offering additional information and engagement means, such as
call button, location etc. They contribute to increasing CTR and eventually affect positively
to overrank competitors.
Lilienthal & Skiera (2010) documented the Average CPC for different industries. They found
that in Germany, companies in the Finance/Insurance industry pay 5.37 EUR for one click.
The CPC amount was 2.94 EUR for Internet Services, 2.36 EUR for the Telecommunication
industry. Together with low conversion rates, these expenses make search engine marketing
unaffordable and unprofitable for advertisers. Therefore advertisers do not only focus on
click as the main KPI, which does not always result in a conversion.
Conversion metric KPIs
Advertisers who rely on leads, purchases etc. set up conversion tracking on their website and
choose the conversion-based optimisation method to increase their advertising effectiveness.
Conversion is a predefined online activity which is valuable for advertiser's business.
Conversion happens after interacting advertisement content (i.e. clicking sponsored text link
on a search engine, viewing a video ad, clicking a banner ad) and taking actions on a website,
such as purchasing product or service, signing for a monthly membership or calling to the
business. Each conversion action has specific value for the business which was defined
during the planning stage before setting up an online campaign as tracking conversions
require initial technical configuration on the website. Advertisers should use code snippets
to track specific online activities of visitors and classify them based on the source they came
from. After integrating the web analytics and advertising platforms, they become able to
monitor results and optimise their campaigns in order to increase conversions and get higher
value from conversions.
In conclusion, conversions can be classified as follows:
- Website actions (e.g. purchases, video plays on webpage, signups etc.)
- Local actions are interactions with the location of a business on applications, such as
Google Map. On Google Ads, even store visits can be tracked and valued
- Phone calls (via using call extension on search engine results or calling tracked
numbers on the website)
- Mobile app installs and in-app actions (i.e. downloading mobile applications or
taking actions inside mobile apps, such as purchase, add-to-cart etc.)
Setting conversion tracking not only helps advertisers to track the performance and make
changes to optimise campaigns regularly but also enable them to give different values to
different actions and calculate return on investment to evaluate the advertising channel for
strategic decisions. Each conversion may bring different value, and the total value of all
conversions will increase return on investment if they are optimised properly.
Conversion Value
Each specific action taken by website visitors or engaged users has specific value for the
business. Even calls and signup do not give a financial outcome for the business; advertisers
should evaluate this action and include web analytics and advertising platforms based on the
long-term profitability of specific conversions. For example, users signed up for an online
newspaper and will receive emails once a week. This creates an opportunity for the company
"About store visit conversions - Google Ads Help."
ads/answer/6100636?hl=en. Accessed 10 May. 2020.
for the value proposition of the brand and to sell the product or service in future. Therefore,
based on the potential newspaper signup gets a value. This value can be $1 or $10 based on
several factors, such as the price of the main product or service offered by the advertiser, the
frequency of sending online newspapers, potential revenue per month (which can be
calculated based on historical data of customer activity) etc.
Overall Performance KPIs
Overall performance KPIs help to determine the profitability of advertising campaigns and
contribute data-driven managerial decisions related to future marketing strategy. These KPIs
combine both costs and revenue and find the exact performance and return on ad spend.
Return on Investment - ROI
Return on Investment is a performance metric that shows the efficiency and profitability of
a certain investment. In the case of online advertising, return on investment helps advertisers
to evaluate overall performance and the net return from online advertising campaigns. This
measure is not limited by sales revenue or a number of ordered items per sales; it gives an
insight into the direct relationship between net income from sales and the overall advertising
spend. The traditional formula of ROI is as follows:
Return on Investment = (Revenue - Cost of goods sold) / Cost of goods sold * 100
In online advertising concept, the cost of goods sold represents the transform into the cost of
advertising campaigns. Revenue comes from the sales generated by these campaigns.
Overall, ROI represents the real effect of online advertising efforts to the business.
For example, an advertiser offers a product with a price of $500, whose production cost is
$100. Five of them were sold, and total sales revenue is $2500, and the total production cost
is $500. If advertising costs is $1000, the ROI will be:
($2500 - $1500) / $1500 * 100 = 66%
As seen on the additional calculation costs, such as cost of production and delivering, content
creation expenses are included in the calculation with advertising costs. Thus, the definition
of ROI is different from the digital perspective, and the metric ROAS is used to distinguish
it. Overall, advertisers attempt to get as high value from conversions as possible in order to
cover the cost of goods sold and advertising expenses.
Return on Ad Spend - ROAS
The return on ad spend is used to measure the return of every dollar spent on online
advertising. The main difference of ROAS from ROI is that the formula of ROAS does not
include the cost of ad spend as a subtraction on the numerator.
Return on Ad Spend = Revenue / Ad Spend x 100
For example, if an advertiser spends $20 dollar and drove $200 dollar in revenue, the ROAS
will be:
200/20 x 100 = 1000%
Return on Ad Spend is shown as a metric called "Conversion value/Cost" on Google Ads
user interface.
Google defines this metric as an indicator of return on investment; however,
when offering automated bidding strategy to increase this metric, it uses "ROAS" instead. If
advertiser sets a target ROAS of 300% - for every $1 spend, the advertiser is going to get
three times that in revenue. For example:
$3 in sales / $1 in ad spend * 100 = 300% target ROAS
In conclusion, ROAS only take advertising cost-effectiveness into account, and in order to
find exact ROI, the advertiser should subtract additional production costs from sales revenue
(i.e. conversion value).
"Understand your conversion tracking data - Google Ads Help."
ads/answer/6270625?hl=en. Accessed 10 May. 2020.
Overall, Advertisers prefer search advertising to promote their products and services while
taking advantage of precise targeting (Chen & Stallaert, 2010), low advertising costs (Telang
et al., 2004) and high ROI (Jansen et al., 2011, Li et al., 2016)
3.2 The AIDA Funnel approach for PPC advertising
Research by Nadjla et al. (2014) confirmed that the Internet had become an essential
information retrieval system for numerous users because it includes different sorts of
information in various formats. Users could access blog posts, academic papers, video files,
movies, accounting reports, images and other formats via online search engines in which a
database of thousands of websites is included. Search results on these engines are recognised
as credible sources of information for users, and engines use several ranking signals to
determine the positions of each website. (Lewandowski, 2013) While on search engines, 91%
of users think that they always or mostly find what they are seeking for on search engines,
73% of them always or mostly find the information reliable and accurate. (Purcell et al.,
2012) The digital culture changed the trust apart from experts and in favour of non-
professional information sources such as search engines, blog and social media posts, search
engines. (Park & Cho, 2012; Wolny & Mueller, 2013). Therefore, after recognising this
consumer behaviour on search engines, advertisers began to impact purchase decisions in
different stages of the sales funnel which represent the whole customer journey from
beginning the awareness of the need for the product to purchase.
Research by Jansen et al., (2007) revealed that more targeted search results increase the
performance of web search engines and in the study, search intents were classified as
information, navigational and transactional. The research also found that 80.6% of search
terms were used for information search, while 10.2% were navigational and 9.2% were
transactional over 5 million search queries. Additionally, Jasen and Mullen's (2008)
discussion about paid search advertising focused on its effectiveness as an information search
instrument, meaning that it can be used as a highly-targeted advertising method during the
information search stage of the customer journey.
AIDA Sales Funnel model is a well-known buyer journey demonstration in marketing offered
by Elias St. Elmo Lewis to understand different phases of the purchase process in detail. The
acronym comes from Awareness, Interest, Desire, Action, the phases in the sales funnel. In
order to analyse and demonstrate the importance of PPC advertising in marketing, the AIDA
model was used below.
Awareness Phase:
Search engines that people mostly use to find information on the Internet are Google,
Yahoo!, Ask, AOL and MSN
. October 2019 Report on the desktop market share of leading
search engines shows that Google is a dominant leader in the search engine market with
87.96%, followed by Bing 5.26%, Yahoo 2.73%.
As the Internet gives an opportunity to
access different sources of information immediately and compare results before making a
purchase decision, many people tend to do research on search engines based on their needs
and want in the initial stages of the sales funnel. Advertisers focus on display banners and
video ads to visually attract consumers while publishing sponsored search ads on search
engines. Research shows that online display campaigns are able to increase site traffic, the
search volume of branded search queries, and both online and offline purchases (Fulgoni &
Mörn, 2009) Type of search terms used by a customer shows the stage of the purchase funnel
and helps to identify the intent. Those who use generic keywords are most likely doing
information research and do not have a clear perception of the brands. Those who use the
name of advertisers on their keywords are probably looking for more information about the
business and want to assess the quality. (Du et al. (2017)
Customers get a first impression about the product or service or are aware of the newly-
introduced feature of the existing one, and showing relevant paid search results make them
aware of a new brand or shopping centre. After recognising their need and wants to be based
on the life condition, their initial awareness of need makes them take action to cover it. Thus
in the stage of information search, many advertisers attempt to show their promotion content
"2 - Comscore, Inc.."
Releases/2011/2/(offset)/574/?cs_edgescape_cc=US. Accessed 12 May. 2020.
"Search engine market share worldwide 2019 | Statista." 25 Mar. 2020, Accessed 10 May.
and offer to capture the audience attention. Advertisers introduce search ads on any positions
on Search Engine Ranking Results (SERP) in order to attract customers, and not all of them
are able to get top positions as the environment is competitive. They use keyword-focused,
and Call-to-Action (CTA) included headlines to attract users and match their needs with
product specifications. Moreover, using display campaigns to show products or services
regularly make consumers purchase and think about covering their needs.
In search advertising campaigns, in order to show headlines relevant to the search intent, it's
recommended to use keywords on the content, so internet users recognise it and click.
Headlines also contain brand names to make aware users that products & services are sold
by recognised brands and trusted companies which users have already purchased before.
Additionally, this strategy is also implemented by new entrants in the marketplace to build a
reputation. One research shows that using moderate explanatory headlines create positive
brand communication (Bergkvist et al., 2012), which may increase clicks/impression rate
(CTR) and result in a lower cost per click for advertisers. As each company attempts to
achieve effective results from advertising campaigns, setting sophisticated goals based on
web metrics and the purpose of a campaign is at the centre of discussion in case of search
engine marketing. Advertising effectiveness is a quite general concept that commonly refers
to the market response to an advertiser company. Another factor affecting KPI choice is the
position of the company in the competition. If brand recognition is lower companies
concentrate on increasing CTR and decreasing CPM (cost per a thousand impressions).
Eventually, spending money wisely on building awareness will be an essential benefit in that
case. Since many advertisers choose a primary objective as attracting web users to their
landing page showing details of their product or service by improving the visibility on SERP
results. (Keng & Lin 2006).
Interest Phase:
In this stage, buyers are interested in what is offered by advertisers. PPC advertising is said
to produce much better results in comparison to the old-school pay per impression methods
such as banner display advertising, or pay per view based method. (Drolias, 2007) In order
to charge for clicks rather than impressions, advertising platforms require high-quality
content and ensure internet users showing related contents and sponsored search results. In
this case, advertisers who create low-quality content pay more for clicks and do not compete
because of higher spends.
This mechanism encourages advertisers to increase the quality
of message and content, as Google defines these parameters as Landing Page Experience, Ad
Relevance and Expected CTR (based on historical data). Finally, advertisements become
relevant and attractive for interested users on this specific topic of search. Customers match
needs with offered product or service and then search more for eliminating less satisfactory
options. In information search, they attempt to get data as much as possible about alternatives
to compare with in the next phase of the sales funnel. Jansen et al. (2011) stated that in the
Research phase (which is equivalent to Intent in given model) advertiser target more focused
keywords including commercial intent and product specifics but without including the brand
Advertiser matches focus keywords to sponsored ads, introductory web pages or visual ad
(in case of Google Ads, Google Display Network ads enable advertisers to target internet
users through Display Network based on custom intent which determined by user's previous
interest on search engines) in order to catch search intent. As the main goal of Search Engine
Marketing is improving the visibility of company websites on search engine results and
generating leads from paid searches (Karjaluoto & Leinonen, 2009), companies benefit the
interest of searchers, converting their interest into the desired action on websites.
Leveraging long-tail keywords which are highly specific and more relevant than broad
matched is one of the recommended ways to attract users, make them recognise and match
their interest with given advertising text. It is considered the way to catch searchers with the
lower cost because of lower competition for rarely used search terms
, and it is argued to be
more valuable for those who run paid search advertising campaigns than using search engine
Leveraging of long-tail keywords was also recommended for avoiding
"What Is Quality Score & How Does it Affect Google Ads?"
Accessed 10 May. 2020.
"How to Use Long Tail Keywords to Boost PPC Search Traffic." 14 Jun. 2019, Accessed 10 May. 2020.
"Long-Tail Keywords - WordStream." Accessed 10
May. 2020.
competition and getting positioning on search results for search engine optimisation
Although advertisers still use long-tail keyword strategy, a study by Skiera et al. (2010)
concluded that it is not suggested to concentrate more on long-tail keywords and
recommended to work on finding top 100 keywords with the highest performance.
Additionally, the research by Clicteq
supports the same conclusion showing that only 2.4%
of conversions came from long-tail keywords that include 4+ words in comparison to the
80%+ of conversions from the top 20% of keywords. This study suggests spending time to
optimise top 20% keywords is 33 times more effective than spending time to add new long-
tail keywords.
Desire Phase:
After learning about product or service buyer and finding the most reasonable choice by
eliminating "worst" ones based on personal preferences such as price, quality factors etc., the
next phase is followed in which buyer wants to find a satisfactory product to take action. This
makes PPC a valuable tool for brands to distinguish their product for information seekers.
Khraim (2015) found that there is a statistically significant impact of using PPC marketing
on bringing new buyers to the website in e-marketing businesses in Jordan.
In the decision phase, Jansen et al. (2011) classified keywords, including brand names,
without including full product names. For example, "Dell Inspiron laptop" includes the brand
name, but the user is still doing research to evaluate possible alternatives between the product
line of one brand. Study shows that consumers always consult different information sources
to identify choice options (Palmer, 2000). They are affected by advertising messages, other
consumers and also previous experiences (Solomon, 2015; Bettman & Park, 1980).
Users could have decided to leave or delay the purchase decision at any time during a
customer journey for one of many different reasons. A better understanding of the customer
"11 Reasons You Need to Focus on Long-Tail Keywords for SEO."
keywords-seo/. Accessed 10 May. 2020.
"Sorry To Burst Your Bubble: The Long Tail Keyword Myth ...." 12 Sep. 2016,
long-tail-keyword-myth-a-data-driven-argument/. Accessed 10 May. 2020.
journey and potential impacts on decision-making processes can contribute to increased
return on investments on marketing efforts with a highly targeted and individually
personalised approach by using a funnel-based model. As PPC advertising contributes to
reminding product specifications and key information about its quality by search ads and
attractive call-to-actions, buyers become prone to reconsider promoted products in the
process of evaluation of alternatives.
Advertisers can convince users in Desire stage by offering discounts and showing previous
satisfied customers. (Gürbüz et al., 2016) They help customers to differentiate offered
products than others and focus on convincing them to continue to the next funnel stage -
Action Phase:
After finishing the previous steps, the buyer is ready to take action to get the desired product
or service. The action can vary based on the objective of advertisers. Keywords were
classified for the Purchase stage when the full product name was included in the search query.
This implies that the users finished the information research and know which product fits the
need. Therefore product and brand name were intended to be found. (Jansen et al., 2011) The
action phase is a final step desired by advertisers, in which the objective is achieved.
Objectives can be newsletter sign-up, lead or completion of the contact form, fundraising or
purchase of product or service. Website analytical tools are used to track user engagement
and measure success in terms of previously mentioned goals.
As discussed above, PPC can contribute different phases of the customer journey in
marketing activities and businesses can leverage benefits by matching their objectives to web
analytics metrics. For each step, different KPIs are determined, and a digital marketing
strategy is built. Based on the objectives and allocated budget, the role of PPC can vary in
each phase.
Companies can easily leverage the way of increasing website traffic in a short-term by using
PPC advertisements which require lower investments and less effort in comparison to SEO -
another branch of Search Engine Marketing as some authors classified. However, it is crucial
to hire a PPC expert or specialist to manage and monitor the campaigns. In addition to PPC
campaign manager, a collaboration of web designers and web developers with the digital
marketing department has an impact on results. Individuals' mental responses to the ad
content differ and are closely associated with behavioural, attitudinal and memory effects of
advertising. (Weber & Schweiger, 2017)
There are studies showing the effectiveness of PPC in marketing activities in different
countries. Although many users think that search engines contain inappropriate information
for their intent and do not fit their needs (Mukhopadhyay et al., 2007) they still use them to
find specific information from the website database of above-mentioned engines and this
makes the search engines a source of information for billions of users each day, creating a
marketplace for dozens of companies to promote their products and services. Considering
the fact that information search is crucial for e-commerce activities (Gefen & Straub 2000),
the increasing PPC costs of brands also cause increased job opportunities in the PPC market,
giving a chance for automation tools as well.
Limitations with the AIDA model concept
Although research revealed that consumers follow sales funnel stages in real life, in online
advertising, it does not always happen in the same way. The effectiveness of the sales funnel
model is useful for planning and creating a strategy when creating certain messages, ad
contents, and for targeting different keywords for each stage, however, it does not represent
the exact flow of internet users in case of PPC advertising. (Jansen et al., 2011) One research
also supported the notion that the behaviour of consumers is not always consistent with the
purchase funnel. It was conducted in South Korea between two popular running shoe brands
Adidas and Nike. The research results showed that the purchase funnel model was not
consistent for both brands. For Nike, the purchase funnel oriented keyword classification did
not perform as expected. However, for the following brand Adidas the framework was
effective, and users follow the stages as expected when different online retailers classified
category-level, brand-level and model-level keywords to attract online searchers and affect
their purchase decision-making process. (Kim et al., 2019)
Additionally, steps included in the AIDA model can be modified based on the industry as
several other models were developed and presented. For example, Engel, Kollatt, and
Blackwell (EKB) model include five steps with different names.
Figure 2:EKB model of sales funnel
EKB model also provides managers with valuable explanations of the customer journey with
detailed insights (Ashman et al., 2015). EKB model has been applied to online shopping
(Darley et al., 2010) and Wolny and Charosensuksai (2014) agreed that 5 stage consumer
journey best explains the highly-involved purchase activities which lead to being more
lasting processes than more habitual shopping behaviour. Some studies (Vuong K. T., 2015;
Abdurrahim et al., 2019) separated the Interest and Search in AISAS (Attention, Interest,
Search, Action and Share) model derived from AIDA, describing the search as a further step
after interest as users do a search after getting a positive sense of products in their minds. In
the study by Poyraz et al., (2017) which aimed to match KPIs with AIDA funnel, Attraction
(referring to Awareness) was matched to Impression, Interest was matched to View, Desire
was matched to Click, and finally, Action was matched to Conversion.
(n.d.). EKB model - PDFSLIDE.NET. Retrieved May 15, 2020, from
Although other customer decision-making models were designed to describe the buyer
journey process, this does not deny the fact that PPC advertising has its own effectiveness
during the process impacting the evaluation of alternatives, awareness about new brands,
finding new information and purchase (with discount promotions and special offers.) In each
stage of the process, advertisers present different ad contents to the users and implement
strategies based on several factors that have a potential impact on attracting users to the brand
and finally make customers purchase.
Summary of the benefits of PPC advertising
- It gives more control over what will be spent on clicks and conversions ( i.e. advertising
- The availability of using data analysis tools and integrations with web analytics and CRM
databases make PPC apparently preferable than traditional advertising in terms of
- Availability of configuring campaigns by using a wide range of variables that affect
performance and targeting. These variables are demographic, behavioural and technical
dimensions, including location, time, keyword, household income, age, device category etc.
Switching among options may increase the profitability and decrease the amount of spending
for the final outcome (goal or conversion). This also creates additional spaces to enter the
market with a specific combination of those variables. Research (Jansen et al., 2013) found
that even changing an intent keyword type from gender-oriented to gender-neutral may
increase 20 times the performance metrics of search campaigns.
- PPC collects accurate data for further marketing decisions based on data. Companies that
want to enter any particular industry can leverage the capabilities of search engine marketing
platforms to get initial information about brand awareness and market fit of any product &
service by spending less amount of money rather than spending time and effort traditional
advertising by using TV and street billboards.
- Providing a quick brand positioning on SERP results for small businesses. Even small
companies can overrank famous and market leader brands in a short time. Although there
was a court case on prohibiting using competitor keywords in search campaigns and policy
of Google about prohibiting using some brand names keywords on advertising content (ad
text and ad image), companies still can show their ads to those who search for products or
services of competitors. One study showed the importance of search engine marketing for
less popular brands. (Dou, W.s, Lim, K. H., Su, C., Zhou, N., & Cui, N. (2010)
- PPC can generate sales in a short-time for companies that use highly attractive advertising
content, well-designed landing pages and enable better user experience for potential
4. PPC Automation
4.1 Importance of Automation in Digital Marketing
Some studies (David, 2015; Bessen, 2016) show that computer automation causes job losses
but at the same time creates new ones for others who can learn new skills in increasing
demand for innovations. Automation causes economic inequality, so employees need to get
new skills in order to survive in the digital era. Machine learning revolutionised the
technology market, and nowadays, 84% of marketing companies leverage machine learning
benefits and enhance their capabilities in communication and services. Predicting consumer
behaviour became more precise and faster by analysing historical activities of consumers on
websites, in retail shop centres etc. (Bayoude et al., 2018)
The implementation of Machine Learning technologies to the business environment gives
more detailed analysis opportunities and gaps in communication and sales process. After
defining business objectives on customer behaviour, predictive analytics contribute to the
sales and marketing department with better forecasts and trend reports using for data-driven
strategies. Predictive analytics provides advertisers with behavioural analysis which helps to
understand technical, psychological and demographic parameters that affect a user's purchase
behaviour determining their wants, needs and personal preferences. Automated systems
extract information from datasets and process them based on given business objectives by
the company itself. Therefore, data scientists should train their machine learning algorithm
with the help of specialists with domain knowledge of the field.
Machine Learning models are trained regularly to get better results considering more patterns
and correlations to form ideal customer personas. Gartner anticipates
that by 2020, almost
30% of companies will be utilising AI and machine learning in at least one of the sales
processes. The machine learning reforms on advertising, email marketing, content creation
and social media create new opportunities while making many tasks automated and
"Gartner Says AI to Have Significant Impact on Sales Training and Coaching" 19 Sep. 2019,
on-sales-t. Accessed 10 May. 2020.
straightforward and reducing human participation. The main benefits of machine learning in
digital marketing are as follows:
- Personalised solutions
In the digital world, brands can distinguish themselves by the well-designed user experience
on their websites, application or on social media channels. Machine learning increases the
superiority of this experience by adding valuable data resources to the process. Personalised
messages, call-to-actions, webpages, suggestions to the end-user create a direct-marketing
effect and users recognise the customised experience and feel special care from the brand.
Brands use this strategy on email marketing to create special messages and promotions to
increase the attractiveness of their content. Netflix uses recommendation algorithms to
validate their improvement plans with existing data to discover how users react to the
recommendations in their journey through Netflix platform by implementing online A/B tests
and analysing long-term behavioural patterns. Based on the analysis platform, personalised
content offering, ranking, titles, page generation, searches, messages etc. and make them
appealing and likely to enjoy over 100 million subscribed members.
- AI in Customer Service
Chatbots nowadays are available in many customer services and help customers to find
relevant information and even help customers in their purchase journey by offering
demanding products and broadcast them when new products and services are available. Every
year, chatbot algorithms are improved and replace customer service workers with their fast
responses and simultaneous working capabilities. Chatbots not only save money and time but
also provide better results if users are satisfied and regularly engage.
Earlier chatbots solely used common basic reply messages, but as they are improved with
new algorithms and increased their knowledge base, they have transformed into intelligent
tools understanding the needs of customers. E-commerce websites eventually have chatbots
to support their users in purchase decisions. Even the applications as Facebook Messenger
have chatbots on which Facebook users can contact chatbots that are sales representatives in
"Machine Learning - Netflix Research."
Accessed 10 May. 2020.
the process of choosing the desired products or services for customers. Intelligent bots chat
by sending speech or text messages in order to communicate with end-users. (Girdher, 2019)
- Content Creation
Machine Learning enables content publishers to find specific patterns between most
attractive and engaging contents and suggest more productive solutions by forecasting based
on historical data. The process is analysing big data including colours on photos, paragraph
sizes, messages, titles etc. (the variables that have an effect on content performance) and
predicting personalised content for each customer journey in sales or impactful message in
the example of advertisement contents. Creation of content is not a straightforward task
because of including initial research, data analysis, following trends and creativity. Repetitive
tasks, such as visiting certain websites to follow trends, collecting contents from different
sources and finding similarities among content, can be easily automated. However, creativity
requires human intervention in the process in order to mix business objectives, the message
of product or service and current trends. However, Machine Learning is able to create its own
content by analysing given input, such as novels.
In 2016 Ryan Kiros from the University of Toronto shared her open-source project "Neural
Storyteller" on GitHub based on the academic research (Kiros et al., 2015). This tool is
trained to read articles and process them in order to create descriptive short stories for images.
Such innovations open a door for improvements on content creation and generation of
product and service descriptions on business websites.
Overall, the AI and ML-based technologies may impact the digital marketing fields as follow:
- Website Design & UX
- Marketing Automation
- Advertising Optimisation
- Email Marketing
- Social Media Management
- Influencer Marketing
- Search Engine Optimisation
- PPC Management
4.2 Management of PPC Advertising:
Sponsored search advertising campaign development involves organising keywords into ad
groups and developing ad copies for ad groups. (Chatwin 2013) Another study also stated
that in a duration of search campaigns, PPC managers encounter necessary keyword-related
decisions including keyword mining, selection of best combinations, grouping and
adjustments. (Yang et al., 2019) Each ad group involves different ad contents that are
intended for particular promotional intent, and also targeting options is configured for one
specific user audience. This classification helps to track the performance better for
advertising efforts.
Google Ads platform includes Search, Display and Video networks. As
this study focused on the management of search and display ad campaigns, the steps of
management processes are classified as follows:
Keyword research
In the case of search advertising, in this step, advertisers use Keyword Planner by Google
Ads or third-party keyword research tools in order to get CPC and search volume data of
specific keywords. Based on collected data in this stage, which is an estimation and can vary
due to several impacting variables, advertisers make decisions in the next steps.
In the case of display advertising, in this step, advertisers analyse their buyer personas to
predict the online behaviour flow of the target audience. Which websites do they browse
regularly? Which type of lifestyle and affinity they have? What are their search preferences?
Which type of contents do they browse? - and such questions determine the digital persona
of the target audience in case of Google Ads.
Some studies along the line of keyword research essentially concentrate on keyword selection
(Kiritchenko et al., 2008, Lu & Zhao 2015) and keyword generation (Ravi et al., 2010).
Several studies (Lee et al., 2018; Scholz et al., 2019; Qiao et al., 2017; Fan et al., 2017)
conducted to generate keywords from queries.
(n.d.). Organize your account with ad groups - Google Ads Help. Retrieved May 9, 2020, from
Campaign structuring
With well-organised keywords help to show the right search ad texts to the right audience
while resulting in increased traffic and profit. (Yang et al., 2017) Keywords are grouped
based on the user intent, and campaigns are structured for sophisticated budget allocation for
better comparisons.
Content creation
Haans et al. (2013) found that the type of evidence given on the ad content affects the CTR
and conversion rate. The included evidence types in this research are anecdotal evidence -
using case stories to strengthen arguments quality, expert evidence - citing experts to increase
reliability, causal evidence - explanation of occurrence, statistical evidence - using numerical
explanation samples. Contents are chosen based on the funnel stage and the message of the
promotional campaign by the advertiser. Content types are classified in the campaign
structuring step and become ready for optimisation after choosing the right contextual signals
for them.
Campaign activation and Optimisation
After completion of content creation and campaign structuring, the most important part for a
Manager is activating the campaigns, regularly monitoring the results and optimising for
better outcomes. The concept of AiAds was analysed in the study by Yang et al. (2019, July),
which focused on bidding optimisation. Some studies focused on deep reinforcement
learning in real-time bidding and second priced auction to predict bidding behaviour based
on historical data and optimise campaigns. (Chen & Rabelo, 2017; Zhao et al., 2018, July)
These studies mainly focused on the technical side of bidding optimisation.
4.3 Automation of PPC management
As automation is about increasing the operating efficiency of digital marketing processes and
includes building algorithms and rule-based models to decrease and even eliminate human
intervention, it is also applicable on PPC management because of its technical nature.
Real-time monitoring and changing bids for each parameter (keywords, audiences, locations
etc.) are a main time-consuming part of PPC that can be automated and by using the objective
function (the function that it is wanted to minimise or maximise) variables can be adjusted
with after proper data analysis and statistical methods. The importance of automation
emerged from several factors:
Lack of time and human resources: In order to implement manual bidding PPC managers
should invest time and effort on it. Implementation of manual changes, monitoring daily
performance and controlling each detail of the campaign requires more time than automated
Lack of expertise to use data resources: Being under-informed about the correlation
between performance metrics and contextual signals that Google tracks over the web search,
which makes it harder to find exact targeting criteria for getting the highest return on ad
spend. In that case, more historical data helps to make more data-driven decisions after
sophisticated data analysis, which makes it difficult to do for small businesses because of a
lack of expertise. In order to find gaps and reasons of fluctuations on the KPIs PPC manager
should have certain expertise to find a root of the problem and should design several
hypotheses to check and validate based on historical performance data. When a PPC manager
lacks many essential skills and can not use the resources that the company has, a certain
percentage of ad spend is wasted, and it damages ROI.
Having less data than Advertising Platform: Advertisers should spend in order to get an
insight about KPI performance of certain keywords, placements, topics, audiences, devices
and other parameters available on campaign configuration of advertising platforms. For
example, Google Ads provide historical data - search volumes of keywords in order to make
forecasts about future campaigns for certain chosen key phrases typed by users on their web
search activity on Google. However, it is not possible to know in which device types, hours
these keywords will perform better. In order to have this information, advertisers should
spend budget and configure the future campaigns based on this data. Each time, analysing
the historical data and making adjustments on bids, especially on bigger accounts, is a time-
consuming process. Additionally, based on the objectives, these adjustments will vary. For
example, if a company wants to get as many clicks as possible rather than conversions, PPC
managers should track all signals and parameters in order to react immediately to get a higher
amount of clicks in the right time, in the right device for the right keyword. At least search
advertising is more straightforward to implement than Display advertising as advertisers are
not aware of whole website names included in Display Network of Google Ads services and
PPC managers should spend their whole day monitoring thousands of website, mobile app
and other display placements for several campaigns regularly to improve performance. As
Google Ads have more filters to analyse in which device certain websites perform better to
get optimised for certain KPI (i.e. clicks, conversion, conversion value etc.) and users only
access the unfiltered data on placements report, advertising platform itself has the better
advantage to access all filtered and structured data to optimise results.
To eliminate the time-consuming tasks and improve the overall performance of advertising
campaigns, Google Ads introduced an automated way of this process called Automated
bidding strategies whose current name is Smart bidding strategies.
4.4 Automation of PPC Management on Google Ads
Google introduced Automated bidding strategies for better PPC management experience of
the advertisers. Increasing demand for search and display ads make users create large client
accounts in Google Ads platform, which make it difficult to manage, monitor and regularly
adjust to adapt to new changes in campaign information (e.g. doing keyword bid adjustments,
allocating budget through different audiences etc.) or parameters which affect campaign
performance. In this case, by understanding the demand, Google Ads developed its
Automated bid management models which are called Smart Bidding. Smart Bidding
introduces result-based campaign objectives to users which is focused on improving
campaign performance based on chosen objectives, such as Target CPA (for those who want
to get conversions or desired actions by a user by setting a higher cost per action limit),
Target ROAS (for those who want to achieve a certain ratio between value and cost per
conversion), Maximize Clicks (for those who want to get as many clicks as possible within
daily budget) Maximize conversions (for those who want to get most conversions within
daily budget), Maximize conversion value (for those who want to achieve the highest
conversion value within the daily budget), Enhanced CPC (for those who want to get more
conversions within daily budget while controlling Max. CPC by setting it manually) and
Target Impression Share (for those who want to show ads on the top or anywhere on the
search results based on the preferences).
Apparently, Google has more data than individual platform users or PPC managers as it
collects search data from billions of actions on the search engine. Google uses machine
learning algorithms to analyse industrial data individually and find patterns and correlations
between conversions and campaign parameters configured by advertisers. Industry-specific
data helps to identify benchmarks and similar patterns through the websites. Applying years
of practice and data, Google accurately designs its algorithms to manage campaigns by smart
bidding strategies, which result in a higher return on investment and data-driven allocation
of budget to increase ultimate profits for companies. Several case studies were introduced to
show the success of Smart bidding campaigns:
- Device
Devices include desktop, mobile and tablet devices. The study by Song et al. (2013) shows
that there is a difference in users' search patterns, the average length of search queries, query
categories, hourly distribution of the search volumes, CTR and other points. The target
audience of advertisers use certain device types when searching or browsing information on
the internet and suddenly see advertising content. The size of content, the position of the
advertising banner on the ad placement, mobile-friendliness of advertiser’s website matter
when considering campaign performances for different devices. Bousquet et al., (2018)
introduced mathematical approach by clustering mobile campaigns to optimize KPIs and
target the audience with relevant context variables. For example, if advertiser uses Display
and Video campaigns rather than Universal app campaigns (which is for only app promotion)
on Google ads to promote the mobile application to get more leads or in-app events (i.e.
purchase, add-to-cart, product view) previously mentioned technical details related to ad
content and the system compatibility of a mobile application with device types should be
considered. Otherwise, given ad content will direct desktop users to mobile applications,
which is not available to use on desktop devices and part of the advertising amount will be
wasted. Such mistakes by advertisers damage the performance, and it’s nothing related to
Another issue related to optimization (i.e. increasing ROI) is comparing conversions for
device types. In this case, each device type receives conversions or other positive results
based on the advertiser's campaign objective, but certain ad campaigns perform differently
for different devices. The main goal of advertisers is to prefer the best performing device
type to others in order to increase returns on ad spend and reduce cost per actions or cost per
conversions. In Smart bidding strategies, there is no need to adjust bids manually because
automated bidding does it itself for advertisers and target device types that are more relevant
to advertiser’s objective (e.g. getting more impressions, driving more traffic, sales, specific
conversion actions etc.).
- Operating System
Available options on operating systems are Android, iOS, WebOS, Blackberry, Windows
Phone. Based on the technical compatibility of an advertiser's website or application,
performances of operating systems can vary from different perspectives. Global statistical
data provided by Monetate (2019) show the difference in conversion rates through operating
Smart bidding strategies analyse historical and present data and target the most appropriate
devices used by the target audience of the advertiser in order to achieve more benefit. For
example, if the cost per conversion for iOS users is higher than Android users, investing in
Android users may result in higher ROI, and it’s desirable by advertisers. However, if the
goal is to get more impressions, machine learning will focus on getting lower CPM (cost per
million of impressions) result for advertisers. It should be considered that Google Ads UI
doesn’t let advertisers change bids for operating systems, but this parameter is used as a
signal in smart bidding.
- Browser
Chrome, Safari, Internet Explorer and other browsers are variables that can affect ad
performance but in Google Ads user interface advertisers, who do not use premium Google
Marketing platform products, cannot change and adjust bids for specific browsers. However,
as Google itself has access to this information, smart bidding can use those signals to
optimize campaigns.
- Web placement (for Display campaigns)
The Display Network of Google Ads consists of 2 millions of websites, apps and videos
which reach over 90% of all web users over the world and are available for advertisers to
show their ad contents. This contextual targeting enables advertisers to target specific website
users and limit their display campaigns only with selected websites, apps and videos, which
is called managed placements. However, if advertisers do not use managed placements ads
appear on placements which are automatically chosen by Google Ads itself. If a user did not
limit placements with specific managed one, there would be differences in performance for
each of them. Therefore there will be a space for optimization. As websites can be classified
for topics, which is called contextual targeting, placement can be optimized by changing bids
for topics that they belong to.
- Interface language
Language preferences are chosen by users in the browser or in a Gmail account. Based on
the language of the target audience and the content published by the advertiser, the
performance may vary for languages people use. Especially for advertisers who target the
countries and cities in which people use two or more languages on the browser, it’s
impossible to adjust bids for interface language on Google Ads user interface. Based on the
historical and present data machine learning use the performance report and target more
sophisticatedly to improve results user interface language as a signal.
- Site behaviour
Users’ activity on advertisers’ sites, including a number of visited webpages, the progress of
achieving a conversion, the value of products on product pages, and other websites which
have been visited before. After clicking any ads, users take some actions on the advertiser’s
website. Google analyzes this activity and decides whether the landing page is relevant or
not for the user. Additionally, users who are more prone to convert are targeted on the next
"About Smart Bidding - Google Ads Help - Google Support."
ads/answer/7065882?hl=en. Accessed 10 May. 2020.
phases of the campaign. This data is also known by Google Ads machine learning, and only
users who leverage analytics tools can get the site behaviour data of those who came from
paid traffic.
- Search Network partner
Search partner websites of Google the ad appears on. Search partner websites include ad
placements which registered ad Custom Search Engines on Adsense platform and they also
serve as placements for sponsored text ads.
- Actual search query
Search term or search query that the user typed on a search engine, which resulted in showing
a sponsored ad or matched the given keyword by the advertiser. Advertisers are capable of
eliminating irrelevant search queries which matched their keywords used on search
campaigns. However, machine learning eases this process and reduces time waste by
analysing all search queries by using other signals, such as site behaviour of those queries,
conversion results etc. Spending more on relevant and converting search queries increase
keyword performances.
In order to maintain regular traffic and positive emotional reactions towards websites,
website owners rely on the intents of users. Based on search intents, they focus on emotional
responses on websites which affect the user behaviour and increase the effectiveness of
traffic. (Deng & Poole, 2010) (Lowry et al., 2015). When a user clicks on the ad on the search
result and lands on the webpage, potentially useful information on the search result headline
and description has already affected the perception about the website and offer. The ad
content itself is designed based on search intent after dedicated keyword research.
- Demographics
Demographic data including gender, age, household income, affinity and in-market
audiences which were given on Google Analytics as well. All demographics factors are
combined and analyzed by machine learning in order to find the right patterns between
"Custom Search Ads - AdSense Help - Google Support." Accessed 10 May. 2020.
converting users who bring the most benefit to the business and more relevant to see ad
- Ad characteristics
Ad characteristics include size and formats of given advertising content. As advertisers use
different types of ad contents, such as .png, .jpg, .gif, html5 banner ads, expanded text ads
and responsive search ads, every single factor related to size and format may affect the overall
results of the campaign combined with other parameters or signals. For example, based on
the target device types, the chosen ad content format may vary and perform differently.
- Remarketing list
Remarketing list includes previous visitors of the advertiser's website. The specific bid can
be set for this audience. Remarketing tag which is set by advertisers on the website to collect
visitors data by using cookies and match this data with Google Ads platform when targeting
this audience. Users in the remarketing audience who visited the website before are already
aware of the brand. One research (Arya et al., 2019) shows that brand association with brand
connection is easily recognizable for those who use remarketing campaigns. Another study
(Lambrecht & Tucker, 2011) found that retargeting users with personalized ads are four times
more effective than targeting with generic ad contents and six times more effective than
targeting with standard banners. Other studies also show that retargeting has a positive effect
on the increase in web visits and sales (Lewis & Reiley, 2014; Lobschat et al., 2017).
- Weekday & time of day
It is determined by the Internet user’s local time in specific time zones. Time is also another
important factor that affects the user behaviour as a whole. Even the frequency of shopping
differs between weekdays and weekends (Sugie, Y. et al., 2003). Therefore any increase in
search behaviour on the internet and intent of customers directly affect the results of
advertising campaigns. Another important factor is the hour of the day. Moreover, sometimes
specific hours of a day can give different results and machine learning can get better results
by increasing bids and outbidding competitors in order to get traffic from converting
customers in specific parts of a day. Huang et al., (2018) found that people from business
districts tend to purchase and browse more in work time while ones from residential areas
tend to purchase more in work time. Therefore based on the location targeting, advertisers
should target more converting hours in order to get effective results. In such cases,
automated strategies perform better than manual methods considering the fact that PPC
managers can not recognise the specific time of a day to react quickly by increasing bids to
outbid competitors if there is no statistically significant historical data on the account which
is almost unable to predict particular circumstances and change in customer behaviour other
than seasonality.
- Location intent
The location intent on the search query in addition to the physical location of the user, such
as “car dealers in Toronto”. In such cases, advertisers should add negative keywords to
eliminate irrelevant matches for search campaigns if one specific location is targeted or this
location can be specified on keywords in order to avoid showing ads to users who are looking
for different locations. Eventually, the intent of the user affects the performance of the
keywords. Machine learning identifies such performance issues with irrelevant searches and
makes a decision based on searcher’s behaviour towards sponsored search ads. Even general
search keywords can match specific location intent terms when the location targeting is
configured for the intended location.
- Physical location
The location the user is located in. Even the advertiser did not specify the campaign location
settings Google collects and analyses this information. Collected data can be used to
determine which locations give more interest, conversion, conversion value etc. Next time
machine learning allocates a budget for winning location in order to catch converting users
if this signal triggers the desired results. Findings of Huang et al., (2018) included the
different user behaviour based on location.
All contextual real-time signals are combined and form smart bidding strategies to
understand user behaviour and help the advertiser to get desired results in a given budget.
Machine learning takes the guesswork out of the targeting strategy on campaigns and enables
advertisers to switch from intuition-based approach to data-driven one. Contextual targeting
not only helps advertisers to show ad content to users in the right place, in the right time and
on the right device but also decreases dissatisfaction of the target audience as numerous
irrelevant and inappropriate advertisements frustrated internet users when users are not
interested in the topic, product & services at all.
Figure 3:Contextual signals on Google Ads
It's recommended
to have historical data to use Smart bidding in order to have enough
previous conversion data to optimize for. Because algorithms need time
to adapt changes
and learn, advertisers should not make significant changes to the campaign, such as adjusting
daily budget and bid strategy, which relaunch the learning period of the campaign.
4.5 Smart Bidding Strategies:
Target Impression Share
This bidding strategy is for advertisers who want to show advertisements on the absolute top
of the search results page (1st position), top of the search results page (first four places above
"Going head to head with Google Smart Bidding: The good, the bad and the weird" 11 Apr. 2019,
weird-315242. Accessed 10 May. 2020.
"id Automation: The Basics of Machine Learning & A Guide to Smart Bidding" 20 Mar. 2018, Accessed 10 May. 2020.
"About campaign statuses - Google Ads Help - Google Support."
ads/answer/1722131?hl=en. Accessed 10 May. 2020.
the organic results) or any position on the search results page (below or above the appeared
organic search results).
Advertisers should choose one of the given options and percentage
of impression share to this target based on the campaign objective and business strategy and
specify maximum CPC amount as a limit.
Research findings show mixed results regarding ad rank and performance. Liu et al. (2009)
found that ads appearing at the top of a page resulted in higher clicks than ads at the bottom
of the same page. Ghose and Yang (2009) claimed that the position of the sponsored ad serves
as a quality sign and consumers associate trust and higher quality with the higher-ranked ads
on search results while Narayanan and Kalyanam (2015) questioned the association between
conversion behaviour and position of search ad in general. A study by Agarwal et al. (2011)
evaluated the impact of ad placement on revenues and profits, reporting that while CTR
decreases with an ad’s position, conversion rate’s increase, especially when more specific
keywords are used. Interestingly, middle positions can also produce powerful results.
Figure 4:Top ad positions on Google Ads search campaigns
(n.d.). About Target impression share bidding - Google Ads Help. Retrieved May 9, 2020, from
(2018, November 6). Introducing four new search ad position metrics - Google Ads Help. Retrieved May 9,
2020, from
For example, if Max CPC bid limit is $2 and Impression Share target is 90% on the absolute
top the search results page, Google Ads will set a CPC amount automatically to get 90% of
absolute top position without exceeding the $2 limit. However, setting Max CPC too low will
restrict the bid management strategy, and the campaign will not be able to achieve its goal.
On the other hand, if competitors are also bid aggressively for top positions, the budget can
be over in half of a day because of higher CPC.
This strategy is helpful when advertisers do not focus on conversions, and the purpose is
getting a specific position for brand awareness or visibility. Even if it is costly, many
businesses choose to use it without concern about advertising expenses because they
recognized this method as a more influential way to increase firm value, brand recognition
and long-term effectiveness. Additionally, advertisers, who considered that top positions
drive more conversions for the business they leverage this bidding strategy to be on the 1st or
top position.
Advertisers use competitive poaching on search engine advertising which refers to using
competitor-brand keywords to show search ads and this method is usually implemented by
smaller-budget firms to drive traffic from the keywords with competitor names (Sayedi et al.
2014). Desai et al. (2014) analyzed the condition under which advertisers leverage the search
engine advertising platforms for this purpose. These theoretical studies differentiated brand
keywords and competitor keywords used in the search campaigns. By using competitor
keywords, advertisers drive additional traffic and users who have purchase intent to see the
competitor’s product and the purchase decision is influenced in this type of scenario. Du et
al. (2017) found that competing-brand keywords are associated with lower CTRs but higher
CRs, proving that keyword poaching can be used to drive conversions. Considering the ad
position advantage by using Target Impression Share bidding strategy, advertisers can benefit
from competitive poaching opportunities by showing their ads on the first position of the
search results when users are looking for competitor products or services.
In addition to this strategy, for automation purposes, Google Ads Scripts introduced a script
to bid to impression share for accounts that want to achieve impression share goals. It Is
executed once a week and does two things:
- Increases the bids for keywords whose impression share is too low
- Decreases the bids for keywords whose CTR is better than 1% and impression share is
too high
After the announcement
about removing Average Position (Avg. Pos.) metric from Google
Ads, the platform introduced the new Impression (Absolute Top) % and Impression (Top)%
metrics which became available from November 2019. These metrics and additional ones
Search Impression Share, Search Top Impression Share etc. were also used to optimize ads
for the position on the search results. However, an article
published by Lyubomir Popov in
May 2019, illustrated that these metrics are not real-time even after getting enough
impressions for ads and are not able to replace Average Position metric from an optimization
(2019, November 19). Bid to Impression Share - Manager Account | Google Ads scripts. Retrieved May 9,
2020, from
(2019, February 26). Prepare for average position to sunset - Google Ads Help. Retrieved May 9, 2020,
(2019, May 15). How Google will kill manual bidding or Prepare for average position to sunset in Google
Ads - Medium. Retrieved May 9, 2020, from
Figure 5:Screenshot by Lyubomir Popov to prove that Search Impression Share, Search Top
Impression Share and other related metrics are not updated real-time.
This delay and non-real-time data decrease the potential performance improvements of
manual edits; therefore, make advertisers use Smart bidding instead. A study by
Clixmarketing proved that Target Impression Share bidding strategy is effective to decrease
CPCs and cost per leads when keeping at a lower Target Impression Share of 90%.
Disadvantages of Target Impression Share bidding strategy:
Not focusing on conversion: As the main objective is focusing on the positioning on the
search result page, this strategy does not take into consideration getting more conversions or
clicks within the budget which can drive traffic in time and on a device in which conversion
probability is low and getting clicks are expensive.
Budget constraints: Advertisers should be careful when choosing this bidding strategy
because the daily budget can be over during the day without getting as many clicks as
possible for the sake of being on the 1st or top position on the page. Additionally, contextual
(2018, November 29). Target Impression Share Bidding: What We've Seen So Far. Clixmarketing.
Retrieved May 9, 2020, from
signals which are related to conversions are not considered, and the advertising expenses may
exceed revenue, therefore negatively impact ROI.
Maximize Clicks
This bidding strategy focuses on getting as many clicks as possible within the campaign
budget, and the main objective is getting clicks.
As Karjaluoto & Leinonen (2009) stated
that the main purpose of search engine advertising is increasing the visibility of websites on
search results and bringing more traffic from search engines to websites, many advertisers
leverage this bidding strategy to get as much traffic as possible in short-term.
Advertisers can set Maximum CPC bid limit in order to prevent spending budget earlier than
expected and get maximum clicks within the specified amount. However, this limit may
negatively affect ad position and the number of clicks the ad receives. Companies who want
to test their website with new users usually use this strategy in order to drive more traffic
from paid search as the main goal of this strategy was determined as “increase site visits” on
Google Ads documentation. Additionally, businesses that have higher conversion rates on
their website trust this strategy and they bid for clicks rather than conversions as they think
they have already found the right patterns of users (search terms, an hour of a day etc.). They
target them based on these preferences. They cannot bid adjustments based on time and day
in this strategy but can leverage Ad schedule features to specify the time to show their ads.
As many factors affect CPC bid for certain keywords, setting max CPC bid limit may restrict
campaigns to spend daily budget when the competition is fierce, and CPC is higher than Max
CPC. Therefore, a regular monitor is needed to adjust bid limits based on historical data and
daily performance goals.
Disadvantages of Maximize Clicks bidding strategy:
Low-quality traffic: Advertisers should choose right keywords which are related to their
business because Maximize Clicks bidding strategy only focuses on clicks and can increase
clicks when lowering conversion rates. In the case of lower conversion rates, advertisers
(n.d.). About Maximize clicks bidding - Google Ads Help. Retrieved May 9, 2020, from
should react because, for the strategy, the goal is achieved, and Machine Learning will not
consider lower conversion rate as a negative signal. Internet users who have not conversion
intents and only surf on the internet may not be interesting for brands who want qualified
clicks and leads on their websites. Therefore, maximize conversions bidding strategy was
introduced as an alternative way to optimize for conversions rather than clicks.
Budget constraints: To find an optimal amount of spending is not possible without
experiments and real data as keyword research tools give an estimation, which may vary
based on seasonality, competition and other factors. For example, if one of the competitor
websites has discount promotion and drives more clicks because of it, competing with this
brand is not costly friendly for the advertiser who uses Maximize Clicks bid.
Using this strategy will result in either spending lower than the daily budget while getting
lower clicks as expected or spending more than usual if the maximum bid is not limited. In
other cases, when a campaign includes both competitive and uncompetitive search keywords,
the algorithm will focus on driving clicks from keywords whose CPC is lower. As a result,
less competitive keywords can be less valuable and cause the low quality of traffic because
the budget is not equally distributed between search keywords. Therefore, using manual CPC
and getting historical data to identify valuable and converting keywords is essential to
organize the campaign keywords before switching to this Smart bidding strategy. Otherwise,
irrelevant keywords will result in a high Bounce Rate (which refers to the percentage of
website visitors who exited after viewing on a page) and low-Quality Score. Additionally,
certain days of the week and hour of a day, devices, age groups may be less valuable for
business and less appropriate target for getting conversions. In that case, spending the part of
the budget to these audiences for the sake of clicks damages overall campaign performance
and is a waste of money.
Maximize Conversions
Maximize conversions is similar to maximize clicks in terms of processing. This strategy
also concentrated on increasing one metric - conversion within the daily budget.
(n.d.). About Maximize conversions bidding - Google Ads Help. Retrieved May 9, 2020, from
Maximize conversions strategy do not have Max CPC bid limit, and one conversion or click
may cost unexpected amounts for advertisers. Additionally, to use this strategy advertiser
should track conversions in order to optimize for them.
Wayfair case study by Google revealed that there was a 75% decrease in CPA (cost per
conversion) by using Smart bidding strategy - conversion optimizer, including programmatic
bidding based on customer signals (location, device, time etc.) and specific target audience
(in-market and similar audiences, contextual keyword targeting).
Maximize Conversion strategies do not require any historical conversion data to optimize,
unlike Target CPA and Target ROAS bidding strategies
, but having historical data helps to
optimize by considering historical contextual signals. The recommended goal is to switch to
Target CPA or Target ROAS after getting certain conversions needed to meet requirements.
Advertisers found this strategy effective for both B2B and B2C companies when having well-
optimized campaigns with historical data, including converting keywords with high-quality
Disadvantages of Maximize Conversion bidding strategy:
Budget constraints: As there is no limit for individual conversion and clicks, in a
competitive industry cost per conversion can be undesirably high, and this strategy can be
unprofitable at the end of the campaign for those who have certain limits for cost per
conversion. It is argued
that if advertisers decrease a daily budget, the machine learning for
conversion optimization is restricted, and eventually, this negatively impacts the overall
conversion performance. Additionally, maximize conversion bidding strategy cannot be
applied to all industry types, therefore requires split testing.
(n.d.). Wayfair Decreases CPA by 75% With Programmatic Bidding on the Google Display Network.
Retrieved May 9, 2020, from
(2019, September 17). Google Ads Tutorials: Intro to Smart Bidding - YouTube. Retrieved May 9, 2020,
(n.d.). About Maximize conversions bidding - Google Ads Help. Retrieved May 9, 2020, from
(n.d.). Advantages & Disadvantages of Maximize Conversions A New Bidding Strategy for Google Ads.
Retrieved May 9, 2020, from
Maximize Conversion Value
Maximize conversion value works the same way as maximizing conversion bidding strategy
while increasing the conversion value rather than conversion metric itself.
The value set by
advertisers determines the actual benefit or revenue derived from the user interaction and
overall marketing strategy heavily based on increasing business profits. Those who want to
increase revenue within the daily budget maximize conversion value bidding strategy and the
main thing is not focusing on the number of conversions but the value brought by these
conversions. As e-commerce websites include products with different prices, using
maximum conversion value, they attempt to get higher value by selling the units with higher
This method decreases the budget constraints because machine learning algorithms now are
able to know the financial outcome of conversions - the transaction revenue.
Disadvantages of Maximize Conversion Value
Not using ROI calculation: Although maximize conversion value bidding strategy uses the
conversion value for the basis of the optimization, it does not consider the return on
investment measurement and does not optimize for actual income derived from the
conversions. Google Ads itself recommends to use Target ROAS if there is a goal to get
higher ROI by mentioning that:
Check your return-on-investment (ROI) goals. If you have an ROI goal for your campaign, such as a
target return on ad spend (ROAS), you may want to add a Target ROAS to your bid strategy.”
Target CPA
Target CPA is another automated bidding strategy helping advertisers to get as many
conversions as possible within their daily budget but limited by cost per conversion bid.
(n.d.). About Maximize conversion value bidding - Google Ads Help. Retrieved May 9, 2020, from
(n.d.). About Target CPA bidding - Google Ads Help - Google Support. Retrieved May 9, 2020, from
Advertisers set a budget and cost per conversion bid for the campaign. Machine Learning
replaces the monitoring and bid adjustment works of PPC managers and chooses the best
combination of signals to achieve desired results. Advertisers set a lower CPA than usual to
achieve better ROI and test the comparative data.
Some case studies showed that Target CPA had a substantial effect on the campaign results:
FRU.PL case study by Google resulted in a 33% lower CPA and 90 increased conversion
rate with Target CPA Smart bidding strategy. In this study, the previous bidding strategy was
Enhanced CPC, and after switching to Target CPA, the advertiser experienced better results.
The study also showed the importance of using Dynamic Search Ads and long-tail search
terms in their campaigns.
Another case study by Google also resulted in a 33% lower CPA
with Target CPA bidding for the international distributor of Toyota.
Experiment by Nordicclick
revealed that Target CPA bidding strategy decreased the cost
per conversion by 40% while maintaining a similar conversion rate for the campaign. In this
case study, Avg. The position was available on Google Ads, and the advertiser found that the
experimental campaign moved the average position down by 0.3 to reduce CPC while also
decreasing the CPA amount to achieve a target CPA.
Another experiment conducted by
revealed that Target CPA was
better than Maximize click bidding strategy for Call Only campaign for one of the local green
cab companies in Michigan. Results are not statistically significant because the sample data
do not support generalized judgement; however, showed that even in a short period of time
(n.d.). FRU.PL boosts conversions while lowering CPA with smart bidding strategy. Retrieved May 9,
2020, from
(n.d.). Automated Smart Display Campaigns Help Toyota Cut CPA by 33%. Retrieved May 9, 2020, from
Garrett Taylor (2018, July 24). PPC Testing: A Case for Target CPA and Target ROAS. Nordclick.
Retrieved May 9, 2020, from
(2018, June 14). PPC Optimization: Enhanced CPC & Maximize Clicks. Retrieved May 9, 2020, from
Target CPA bidding strategy is able to increase profitability.
Figure 6: Case study results from
Google suggested a target CPA of $1.34, and as a result, the cost per conversion decreased,
the conversion rate increased more than 15% for the Experiment campaign using Target CPA
bidding strategy. Especially, cost per conversion and conversion rate were considered to
determine profitability.
Another case study by
included both Display and Search campaigns
resulting with positive outcomes. On the display campaign, the cost per conversion dropped
from 5 EUR to 3.5 EUR with 7.27% conversion rate at the end. On search campaign cost per
conversion decreased by 66% and it was recorded as 13.89 EUR cheaper than the original
campaign over the past 30 days.
Figure 7: Case study results from
Disadvantages of Target CPA bidding strategy
Not using ROI calculation: As Target CPA bidding strategy only focuses on getting
conversions within the specified target cost per conversion, if advertisers do not measure
conversion values and ROI, this strategy can be useless. Actually, this is not a fault of bidding
strategy, but possible shortcoming when the advertiser is not aware of conversion values.
(2019, February 18). Supercharge Your Search Campaign with Target CPA. Retrieved May 9, 2020, from
Short-term ineffectiveness: In the case study of, it was revealed that in
the short-run Target CPA bidding strategy could be found as ineffective, but this is due to
the learning period. In a 30 days period, the bidding strategy showed its effectiveness and
achieved the target cost per conversion.
Target ROAS
This strategy focuses on getting higher conversion value or transaction revenue at the given
target return-on-ad-spend (ROAS). Using real-time data, Google Ads change bids to
maximize conversion value while trying to achieve an average ROAS equal to the target.
Experiment by Nordicclick
showed that Target ROAS bidding strategy increased
conversion rate by 85% while getting 38% more conversions and decreasing the cost per
conversion from $81.25 to $26.22.
FishingBooker case study by Google resulted in a 49% increase in ROAS and 44% shift in
conversion rate by applying the ROAS Smart bidding strategy, including 10 separate tests.
The study also showed that the amount of time spent on manual optimization decreased and
automated bidding delivered better results than in the past.
Another case study by Google also resulted in higher ROAS, which increased by 47%. The
conversion was PC game downloads, and the company used mixed bidding strategies from
Maximize conversions to Target ROAS as a multi-strategy approach.
Target ROAS requires precise calculation and regular adjustments in order to guarantee a
positive ROI from advertising spend. As Google recommended
to adjust those values when
(n.d.). About Target ROAS bidding - Google Ads Help - Google Support. Retrieved May 9, 2020, from
Retrieved May 9, 2020, from
(n.d.). Smart Bidding boosts FishingBooker's ROAS by 49%. Retrieved May 9, 2020, from
(n.d.). Point It Innovates with Flexible Bid Strategies, Boosting Revenue for Clients. Retrieved May 9,
2020, from
"Setting Smarter Search Bids - Google." Accessed 11 May. 2020.
there is a potential change in conversion rates that may affect overall performance during the
seasonal shopping dates, such as Black Friday, Christmas etc. This cannot be considered as
a disadvantage but requires additional attention and consume time, for which third-party
tools, such as Optmyzr offered
an automated solution to manage ROAS and CPA targets
in these periods.
Enhanced CPC
This bidding strategy focuses on getting more conversions from manual bidding by
automatically adjusting bids for clicks when there is a higher likelihood of getting
conversions. Unlike other conversion-based optimization strategies, Enhanced CPC tries to
get lower CPC than max CPC set by the advertiser.
The experiment conducted by
compared the results between
ECPC and Target CPA bidding strategies. A hybrid approach of this experiment was to lower
the maximum CPC (from $1.34 to $1.00) and switch to ECPC bidding. An increase of the
bids without a limit for the higher conversion likelihood and decrease for lower likelihood
were expected before the experiment.
Figure 8:Case study results from
After about a month, the results showed that ECPC lowered the cost per conversion by
35.55% (below $1.00 per call action) and conversion rates increased by more than 11%.
"If you're not updating ROAS and CPA targets frequently, you're missing out" 12 Jun. 2018,
299821. Accessed 11 May. 2020.
(n.d.). About Enhanced CPC (ECPC) - Google Ads Help. Retrieved May 9, 2020, from
(2018, June 14). PPC Optimization: Enhanced CPC & Maximize Clicks - ECPC .... Retrieved May 9, 2020,
Disadvantages of Enhanced CPC bidding strategy
Budget constraints: Inconsistency with CPC bid in Enhanced CPC creates a difficulty for
PPC managers as any change due to the conversion probability results in different budget
spend at the end of the campaign period. However, manual CPC guarantees the targeted result
without any fluctuation.
Overall, using Smart bidding strategy enables users to automate their advertising campaigns
and create opportunities to focus on strategic and creative parts of advertising, such as
focusing on ad content, competitor analysis etc. Mentioned shortcomings of Smart bidding
strategies prove that they still need improvements and do not fit every specific objective as
long-term and short-term objectives of companies changes and sometimes there is a need to
switch from one to another strategy in the campaign period.
Lack of manual control created a gap in the market; therefore, several third-party tools were
introduced for complete management of Google Ads campaigns with customized automation
solutions. The implementation of AI and ML is still in the development phase and need more
time to develop in order to satisfy customer needs. As more variables impact on campaign
performance, more personalized solutions may emerge on the PPC market.
Figure 9: Relevant KPIs for each Smart Bidding Strategy
Google Ads provided a guideline for advertisers to leverage Smart Bidding Strategies based
on relevant KPIs, which fit the AIDA sales funnel model. Each strategy focuses on one
specific objective to achieve, and machine learning was trained to maximize effectiveness
for this goal.
Figure 10: Lifecycle of Smart Bidding Strategies
Google Ads also provided information about the lifecycle of Smart Bidding Strategies that
require control and learning periods in the beginning for a machine learning model. It is
recommended to wait for enough data to make a decision after a certain period of time.
(n.d.). How the profit driven marketer bids to win - Think with Google. Retrieved May 15, 2020, from
(n.d.). How the profit driven marketer bids to win - Think with Google. Retrieved May 15, 2020, from
To sum up, as advantages smart bidding strategies
- save the time of PPC manager and enable them to spend more time on strategic planning
and creatives tasks
- provide "set and forget" method for some instances which is helpful to focus on other
complex campaigns (while requiring regular control)
- provide a more data-driven approach by using Google Ads platform data through the web
- process and analyze previous data more wisely from a statistical perspective
At the same time shortcomings can be mentioned as:
- Lack of knowledge about the business’s long-term perspective and digital objectives in
detail except for chosen goals and conversions on advertising platforms and analytics tools,
- broad data of machine learning that cannot be specific for some cases
- limited campaign objectives (for example doesn't prove to get more 100% watches on video
campaigns and optimize for this metric)
- inappropriate for brand-new innovative businesses for products which are designed to cover
specific needs, requiring more complex combinations of contextual signals and no or little
data available for such products.
5. Third-party Automation Tools
5.1 Current situation and opportunities
Smart bidding can be perceived as "set and forget" for some PPC managers because it does
auction-time bidding based on given contextual signals in real-time. However, although
Machine Learning technology is better than humans in processing and analysing large
amounts of data, it is unable to process unpredictability, and it cannot solve problems that
have never been seen before. For example, any flash sale of competitors or weather
conditions that affect market prices, sales of advertisers will damage the results of
automatically managed campaigns. Any flash sale ads of competitors will be perceived as
competition signals and result in an increase in the advertising budget. As a result, advertisers
unwillingly will spend more on certain competitive keywords, which are not seasonal and
predicted change in advance. Additionally, revenue in some businesses is dependent on
weather conditions, such as resort hotels in mountains that are visited in winter or beach
places that become crowded in summer. It is predictable that seasonally people are more
interested in going on such vacations and businesses have historical customer data and expect
a certain number of visitors, and based on this data, they invest on advertising campaigns.
However, any change in weather condition, which can also cause natural disaster or even
without disaster the change can decrease the demand of such vacations will result in a lower
return on ad spend or even loss of the whole budget. In such cases, the manual changes to the
advertising campaign are required in order to avoid waste and budget loss. However, this
manual review and edits decrease the importance of Smart bidding and perception of “set
and forget”, increasing demand for additional automation to make changes on campaigns
based on weather conditions by using API to get real-time data from weather forecast
Above-mentioned needs and concerns derived from lack of knowledge of machine learning
systems about business situations and unpredictability problems lead businesses and agencies
to build their own model in order to interpret situational need into algorithms and scripts on
Google Ads platform.
One study (Roman V. 2019) shows that it is impossible to give accurate foresight of specific
actions taken by the smarter-human intelligent system even if the terminal objectives are
acknowledged by humans. Therefore, Unpredictability of AI can cause problems in PPC
management in an unknown and dynamically changing environment, which is largely
affected by external factors. Because it is sometimes difficult to analyse statistical results by
digital marketers with a reasonable degree of confidence, it is also impracticable to follow
each decision-making step and understand the reasons for the judgement of Artificial
Intelligence on Search Engine Advertising platform as it is a black box unknown for
advertisers. PPC managers need to have more control in order to give a detailed report to the
company when any undesirable results, such as overspending, lower conversion rate
occurred. Therefore, internal tools and third-party management tools enter the market and
are largely used by PPC agencies to automate reporting, bid management, optimization, ad
creation etc. These tools give more control and don't require spending more to optimize
campaigns automatically like Google Smart bidding. In contrast, they require more manual
work but with higher accuracy and control to change and react to the market situation.
Some of the advantages of building or using internal and third-party PPC management tools
are as follow:
Finding best practices with pattern recognition: For example, comparing different text
ads and CTA's using historical data help to find the most effective advertising message and
patterns that result in higher ROI in PPC campaigns. This is also applicable to banner ads to
identify the correlation between parameters (device, audience, topic, etc.) and colours or
other variables on the banner content. Bousquet et al., (2018) also used the context variables
(ad size, ad type) and device, aiming to show a relevant ad to the right person, at the right
time and place.
More custom columns: By using API, agencies can build custom columns on internal
systems or can use third-party services to provide it. Custom columns are limited and even
are not available on some dashboards in Google Ads. However, PPC manager can create their
ratios and different formulas via API. Customized dashboards for each business make the
human monitoring process easy and custom ratios and calculations can be used for automated
optimization and data analysis.
Customized KPIs: As PPC managers give a consultation services to businesses by
specifying their KPIs based their short-term and long-term objectives, integrating given
objectives with needs and wants of customers, building a custom strategy after analyzing
competition, in the same way, third-party tools are available to introduce different strategy
types and long-term plans with experiment plans by analyzing previous experiences of other
accounts inside the platform and can build new algorithms derived from big data analysis by
using machine learning pattern recognition algorithm. There can be privacy concerns of
advertising accounts about sharing their data; however, without explicitly using the business
data and sharing to another party, third-party tools can leverage their client’s database to
develop their products.
Data collection: PPC managers can create their own datasets, and by using cloud technology,
they can store and build a custom ML model. Although it requires data science and data
engineering skills and expertise, this process will contribute the competitiveness in the
industry. Park (2020) demonstrated a deep learning-based algorithm for the system which
automates predictions about search advertising price. This study pointed out the importance
of collecting massive amounts of data in a cloud-based server. Additionally, the importance
of further researches for dynamic bidding models was discussed.
Li et al. (2019) found that Keyword level bidding has a higher risk with higher profits
compared to Ad group level bidding. As the budget increases, there is a decrease in marginal
profit and marginal risk. This study used an integrated bidding strategy, including both
methods and further research was required to build a dynamic model, showing that there is a
gap to fill for advertisers and ad platforms in the bidding process.
Usually, PPC managers themselves are able to easily analyze single and even two-
dimensional data. However, when there are several variables or dimensions to consider and
analyze for optimization, the process becomes complicated and overwhelming. In this case,
machine learning and user-generated solutions become alternatives. There are additional
dimensions that create difficulty to optimize as follows:
Genetic algorithms help to allocate the budget across geo-locations for each campaign,
decreasing CPC while increasing return on ad spends (ROAS). Location as a vertical affects
the CPA targets of advertisers as stores in different locations may have different profit
margins and auction competitiveness vary for different location targeting.
In the case of weather changes, many restaurants are affected by this factor. For example, a
restaurant with terrace and outdoor seats mostly have few visitors on rainy days as clients do
not want to sit outside. However, clients on summer days prefer outdoor and restaurants to
increase their sales. How can Google machine learning detect this factor and sophisticatedly
allocate budget based on the weather factor?
Another example related to weather is the effect of temperature on people's desire to visit
car dealerships. Are people prone to stay outside to wait in a queue? In that case, what is the
break-even point? What degree of temperature (4 or 7) affect the conversions? How much
should PPC managers adjust the budgets and bids for each Celsius change in temperature?
Obviously, the temperature may affect clicks and eventually sales; therefore, the relevance
of those determinants should be considered when making bidding decisions.
PPC managers are not able to manually analyze all data of clients’ accounts. In order to
identify such relationships, large datasets should be prepared for analysis and monitoring. In
supervised machine learning, the system automatically analyzes existing data and based on
the input; it finds similarities in historical data. Eventually, machine learning gives new
insights into clusters and patterns. Similar audience” as a targeting feature offered by
Google is one of these examples; however, the relationship between different dimensions and
variables is quite a different topic.
Each data source tracks the different parts of the customer journey and combining all user
interactions and signals is difficult and costly to manage, organize and optimize. Therefore,
several third-party services emerged, offering optimized bidding strategies to fill the gap and
introduced unique solutions.
For example, third party tool QuanticMind use custom machine learning algorithms,
Bayesian modelling, natural language processing, predictive analytics methods to optimize
search advertising campaigns for specific business objectives.
In the Mitsubishi case study presented by blue rank, one of the main objectives included
minimizing complexity and time of monthly budget and campaign planning. Another goal
"Machine Learning Powered PPC Optimization - QuanticMind"
machine-learning-powered-ppc-optimization-the-modern-methodology-to-best-in-class-bidding/ Accessed 16
May. 2020.
was improving cost allocation more strategically in order to decrease CPC amount by
eliminating internal competition between car dealers.
Another automation tool by Takesomerisk helps PPC managers to audit their accounts and
automate management tasks. Tool checks targeting settings (locations, networks, etc.),
budget spend, ad schedule and other important campaign configurations, and it also provides
additional reports for optimization. For example, a custom Search Query Report helps to
identify negative keyword opportunities, which is not provided by Google Ads and only a
customized approach is required to analyze this data.
Shape platform help to manage advertiser budgets across the platforms, such as Bing,
Facebook, Google and Linkedin. Although it does not provide bid management, the platform
includes automation tools from alerts to reports.
A case study published Daisy-ree Quaker on showed that bidding for the third
position via automated Google Ads script resulted in a lower cost per conversion decreasing
from $87 to $16 and higher conversion rate increasing from 6% to 18%. This shows that
changing ad position has a substantial impact on ad performance and even advertisers can
achieve high conversion rates in lower positions within top positions (i.e. first three positions)
and prove the possibility of getting better results with ad script creating an alternative to
Target Impression Share. However, it requires a data-driven experiment to prove this
Another feature of Google Ads is using scripts to automate certain actions inside the ad
platform. The method is similar to using API and getting data from the platform; however,
API is more flexible to use and customize. As many PPC managers lack coding skills, they
prefer to use third-party platforms which offer custom dashboards and ready solutions as
discussed above.
Additionally, Google Ads partially automated many features for advertisers and provided
recommendations and automated rules to decrease the monitoring time and increase the usage
of smart bidding strategies (by offering to use them). According to the latest announcement,
(n.d.). Local Mitsubishi campaigns made simple - Bluerank. Retrieved May 11, 2020, from
(n.d.). Advanced Guide to PPC Account Audits - Take Some Risk Inc.. Retrieved May 11, 2020, from
(n.d.). PPC Budget Management Software - Retrieved May 11, 2020, from
Google Ads require advertisers to achieve +70 optimization points to earn Google Partner
badges for PPC agencies. The optimization score represents the percentage of applying
recommendations, which not only include "one-click process", but also require enriching the
ad content, adding proper ad extensions etc.
Therefore, recent changes make Partner
advertisers apply these recommendations in order to keep their Partnership status while
increasing the usage of recommendations and letting Google Ads system to build more
sophisticated algorithms based on the behaviour of advertisers towards recommendations.
5.2 Impacts of Automation for different players in the
Google Ads framework consist of the three main actors: ad agency, an advertiser (a particular
business that promotes its product or services), the customer (online users). The customer
navigates through web and mobile applications which refer to the ad placement of publishers
and advertisers target these users on chosen placements specified in campaign level. In this
scenario, Google Ads is an intermediary service connecting these actors and all of the benefit
from the service.
Figure 11:Players and roles in the PPC industry by Li et al. (2016)
(n.d.). How to earn the Google Partners badge - Google Ads Help. Retrieved May 11, 2020, from
Li et al., (2016) illustrated the Ad agency’s roles in the institutional context including players
- advertiser business (retailer’s website), ad agency, a consumer (online user), search engine
(ad platform). This illustration also includes the revenue estimation for individual keywords
using attribution metrics as the study focused on the effects of attribution strategies on ROI.
Obviously, ad agencies are responsible for the advertising expenses - the investments for
each individual campaign. The IT-based high-technology industries may always experience
technology and market uncertainty and competitive volatility, which make it difficult to
continue with consistency.
PPC managers (individuals or agencies)
PPC managers are able to automate many tasks, including reporting, monitoring, optimizing
etc. The main area is an optimization which includes data analysis and prediction for better
financial outcomes for the client businesses. As PPC agencies satisfy their clients with data-
driven decisions and sophisticated budget allocation, businesses consider them as reliable
service and preferable than traditional marketing channels which are not as measurable as
digital marketing.
Businesses (advertiser companies)
Businesses can leverage automation tools via PPC agencies and increase their efficiency. It
also creates an additional opportunity to integrate automation tools with their existing
marketing channels to get automated reports via APIs. Additionally, the decrease in human
errors positively impacts the PPC audit of the agency's activity by creating a trust and reliable
partnership. However, all results depend on the expertise and skills of PPC managers who
use automation solutions both by Google and third-party service.
The main focus of advertisers on ad platforms is different in terms of budget allocation, and
each ad platform gets various proportions of total advertising costs from businesses. When
the reserve prices in the auction are low, advertisers allocate their budget asymmetrically
between ad platforms, and therefore, partial differentiation helps advertisers reduce their bid
competition, lower the cost and increase profits. Additionally, bid jamming occurs in each
ad platform when one advertiser increasingly raises bid cost to $0.01 below the competitor
who has set their maximum bid cost too high. This bidding behaviour exhausts competitor's
budgets, and eventually, lower bidders earn more clicks with the lowest CPC amount after
moving up. (Zia & Rao, 2019)
As so many small businesses usually encounter serious budget constraints (Yang et al., 2014;
Yang et al., 2015), there is always a competition among rivals. Shin W. (2015) stated that in
order to eliminate budget-constrained rivals in the auction, advertisers raise the bids as much
as their marginal profit. The analysis of this study showed that advertisers not only raise the
competitor's costs but also intend to maximize its own profits by using this aggressive bidding
strategy. Surprisingly, this study also found that budget-constrained advertisers may set a
higher valuation. Competitors dynamically change their bidding strategy to minimize the
high-valuation of competitors' budget and get a higher position in the bidding game. Given
evidence shows the importance of using dynamic bid adjustments and optimizations to
survive in the competition and increase the effectiveness of PPC campaigns. Therefore,
machine learning algorithms create a new market for agencies to provide dynamic
optimization tools for businesses who want to leverage innovations to integrate their
marketing channels.
Ad Platforms and Users
The current situation increases transparency making advertising content available in more
appropriate places in terms of contextual signals. As ad platforms’ targeting capabilities
enhance, people are bombarded by several advertising contents during a day, and ad publisher
websites encounter reliability problems. Precise targeting by finding a relevant audience and
getting website traffic or visibility in the right time and on the right context improve the
relationship between online users and ad publishers because people think that advertisements
are helpful to find products and services to cover needs. However, there is always a risk that
users may complain about more personal advertising messages by advertisers who found the
specific needs of customers by using data about their online activity. Therefore, in recent
years GDPR compliance made it hard to track every single action of users without their
permission. Goldberg et al. (2019) found that online e-commerce businesses experienced a
drop in revenue and online traffic after GDPR’s enforcement. This makes it necessary to
adopt machine-learning advancements compliance with user privacy in order to maintain
effective advertising results for both parties. Otherwise, users attempt to use adblockers and
block data tracking systems which make it impossible to use personalized advertising.
6. Conclusion
The study analysed the efficiency of PPC advertising in bringing new customers to the
company, and according to previous studies, it was proved that many companies still use
PPC advertising to attract customers online during their research about products or services.
In the shadow of current technological innovations, it is undeniable that Smart bidding
strategies and third-party automated campaign optimization tools decrease human
intervention and eliminate human errors in the process of managing PPC campaigns. Existing
third-party tools and competition between PPC management tools show the current demand
of using automation tools for ad creation, reporting, monitoring and optimization by PPC
managers, However, it is questionable whether the intelligence of AI may consider the impact
of characteristics of the business and market changes on the advertising campaigns
dynamically and apply changes by analyzing the market as PPC manager do. As AI and
machine learning work with data, small businesses still should work on integrating their
marketing activities and collect the data from different channels for better optimization.
Additionally, it was founded that there is a risk for PPC management tools as Google
regularly improve and educate the Machine Learning algorithms with more data and
algorithm changes. Since the last requirements for Google Partners, recommendation feature
makes Partners use Smart bidding strategies more often to meet the requirements and
eventually, this provides ML with more experience. As a result, Google’s Machine Learning
models are improved and get a chance to give higher performance results or KPI
achievements to advertisers. Smart bidding strategies were used by thousands of advertisers
and made it easy to target users and optimize campaigns based on the target metrics KPIs.
Well-designed guidance for advertisers educates them on using the right KPIs based on their
target and Machine-Learning support the optimization strategy. The variety of choices in
terms of bidding strategies enabled PPC managers to enter different auctions, target different
segments of users based on contextual signals and give control to Machine Learning to
analyze and optimize historical performance after a certain learning period. Overall,
switching Smart bidding strategies and using parallel experiments help advertisers identify
their best performing approach to the campaigns as case studies proved the efficiency of
using these experiments. However, third-party and custom internal PPC management tools
are still required as uncertainties exist for a business requiring dynamic changes and
customized solutions. Uncertainties and having a lack of data from outside, Google Ads
Smart Bidding Strategy may have trouble with optimizing campaigns for businesses which
serve in a dynamic market environment and encounter unpredictable consequences. This
study acknowledged the effectiveness of Smart bidding as many businesses leveraged this
machine learning-based advancements while finding gaps in the industry which was filled
by agencies and third-party service providers. Uncertainties always exist, and custom
variables that affect the performance metrics are required to be considered when managing
advertiser accounts. In this case, the cases discussed in the study helped to describe current
situation and opportunities to use custom solutions for businesses.
7. Limitations and Further Research
This paper aimed to create a foundation for further academic researches - articles, papers
with quantitative and market research methods to highlight the impact of automated
solutions. By nature, many arguments were still not investigated by comparative research to
analyze the possible impact of automated solutions on PPC agencies and advertiser
businesses. As agencies use these tools for promotion, the credibility of given case study
results and promises do not reflect the statistically significant positive impact of automated
solutions on PPC management. Further studies are required to prove the effectiveness of each
Smart Bidding Strategy by using primary data of businesses and should point specific
problems which were solved by automated strategies. The thesis recommends that further
empirical studies with large primary datasets are required for more credibility because the
analysis of this thesis is mainly based secondary data, although blogs and credible source by
Google Ads introduced successful case studies and third-party tools are the sign of the need
for custom automated solutions. Further studies should mainly focus on return on investment
measurement when applying Smart Bidding Strategies or any third-party PPC management
software. The possible experiments can be conducted by using drafts/experiments feature on
Google Ads while integrating the web analytics tools, such as Google Analytics to evaluate
the effect of experiment on the behaviour of the users obtained from different bidding
8. References:
Abdurrahim, M. S., Najib, M., & Djohar, S. (2019). Development of AISAS Model to
See the Effect of Tourism Destination in Social Media. Management (JAM), 17(1).
Abou Nabout, N., Skiera, B., Stepanchuk, T., & Gerstmeier, E. (2012). An analysis of
the profitability of fee-based compensation plans for search engine marketing.
International Journal of Research in Marketing, 29(1), 6880
Agarwal, A., Hosanagar, K.., & Smith, M. D. (2011). Location, location, location: An
analysis of profitability of position in online advertisement markets. Journal of
Marketing Research. 48(6). 1057-1073.
Alphabet Announces Fourth Quarter and Fiscal Year 2019 Results -
Anderson, C. (2006). The long tail: Why the future of business is selling less of more.
Hachette Books.
Arya, V., Sethi, D., & Paul, J. (2019). Does digital footprint act as a digital asset?
Enhancing brand experience through remarketing. International Journal of Information
Management, 49, 142-156.
Ashman, R., Solomon, M. R., & Wolny, J. (2015). An old model for a new age:
Consumer decision making in participatory digital culture. Journal of Customer
Behaviour, 14(2), 127-146.
Ayanso, A., & Karimi, A. (2015). The moderating effects of keyword competition on
the determinants of ad position in sponsored search advertising. Decision Support
Systems, 70, 42-59.
Banerjee, A. 1992. A simple model of herd behavior. Quart. J. Econom. 110 797817
Bao Lam & Koppel Verma (2014) "New Research Shows Search Ads Drive Brand
Bartz, Kevin, Vijay Murthi, and Shaji Sebastian (2006), “Logistic Regression and
Collaborative Filtering for Sponsored Search Term Recommendation" in Second
Workshop on Sponsored Search Auctions in Conjunction with the ACM Conference on
Electronic Commerce. Available at:
f Accessed: 15/05/2020
Bayer, E., Srinivasan, S., Riedl, E. J., & Skiera, B. (2020). The impact of online display
advertising and paid search advertising relative to offline advertising on firm
performance and firm value. International Journal of Research in Marketing. Available
at: Accessed:
Bayoude, K., Ouassit, Y., Ardchir, S., & Azouazi, M. (2018). How machine learning
potentials are transforming the practice of digital marketing: State of the art. Periodicals
of Engineering and Natural Sciences, 6(2), 373-379.
Bergkvist, L., Eiderbäck, D., & Palombo, M. (2012). The brand communication effects
of using a headline to prompt the key benefit in ads with pictorial metaphors. Journal of
Advertising, 41(2), 67-76.
Bessen, J. E. (2016). How computer automation affects occupations: Technology, jobs,
and skills. Boston Univ. school of law, law and economics research paper, (15-49).
Bettman, J.R., & Park, C.W. (1980). Effects of prior knowledge and experience and
phase of the choice process on consumer decision processes: A protocol analysis.
Journal of Consumer Research, 7(3), 234-238. doi: 10.1086/208812
Bidding Patterns for Search Advertising,” Marketing Science, 29, 2, 199–215
Bousquet, F., Duong, K., Lavergne, C., Lebre, S., & Lieva, A. (2018). User response
prediction in mobile advertising. Available at:
02014821/document Accessed: 15/05/2020
Broder, A., Gabrilovich, E., Josifovski, V., Mavromatis, G., & Smola, A. (2011,
February). Bid generation for advanced match in sponsored search. In Proceedings of
the fourth ACM international conference on Web search and data mining (pp. 515-524).
Brooks, N. (2004a). The Atlas rank report: How search engine rank impacts traffic.
Retrieved from
Cary, M., Das, A., Edelman, B., Giotis, I., Heimerl, K., Karlin, A. R., ... & Schwarz, M.
(2007, June). Greedy bidding strategies for keyword auctions. In Proceedings of the 8th
ACM conference on Electronic commerce (pp. 262-271).
Chan, T. Y., Wu, C., & Xie, Y. (2011). Measuring the lifetime value of customers
acquired from google search advertising. Marketing Science, 30(5), 837850
Chatwin, R.E. An overview of computational challenges in online advertising. In 2013
American Control Conference. Washington, DC: IEEE, 2013, pp. 59906007.
Chen, F. Y., Chen, J., & Xiao, Y. (2007). Optimal Control of Selling Channels for an
Online Retailer with Cost‐per‐Click Payments and Seasonal Products. Production and
Operations Management, 16(3), 292-305.
Chen, J., & Stallaert, J. (2010). Who Welcomes Behavioral Targeting: An Economic
Chen, M., & Rabelo, L. (2017). Real Time Bidding Optimization in Computational
Advertising. In IIE Annual Conference. Proceedings (pp. 175-180). Institute of
Industrial and Systems Engineers (IISE).
Darley, W. K., Blankson, C., & Luethge, D. J. (2010). Toward an integrated framework
for online consumer behavior and decision making process: A review. Psychology &
marketing, 27(2), 94-116.
Das Sharma, A., Gujar, S., & Narahari, Y. (2012). Truthful multi-armed bandit
mechanisms for multi-slot sponsored search auctions. Current Science, 103(9), 1064-
David, H. J. J. O. E. P. (2015). Why are there still so many jobs? The history and future
of workplace automation. Journal of economic perspectives, 29(3), 3-30.
Dellarocas, C. (2012). Double marginalization in performance-based advertising:
Implications and solutions. Management Science, 58(6), 1178-1195.
Deng, L., & Poole, M. S. (2010). Affect in web interfaces: a study of the impacts of
web page visual complexity and order. Mis Quarterly, 711-730.
Donnellan B., M. Helfert, J. Kenneally, D. Vandermeer, M. Rothenberger, R. Winter,
(2015), New Horizons in Design Science: Broadening the Research Agenda, Springer.
Drolias , B. (2007). Pay-Per-Click: The Complete Guide. London: Lulu.
Du, X., Su, M., Zhang, X., & Zheng, X. (2017). Bidding for multiple keywords in
sponsored search advertising: Keyword categories and match types. Information
Systems Research, 28(4), 711-722.
Ellam, A. (2004). Overture and Google: Internet Pay-Per-Click (PPC) Advertising
Auctions. Accessed 24 July 2014
Even Dar, E., Mirrokni, V. S., Muthukrishnan, S., Mansour, Y., & Nadav, U. (2009,
April). Bid optimization for broad match ad auctions. In Proceedings of the 18th
international conference on World wide web (pp. 231-240).
Fan, M., Guo, J., Zhu, S., Miao, S., Sun, M., & Li, P. (2019, July). Mobius: Towards
the next generation of query-ad matching in Baidu's sponsored search. In Proceedings
of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data
Mining (pp. 2509-2517).
Farris, P. W., Bendle, N. T., Pfeifer, P. E. and Reibstein, D. J. (2010). Marketing
metrics: The definitive guide to measuring marketing performance. New Jersey:
Pearson Education.
Feng, J., Bhargava, H., & Pennock, D. (2007). Implementing sponsored search in web
search engines: Computational evaluation of alternative mechanisms. INFORMS
Journal on Computing, 19(1), 137148. doi: 10.1287/ijoc.1050.0135
Fulgoni, G. M., & Mörn, M. P. (2009). Whither the click? How online advertising
works. Journal of Advertising Research, 49(2), 134-142.
Gefen, D. and Straub, D. (2000) The Relative Importance of Perceived Ease of Use in
IS Adoption a Study of E-Commerce Adoption. Journal of the Association for
Information Systems, 1, 1-28.
Ghose, A., & Yang, S. (2009). An empirical analysis of search engine advertising:
Sponsored search in electronic markets. Management science, 55(10), 1605-1622.
Girdher, S. (2019) Role of Artificial Intelligence in Transforming E-commerce Sector.
Goldberg, S., Johnson, G., & Shriver, S. (2019). Regulating Privacy Online: The Early
Impact of the GDPR on European Web Traffic & E-Commerce Outcomes. Available at
SSRN 3421731.
Goldfarb, A., & Tucker, C. (2011). Online display advertising: Targeting and
obtrusiveness. Marketing Science, 30(3), 389404
Goodwin, K. (2001). Perfecting your personas. Cooper Interaction Design Newsletter,
19, 295-313.
Gupta, Sonal, Mikhail Bilenko, and Matthew Richardson (2009), “Catching the Drift:
Learning Broad Matches from Clickthrough Data,” in Proceedings of the 15th ACM
SIGKDD International Conference on Knowledge Discovery and Data Mining, 1165
Gürbüz, A., Kılıç, İ., & Yeğin, T. (2016). Effects of Remarketing Implementations on
Consumers’ Behaviour. International Journal of Research in Management, Economics
and Commerce, 6(7), 36-45.
Haans, H., Raassens, N., & van Hout, R. (2013). Search engine advertisements: The
impact of advertising statements on click-through and conversion rates. Marketing
Letters, 24(2), 151-163.
Haveliwala, T. H., Jeh, G. M., & Kamvar, S. D. (2012). U.S. Patent No. 8,321,278.
Washington, DC: U.S. Patent and Trademark Office.
Huang, H., Zhao, B., Zhao, H., Zhuang, Z., Wang, Z., Yao, X., ... & Fu, X. (2018,
April). A cross-platform consumer behavior analysis of large-scale mobile shopping
data. In Proceedings of the 2018 World Wide Web Conference (pp. 1785-1794).
Jafarzadeh, H., Aurum, A., D’Ambra, J. and Ghapanchi, A. (2015), “A systematic
review on search engine advertising”, Pacific Asia Journal of the Association for
Information Systems, Vol. 7 No. 3.
Jansen, B. J., & Mullen, T. (2008). Sponsored search: an overview of the
concept, history, and technology. International Journal of Electronic Business, 6(2),
Jansen, B. J., & Clarke, T. B. (2017). Conversion potential: a metric for evaluating
search engine advertising performance. Journal of Research in Interactive Marketing.
Jansen, B. J., & Schuster, S. (2011). Bidding on the buying funnel for sponsored search
and keyword advertising. Journal of Electronic Commerce Research, 12(1), 1.
Jansen, B. J., Booth, D. L., & Spink, A. (2007, May). Determining the user intent of
web search engine queries. In Proceedings of the 16th international conference on
World Wide Web (pp. 1149-1150).
Jansen, B. J., Moore, K., & Carman, S. (2013). Evaluating the performance of
demographic targeting using gender in sponsored search. Information Processing &
Management, 49(1), 286-302.
Jerath, K., Ma, L., & Park, Y. H. (2014). Consumer click behavior at a search engine:
The role of keyword popularity. Journal of Marketing Research, 51(4), 480-486.
Joo, M., Wilbur, K. C., & Zhu, Y. (2016). Effects of TV advertising on keyword search.
International Journal of Research in Marketing, 33(3), 508-523.
Joo, M., Wilbur, K. C., Cowgill, B., & Zhu, Y. (2014). Television advertising and
online search. Management Science, 60(1), 56-73.
Karjaluoto, H., & Leinonen, H. (2009). Advertisers' perceptions of search engine
marketing. International Journal of Internet Marketing and Advertising, 5(1-2), 95-112.
Katona, Zsolt and Miklos Sarvary (2010), “The Race for Sponsored Links:
Keng, C. J. & Lin, H. Y. 2006. Impact of telepresence levels on internet advertising
effects. CyberPsychology & Behavior, 9, 8294.
Keng, C. J., & Lin, H. Y. (2006). Impact of telepresence levels on internet advertising
effects. CyberPsychology & Behavior, 9(1), 82-94.
Khraim, H. S. (2015). The effect of using pay per click advertisement on online
advertisement effectiveness and attracting customers in e-marketing companies in
Kim, A. J., Jang, S., & Shin, H. S. (2019). How should retail advertisers manage
multiple keywords in paid search advertising?. Journal of Business Research.
Kiritchenko, S., and Jiline, M. Keyword optimization in sponsored search via feature
selection. In New Challenges for Feature Selection in Data Mining and Knowledge
Discovery. Antwerp, Belgium: JMLR, 2008, pp. 122134.
Kiros, R., Zhu, Y., Salakhutdinov, R. R., Zemel, R., Urtasun, R., Torralba, A., & Fidler,
S. (2015). Skip-thought vectors. In Advances in neural information processing systems
(pp. 3294-3302).
Klapdor, S., Anderl, E. M., von Wangenheim, F., & Schumann, J. H. (2014). Finding
the right words: The influence of keyword characteristics on performance of paid search
campaigns. Journal of Interactive Marketing, 28(4), 285-301.
Küçükaydin, H., Selçuk, B., & Özlük, Ö. (2019). Optimal keyword bidding in search-
based advertising with budget constraint and stochastic ad position. Journal of the
Operational Research Society, 1-13.
Kwon, C. (2011). Single-period balancing of pay-per-click and pay-per-view online
display advertisements. Journal of Revenue and Pricing Management, 10(3), 261-270.
Lambrecht, A., & Tucker, C. (2011). Executive Summary: The Right Words at the
Right Time. Business Strategy Review, 22(3), 78-79.
Lee, M. C., Gao, B., & Zhang, R. (2018, July). Rare query expansion through
generative adversarial networks in search advertising. In Proceedings of the 24th ACM
SIGKDD International Conference on Knowledge Discovery & Data Mining (pp. 500-
Lee, M., Kwahk, J., Han, S. H., Jeong, D., Park, K., Oh, S., & Chae, G. (2020).
Developing personas & use cases with user survey data: A study on the millennials’
media usage. Journal of Retailing and Consumer Services, 54, 102051.
Lewandowski, D. (2013). Credibility in web search engines. In Online credibility and
digital ethos: Evaluating computer-mediated communication (pp. 131-146). IGI Global.
Lewis, R. A., & Reiley, D. H. (2014). Online ads and offline sales: measuring the effect
of retail advertising via a controlled experiment on Yahoo!. Quantitative Marketing and
Economics, 12(3), 235-266.
Li, H., Kannan, P. K., Viswanathan, S., & Pani, A. (2016). Attribution strategies and
return on keyword investment in paid search advertising. Marketing Science, 35(6),
Li, H., Lei, Y., & Yang, Y. (2019, August). Bidding Strategies on Adgroup and
Keyword Levels in Search Engine Advertising: A Comparison Study. In Proceedings of
the 2019 2nd International Conference on Information Management and Management
Sciences (pp. 23-27).
Li, J., Ni, X., & Yuan, Y. (2018). The reserve price of ad impressions in multi-channel
real-time bidding markets. IEEE Transactions on Computational Social Systems, 5(2),
Lilienthal, M., & Skiera, B. (2010). Decomposing the effect of increases in search
advertising expenditures on prices-per-click and number of clicks. Arbeitspapier,
Goethe Universität, Frankfurt am Main.
Liu, D., Chen, J., & Whinston, A. B. (2010). Ex ante information and the design of
keyword auctions. Information Systems Research, 21(1), 133-153.
Lobschat, L., Osinga, E. C., & Reinartz, W. J. (2017). What happens online stays
online? Segment-specific online and offline effects of banner advertisements. Journal of
Marketing Research, 54(6), 901-913.
Lowry, P. B., Gaskin, J., & Moody, G. D. (2015). Proposing the multi-motive
information systems continuance model (MISC) to better explain end-user system
evaluations and continuance intentions. Journal of the Association for Information
Systems, 16(7), 515-579.
Lu, X., and Zhao, X. Differential effects of keyword selection in search engine
advertising on direct and indirect sales. Journal of Management Information Systems,
30, 4 (2014), 299326.
Mangani, A. (2004). Online advertising: Pay-per-view versus pay-per-click. Journal of
Revenue and Pricing Management, 2(4), 295-302.
Matz, S. C., Kosinski, M., Nave, G., & Stillwell, D. (2017). Psychological targeting as
an effective approach to digital mass persuasion. Proceedings of the National Academy
of Sciences, 114(48), 1271412719
Miaskiewicz, T., & Kozar, K. A. (2011). Personas and user-centered design: How can
personas benefit product design processes?. Design studies, 32(5), 417-430.
Monetate. (November 24, 2019). Conversion rate of online shoppers worldwide as of
2nd quarter 2019, by platform [Graph]. In Statista. Retrieved May 04, 2020, from
Mukherjee, P., & Jansen, B. J. (2017). Conversing and searching: the causal
relationship between social media and web search. Internet Research.
Mukhopadhyay, D., A Banik, S. Mukherjee, J. Bhattacharya, and Y. Kim. 2007. “A
domain spe-cific ontology based semantic web search engine.” In 7th International
Workshop MSPT2007 Proceedings, 8189. Youngil Publication: Republic of Korea
Nadjla Hariri, Maryam Asadi, Yazdan Mansourian, (2014) "The impact of users’
verbal/imagery cognitive styles on their Web search behavior", Aslib Journal of
Information Management, Vol. 66 Issue: 4, pp.401-423
Narayanan, S. and K. Kalyanam (2015). “Position Effects in Search Advertising and
their Moderators: A Regression Discontinuity Approach.” Marketing Science 34 (3),
Nguyen, B. N., Meador, M. A. B., Medoro, A., Arendt, V., Randall, J., McCorkle, L., &
Shonkwiler, B. (2010). Elastic behavior of methyltrimethoxysilane based aerogels
reinforced with tri-isocyanate. ACS applied materials & interfaces, 2(5), 1430-1443.
Nunan, D., & Knox, S. (2011). Can search engine advertising help access rare
samples?. International Journal of Market Research, 53(4), 523-540.
Olivier, N. P. H. E. (2016). Effects of content on Google ad success: the case of
Icelandair (Doctoral dissertation), 62
Palmer, A. (2000). Principles of Marketing. Oxford: Oxford University Press.
Papadimitriou, P., Garcia-Molina, H., Krishnamurthy, P., Lewis, R. A., & Reiley, D. H.
(2011, August). Display advertising impact: Search lift and social influence. In
Proceedings of the 17th ACM SIGKDD international conference on Knowledge
discovery and data mining (pp. 1019-1027).
Park, C. H., & Park, Y. (2014). Investigating purchase conversion by uncovering online
visit patterns. Working paper. Binghamton University.
Park, H. (2020). MLP modeling for search advertising price prediction. Journal of
Ambient Intelligence and Humanized Computing, 11(1), 411-417.
Park, H., & Cho, H. (2012). Social network online communities: information sources
for apparel shopping. Journal of Consumer Marketing, 29(6), 400-411.
Pashkevich, M., Dorai-Raj, S., Kellar, M., & Zigmond, D. (2012). Empowering online
advertisements by empowering viewers with the right to choose: the relative
effectiveness of skippable video advertisements on YouTube. Journal of Advertising
Research, 52(4), 451-457.
Poyraz, E., & Çetintürk, N. (2017). Defining search engine advertising metrics
according to AIDA advertising model.
Purcell, K., Rainie, L., & Brenner, J. (2012). Search engine use 2012.
Qiao, D., Zhang, J., Wei, Q., & Chen, G. (2017). Finding competitive keywords from
query logs to enhance search engine advertising. Information & Management, 54(4),
Qin, R., Ni, X., Yuan, Y., Li, J., & Wang, F. Y. (2017, October). Revenue models for
demand side platforms in real time bidding advertising. In 2017 IEEE International
Conference on Systems, Man, and Cybernetics (SMC) (pp. 438-443). IEEE.
Qin, R., Yuan, Y., & Wang, F. Y. (2019, November). Exploring Optimal Revenue
Models For DSPs In Real Time Bidding Advertising. In 2019 IEEE International
Conference on Service Operations and Logistics, and Informatics (SOLI) (pp. 181-185).
Ravi, S.; Broder, A.; Gabrilovich, E.; Josifovski, V.; Pandey, S.; and Pang, B.
Automatic generation of bid phrases for online advertising. In Proceedings of the Third
ACM International Conference on Web Search and Data Mining. New York: ACM,
2010, pp. 341
Regelson, M., & Fain, D. (2006). Predicting click-through rate using keyword clusters.
In Proceedings of the Second Workshop on Sponsored Search Auctions. Available:
df Accessed: 08/10/2019
Riley, J. (2012) Buyer behavior - The decision-making process.
making-process Accessed on 22 April 2020
Rutz, O. J. (2011). Zooming in on paid search ads-a consumer-level model calibrated
on aggregated data. Marketing Science, 30(5), 789-800.
Rutz, O. J., & Bucklin, R. E. (2011). Does banner advertising affect browsing for
brands? Clickstream choice model says yes, for some. Quantitative Marketing and
Economics, 10(2), 231257
Rutz, O. J., & Bucklin, R. E. (2011). From generic to branded: A model of spillover in
paid search advertising. Journal of Marketing Research, 48(1), 87-102.
Rutz, O. J., Bucklin, R. E., & Sonnier, G. P. (2012). A latent instrumental variables
approach to modeling keyword conversion in paid search advertising. Journal of
Marketing Research, 49(3), 306-319.
Rutz, O. J., Trusov, M., & Bucklin, R. E. (2011). Modeling indirect effects of paid
search advertising: Which keywords lead to more future visits?. Marketing Science,
30(4), 646-665.
Sayedi, A., Jerath, K., & Srinivasan, K. (2014). Competitive poaching in sponsored
search advertising and its strategic impact on traditional advertising. Marketing
Science, 33(4), 586608
Scaiano, M., & Inkpen, D. (2011, September). Finding negative key phrases for internet
advertising campaigns using wikipedia. In Proceedings of the International Conference
Recent Advances in Natural Language Processing 2011 (pp. 648-653).
Schäfer, C., Zinke, R., Künzer, L., Hofinger, G., & Koch, R. (2014). Applying Persona
method for describing users of escape routes. Transportation Research Procedia, 2, 636-
Scholz, M., Brenner, C., & Hinz, O. (2019). AKEGIS: automatic keyword generation
for sponsored search advertising in online retailing. Decision Support Systems, 119, 96-
Shin, W. (2015). Keyword search advertising and limited budgets. Marketing Science,
34(6), 882-896.
Skiera, B., Eckert, J., & Hinz, O. (2010). An analysis of the importance of the long tail
in search engine marketing. Electronic Commerce Research and Applications, 9(6),
Solomon, M.R. (2015). Consumer behaviour: buying, having, and being,11th edition.
Upper Saddle River, NJ: Pearson Education.
Song, Y., Ma, H., Wang, H., & Wang, K. (2013, May). Exploring and exploiting user
search behavior on mobile and tablet devices to improve search relevance. In
Proceedings of the 22nd international conference on World Wide Web (pp. 1201-1212).
Sugie, Y., Zhang, J., & Fujiwara, A. (2003). A weekend shopping activity participation
model dependent on weekday shopping behavior. Journal of Retailing and Consumer
Services, 10(6), 335-343.
T. Haveliwala, G. Jeh, and S. Kamvar, “Targeted advertisements based on user profiles
and page profile,” Nov. 27 2012, US Patent 8,321,278
Telang, R., Boatwright, P., & Mukhopadhyay, T. (2004). A mixture model for Internet
search-engine visits. Journal of Marketing Research, 41(2), 206-214.
Turner, P., & Turner, S. (2011). Is stereotyping inevitable when designing with
personas?. Design studies, 32(1), 30-44.
Vuong, K. T. (2015). An exploration of the customer behavioral model for e-marketing
strategy in Vietnam. Review of Management Innovation and Creativity: Published and
Sponsored by: Intellectbase International Consortium, 8(23), 98-104.
Weber, P., & Schweiger, W. (2017). Content Effects: Advertising and Marketing. The
International Encyclopedia of Media Effects, 1-13.
Wolny, J. and Charoensuksai, N. (2014) Mapping customer journeys in multichannel
decision-making. Journal of Direct, Data and Digital Marketing Practice 15 (4) pp. 317-
Wolny, J., & Mueller, C. (2013). Analysis of fashion consumers’ motives to engage in
electronic word-of-mouth communication through social media platforms. Journal of
marketing management, 29(5-6), 562-583.
Xue, L., Ray, G., & Gu, B. (2011). Environmental uncertainty and IT infrastructure
governance: A curvilinear relationship. Information Systems Research, 22(2), 389-399.
Yang, X., Sun, D., Zhu, R., Deng, T., Guo, Z., Ding, Z., ... & Zhu, Y. (2019, July).
AiAds: automated and intelligent advertising system for sponsored search. In
Proceedings of the 25th ACM SIGKDD International Conference on Knowledge
Discovery & Data Mining (pp. 1881-1890).
Yang, Y., Qin, R., Jansen, B. J., Zhang, J., & Zeng, D. (2014). Budget planning for
coupled campaigns in sponsored search auctions. International Journal of Electronic
Commerce, 18(3), 39-66.
Yang, Y., Zeng, D., Yang, Y., & Zhang, J. (2015). Optimal budget allocation across
search advertising markets. informs Journal on Computing, 27(2), 285-300.
Yang, Y.; Jansen, B.J.; Yang, Y., Guo, X.; and Zeng, D. Keyword optimization in
sponsored search advertising: A multilevel computational framework. IEEE Intelligent
Systems, 34, 1, (2019), 3242.
Yang, Y.; Yang, Y.C.; Jansen, B.J.; and Lalmas, M. Computational advertising: A
paradigm shift for advertising and marketing? IEEE Intelligent Systems, 32, 3 (2017),
Zhang, W., Zhang, Y., Gao, B., Yu, Y., Yuan, X., & Liu, T. Y. (2012, August). Joint
optimization of bid and budget allocation in sponsored search. In Proceedings of the
18th ACM SIGKDD international conference on Knowledge discovery and data mining
(pp. 1177-1185).
Zhao, J., Qiu, G., Guan, Z., Zhao, W., & He, X. (2018, July). Deep reinforcement
learning for sponsored search real-time bidding. In Proceedings of the 24th ACM
SIGKDD international conference on knowledge discovery & data mining (pp. 1021-
Zhu, H., Jin, J., Tan, C., Pan, F., Zeng, Y., Li, H., & Gai, K. (2017, August). Optimized
cost per click in taobao display advertising. In Proceedings of the 23rd ACM SIGKDD
International Conference on Knowledge Discovery and Data Mining (pp. 2191-2200).
Zia, M., & Rao, R. C. (2019). Search advertising: Budget allocation across search
engines. Marketing Science, 38(6), 1023-1037.
Zigmond, D., & Stipp, H. (2010). Assessing a new advertising effect: Measurement of
the impact of television commercials on Internet search queries. Journal of Advertising
Research, 50(2), 162-168.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
This research examines the impact of online display advertising and paid search advertising relative to offline advertising on firm performance and firm value. Using proprietary data on annualized advertising expenditures for 1651 firms spanning seven years, we document that both display advertising and paid search advertising exhibit positive effects on firm performance (measured by sales) and firm value (measured by Tobin's q). Paid search advertising has a more positive effect on sales than offline advertising, consistent with paid search being closest to the actual purchase decision and having enhanced targeting abilities. Display advertising exhibits a relatively more positive effect on Tobin's q than offline advertising, consistent with its long-term effects. The findings suggest heterogeneous economic benefits across different types of advertising, with direct implications for managers in analyzing advertising effectiveness and external stakeholders in assessing firm performance.
Conference Paper
Full-text available
This research aims to explore bidding strategies on two different levels (i.e., adgroup and keyword) in search engine advertising (SEA). With consideration of uncertainty in advertising performance, we build a stochastic bidding model that can be applied to adgroup and keyword levels. Then we develop an integrated strategy to seek out a feasible solution based on the tradeoff between the expected profit and advertiser's computational cost (or operational time). Using a panel dataset collected from field reports and logs of search advertising campaigns, we conduct computational experiments to evaluate the performance of our models. Experimental results show that 1) bidding on the keyword level leads to higher profit with higher variability, compared to that on the adgroup level; 2) the integrated strategy of optimal bidding can help advertisers obtain the highest profit under different constraints of computational costs; 3) for adgroups and keywords with better performance indexes, bidding prices are higher, and increase faster with the budget; 4) as the computational cost increases, the marginal profit initially increases sharply and then decreases after a certain point.
Consumers’ online searches usually involve multiple keywords about their purchases, which vary depending on the purchase stage. Similarly, retail advertisers use a set of related keywords for competing brands. Thus, understanding how consumers search using keywords for competing brands at different purchase stages is important for retailers seeking to use multiple keywords more effectively. We examine consumers’ click behavior and retailers’ bids across multiple keywords. We empirically show that, while consumers search in a manner generally consistent with the purchase funnel, their behavior differs between market leader and follower brands. We also find that retailers consider the different keywords to be strategic complements, but this does not hold when consumers are close to making a purchase decision. Interestingly, retailers’ bid allocation across keywords may be inconsistent with consumers’ click behavior, revealing a potential opportunity to improve the performance of search advertising campaigns.
Conference Paper
Baidu runs the largest commercial web search engine in China, serving hundreds of millions of online users every day in response to a great variety of queries. In order to build a high-efficiency sponsored search engine, we used to adopt a three-layer funnel-shaped structure to screen and sort hundreds of ads from billions of ad candidates subject to the requirement of low response latency and the restraints of computing resources. Given a user query, the top matching layer is responsible for providing semantically relevant ad candidates to the next layer, while the ranking layer at the bottom concerns more about business indicators (e.g., CPM, ROI, etc.) of those ads. The clear separation between the matching and ranking objectives results in a lower commercial return. The Mobius project has been established to address this serious issue. It is our first attempt to train the matching layer to consider CPM as an additional optimization objective besides the query-ad relevance, via directly predicting CTR (click-through rate) from billions of query-ad pairs. Specifically, this paper will elaborate on how we adopt active learning to overcome the insufficiency of click history at the matching layer when training our neural click networks offline, and how we use the SOTA ANN search technique for retrieving ads more efficiently (Here "ANN'' stands for approximate nearest neighbor search). We contribute the solutions to Mobius-V1 as the first version of our next generation query-ad matching system.