Analysis and illustration of the practical impact of Artificial Intelligence and Intelligent Personal Assistants on business processes in small- and medium-sized enterprises

Abstract

Artificial intelligence (AI) and intelligent personal assistants (IPAs) are becoming more and more important. This is no longer limited to private use but also becoming increasingly important in everyday business life. The identification of optimization potentials through the use of AI and IPAs is therefore relevant from both a theoretical and a practical point of view. This paper, therefore, identifies concrete use cases for relevant processes in small-and medium sized enterprises (SME) in the service industry and enhances them with AI and IPA capabilities. Based on a prototype, the use cases were presented to 10 experts who were interviewed regarding their usefulness and influencing factors. Subsequently, the results of the interviews were categorized and validated again by a quantitative survey within the expert panel. As a result, the use cases were evaluated with regard to the specific influencing factors and the potential for optimization was determined. The use cases were evaluated based on this data. It was shown that IPA features in particular are perceived as useful. On average, AI and IPA features have a cost savings potential of over 31%. This shows the importance of these features and the need to consider them when modeling modern business processes.

Full text

Abstract— Artificial intelligence (AI) and intelligent personal
assistants (IPAs) are becoming more and more important. This
is no longer limited to private use but also becoming
increasingly important in everyday business life. The
identification of optimization potentials through the use of AI
and IPAs is therefore relevant from both a theoretical and a
practical point of view. This paper, therefore, identifies
concrete use cases for relevant processes in small- and medium
sized enterprises (SME) in the service industry and enhances
them with AI and IPA capabilities. Based on a prototype, the
use cases were presented to 10 experts who were interviewed
regarding their usefulness and influencing factors.
Subsequently, the results of the interviews were categorized
and validated again by a quantitative survey within the expert
panel. As a result, the use cases were evaluated with regard to
the specific influencing factors and the potential for
optimization was determined. The use cases were evaluated
based on this data. It was shown that IPA features in particular
are perceived as useful. On average, AI and IPA features have
a cost savings potential of over 31%. This shows the importance
of these features and the need to consider them when modeling
modern business processes.
I. INTRODUCTION
After two waves of Artificial Intelligence (AI) in the 1960s
and 1980s, the third wave is ongoing since the early 2010s
[1]. This wave will - according to recent studies - not only
change our private everyday life but will also entail
comprehensive changes in business processes, as this
technology has made advances in logic and abstraction and
has reached a new level of autonomous decision-making [1],
[2]. For example, companies are increasingly using chatbots
to meet the need for constant availability for technical
support or information [3]. AI can already be used in a
variety of business processes, ranging from marketing while
accurately addressing brand messages [4] to automating
accounting processes [5] or supporting complex decisions
Dr. Daniel Hüsson is head of consulting at a German ERP system
provider and research fellow at the Institute for IT Management &
Digitization. His research focuses on the impact of artificial intelligence on
SME business processes. E-Mail: daniel.huesson@fom-net.de
Prof. Dr. Alexander Holland is a Full-time professor at FOM University
of Applied Sciences. His main research is knowledge management. E-Mail:
alexander.holland@fom.de
Prof. Dr. Fathi has established the Institute of KBS &KM at the
University of Siegen in Germany his research focus is Applied Knowledge
Management. E-Mail: fathi@informatik.uni-siegen.de
Prof. Dr. Rocío Arteaga Sánchez is an Assistant Professor at UCAM
(Catholic University of Murcia), Spain. Her research focus is the acceptance
and usage of new information technologies. E-Mail: rarteaga@ucam.edu.
[6] especially the application possibilities of machine
learning (ML) to optimize processes. They are highly
regarded by many companies from different industries.
Pattern recognition, image analysis, text analysis or voice
control offer extensive potential in business processes [7].
Surprisingly, AI is currently used relatively seldom in
German companies and those companies with less than 500
employees and a turnover of less than 1 billion Euros rarely
use AI. However, AI is generally more common in service
companies than in the trade and industry sector [8], [9]. It is,
therefore, necessary to analyse the reasons why AI is not yet
more widely used. Considering the service sector as the
current pioneer in AI serving it is, therefore, the objective of
the research to answer why there exists a gap in the use of
AI. According to a PwC study from 2019 [8], this passivity
of many companies lies in the lack of knowledge of
decision-makers about the potential of AI solutions. Other
studies [10] also consider the diffusion of AI in companies
to be similarly low but forecast a strong increase in the next
5 to 10 years. This is evident since expert surveys also
estimate the potential of AI use in small- and medium sized
enterprises (SME) to be large in many areas such as process
efficiency, customer service and work quality [10]. A
summary study shows [11] that AI is a long-term trend that
is both a growth-securing factor and a driver of innovation
for SMEs. As mentioned above, different areas of
application of AI are shown including in depth use of cases
that is relevant for SMEs in everyday business. It is therefore
consistent to extend the research into the area of SMEs in the
service industry and to build a link between theory and
practice. This should be based on concrete use of cases in
order to analyse the added value of the practice of AI in
specific business processes. Therefore, it is important that
the transfer of AI technologies to create value in SMEs is
successful in order to remain competitive [12]. The aim of
this study is to answer the following research questions that
resulted from the above-mentioned research gaps of the
missing use-cases and the low uptake of AI supported
processes in SMEs [13]. RQ: “Which specific AI- and IPA-
features are being recognised as useful in a business context
of SMEs in the service sector?”
To answer the questions in this study, relevant process areas
in service-oriented SMEs are identified and concrete AI use
cases are derived from the literature and combined with
requirements surveys from a previous study [14]. As part of
a study already completed, an artifact in the form of an
Analysis and illustration of the practical impact of Artificial
Intelligence and Intelligent Personal Assistants on business processes
in small- and medium-sized service enterprises
Daniel Hüsson, FOM, Institute for IT-Management and Digitalization, Essen, Germany
Alexander Holland, FOM, Institute for IT-Management and Digitalization, Essen, Germany
Madjid Fathi, University Siegen, Faculty IV Knowledge Based Systems and Knowledge Management, Germany
Rocío Arteaga Sánchez, UCAM - Catholic University of Murcia, Spain
2021 IEEE International Conference on Systems, Man, and Cybernetics (SMC)
17-20 October, 2021. Melbourne, Australia
978-1-6654-4207-7/21/$31.00 ©2021 IEEE 3287

Intelligent Personal Assistant (IPA) was realised. This
prototype named V-IP-A is part of the next generation of the
ERP system Vemas and supports speech commands to
provide complex processes such as information retrieval and
input in a sales context, as well as calling up overviews and
explanation graphical reports. Within the framework of the
study, an evaluation was carried out based on a quantitative
survey of ERP users. The evaluation has shown which
enhancements are necessary to create added value and
increase user acceptance according to the feedback
extensions such as a speech command for customer search
and also prediction functions were implemented. [14] Fig I.
shows the prototype V-IP-A as a plug-in for an ERP system.
Figure I: Prototype V-IP-A in ERP-System
Based on the final version of the prototype, 10 experts from
various industrial fields such as IT & Management
consulting, IT Service providers, technical documentations
and special machine construction were interviewed within
the framework of a qualitative study and the prototype was
considered in the context of the use cases regarding the
application potential and the influencing factors.
Subsequently, a content analysis was carried out and the use
potential, as well as the corresponding influencing factors,
were extracted. To verify the results of the qualitative
analysis, the findings were processed in the form of a
questionnaire and the experts were interviewed again as part
of a quantitative study. The corresponding results of both
studies are presented within this paper.
II. RELATED WORK AND CONTRIBUTIONS
Current scientific literature is already focusing on the use of
AI in business processes in the context of specific industries
and working out AI use-cases for improving processes in the
specific process for banks [15]. Extensive cross-industry
research is also carried out, for example, to evaluate the
general use of AI in the area of project management [16],
[17]. In addition, there is also broad-based, general research
into how AI affects business processes and how this
potential can theoretically be exploited [18]. Detailed
research is also carried out by SMEs in industry - for
example in the IoT environment - to transfer application
cases from science to practice [19]. Within the literature, it
has also already been investigated why AI has not yet been
used in SMEs - even in a high-tech country like Germany
[13]. In addition to other challenges, such as how to initially
master digitisation, a study also identified a lack of experts,
a lack of data and a lack of relevant use-cases [13]. Even
though there are many descriptions of use-cases in the
literature, usually only rough areas are described, but no
detailed processes within a company context are developed
and evaluated [2]. It can therefore already be proven from
the literature that identification of suitable use cases is an
important research aspect to show especially SMEs what
potential is available in AI. Through the development of
practical use cases and the identification of influencing
factors and the evaluation of the optimisation potential, this
work contributes to developing a profound understanding of
the impact of AI and IPA functions on business processes.
Table I provides an overview of studies using similar
research methods, it can be shown that the methodology
presented in this paper is in line with current research.
TABLE I. RELATED RESEARCH METHODS
Research methodology
Research objective
Qualitative findings from a
semi-structured interview
combined with the quantitative
findings from ∼10,000
messages that the participants
exchanged with the chatbots
Inform and guide the design of future
chatbots [20]
Semi-structured interviews of
employees from different
backgrounds
A better understanding of the needs
and demands of potential users of
smart assistants for private- and work-
related availability [21]
Interviews with CEOs of
SMEs, grounded in the
technology-organization-
environment theory and the
Knowledge-Based View
Organizational impact of Internet
technologies by analysing factors
affecting e-business use and its effect
on organisational innovation in
manufacturing Small and Medium-
Size Enterprises [22]
III. INTRODUCTION OF THE USE CASES
Within this chapter, the use cases and their theoretical basis
are presented. This is necessary to be able to better classify
the results of the qualitative and quantitative studies already
mentioned in the introduction. The use cases are categorized
according to business divisions identified as especially
relevant in SME in the literature: sales & marketing [23],
project management [24], customer service [25] and
controlling [26], [27]. The areas of supply chain
management [28] and accounting [29] were also identified,
but the expert interviews showed that the aforementioned
areas have greater relevance. The use cases relevant for
further consideration are therefore described below.
A. Sales & Marketing
Use-Case 1.1: Lead management (LM)
Lead management is the process of managing and
tracking prospective customers, especially to identify their
revenue potential. AI especially ML is also able to verify the
impact of different factors to approve the lead scoring
parameters and optimise targeting and segmenting [4], [30],
[31].
Use-Case 1.2: Forecasting (FC)
Sales forecasting is being used by companies as a basis
for sales revenue estimation, decision making, operation and
marketing strategies development. ML can support sales
forecasting by analysing large and high dimensional data to
predicting trends and improve the validity of the forecast.
Enriched with historical sales figures, product features and
3288

characteristics as well as publicly available data, it is possible
to predict short and long-term company performance [32],
[33].
Use-Case 1.3: Information input and retrieval (IIR)
Collecting information about current and potential
customers is an important task in the sales context. When a
sales representative has had a meeting with a customer, it is
important that the key information is captured in a software
system so that other employees can also gain insight into
relevant topics or ongoing negotiations. The input of this
information can be simplified by speech recognition so that
this task can be done hands-free. Information can also be
requested by speech command, allowing natural interaction
with the software system [14], [34]–[36].
B. Project management (PM)
Use-Case 2.1: Project planning
Project planning is a complex task, the relevant resources
must be available at the right time, in the necessary quantity
and at the right location. By using AI, the planning process
can be supported by determining suitable resources based on
previous projects and identifying project risks by analysing
milestone shifts or cost and effort overruns in similar projects
[37], [38].
Use-Case 2.2: Project implementation
Long-term decision-making processes and poor
communication between project parties and contractors are
the risks that delay the project. With the assistance of AI,
real-time analysis of the project can provide a complete
picture of the decisions to be made based on data from
similar projects. All available information can be compiled
and made available to the decision-maker to speed up the
process of the upcoming decision. Based on communication
entries from previous projects, suggestions for the necessary
project communication can be derived to keep the contractor
and the project team informed. The extraction of best
practices after project completion are also conceivable use
cases to make the knowledge available to other project
sponsors [38].
C. Customer Service (CS)
Use-Case 3.1: Ticket classification
When customers have a problem or question about a
product or service, customer service is usually contacted for
assistance. The contact is established through various
channels, such as telephone, e-mail, portals, or APIs.
Depending on the input channel, tickets must be classified
based on the available data to control further processing. In
this process, machine learning can be used to learn the
classification based on already classified tickets and to
independently classify new tickets based on the trained model
[39].
Use-Case 3.2: Ticket assignment
Assigning tickets to an appropriate agent or group to
resolve the case is an important and time-consuming process
that requires knowledge of the IT portfolio under
management, the responsibilities and roles of each group, and
the ability to quickly analyse the ticket content that describes
the problem and assign it to the appropriate agents. In
particular, the forwarding of time-critical, complex tickets to
an expert who has the specific knowledge to solve the
problem holds optimization potential through AI. Based on
past assignments, AI can identify which tickets have been
successfully completed by which agent and use this data to
carry out an automated assignment considering availability
and workload [40].
Use-Case 3.3: Solution suggestion for tickets
Finding solutions for tickets can be time-consuming;
many man-hours are spent searching through old tickets to
find a solution for a current problem. Depending on the
availability of reliable data within the problem to be solved,
AI can identify similar tickets or FAQs to make suggestions
for solving the problem. As a result of the suggestion
identification process, the service desk user retrieves a list of
similar tickets and the corresponding solutions to select the
most appropriate one. Based on the user feedback, the
algorithm can learn and increase the relevance of the selected
solution in the context of the search parameters. To justify the
reasons why the specific solutions were suggested by the
framework, a combination of case-based reasoning and in-
depth learning can explain the result and enable the user to
understand the decision-making process [41], [42].
Use-Case 3.4: Chatbot
Companies can use chatbots to set up 24/7 customer
support to resolve product or technical issues. The chatbots
are based on a dialog system that has been trained using a
database of questions and corresponding answers. Users can
report problems and receive solutions at any time, and the
workload of the support staff is reduced when problems are
handled partially or fully autonomously by the chatbot [43].
Use-Case 3.5: Support volume prediction
For the most cost-efficient operation of customer service,
not only is an effective process organization essential, but
also the avoidance of over- and under capacities. The
expected support volume plays a role here, for which ML
procedures can also be used, analogous to procedures that
determine e.g. delays in ticket processing and the forecast of
solution times [44].
D. Controlling (CT)
Use-Case 4.1: Interpretation of repots
The correct interpretation of reports is an essential aspect of
decision making. For example, the contribution margin is an
important indicator of the company’s health and it is the
foundation for break-even analysis. Using configurations for
KPIs, it is not only possible to prepare the underlying data
visually, but it is also possible to interpret this data in
context and explain the displayed information to the user via
speech synthesis in combination with visual highlighting
[14], [36], [45].
Use-Case 4.2: Data-informed decision making
To make sound decisions, it is necessary to use an
extensive database as a basis for the decision-making
process. Nevertheless, the gut feeling, if developed over
years of experience, is also an important factor in the
decision-making process. The combination of a data-driven
decision-making process in combination with intuition is
called data-informed decision making. ML can also be used
3289

here to enable extensive drilldowns in data. This process can
be simplified by a natural interaction with the system via
voice command [46].
IV. INFLUENCING FACTORS AND OPTIMIZATION POTENTIALS
The use cases listed in the previous chapter serve as the basis
for further analysis of the influencing factors and
optimization potential and thus form the foundation for
answering the research question posed within the
introduction. In the period of two months between August
and September 2020, 10 experts were interviewed using a
guided interview. The experts have more than 10 years of
experience in the field of business management and
profound knowledge in business process modeling. 9 of the
10 experts are managing directors of a company in the
service segment, one expert is a freelance management
consultant. Most of the experts also work in a consulting
capacity for other customers, and three experts are
professors of business information systems and have
extensive insights into other companies through lectures and
the supervision of final theses by part-time students. The
individual use cases were presented to the experts and then
asked for an assessment with regard to the potential for use
and the influencing factors. The interviews were conducted
online, a total of 18.5 hours were recorded, logged and then
inductive categories were formed using qualitative content
analysis according to Mayring [47], [48].
A. Influencing factors
Within this section, the factors influencing the usefulness of
functions are analyzed. As a result, from the qualitative
content analysis, 21 different influencing factors were
identified, and the top 5 – as explained later - are presented
in Table II.
TABLE II. INFLUENCING FACTORS
Influencing factors
Definition
Dynamic interaction
Dynamic interaction influences the application
potential. Here the focus is on the flexible use of
functions, which can also recognise an existing
context and include it in the use.
Expertise
The knowledge or expertise of the user influences
the application potential
Field of application
The application environment influences the
application potential. This includes, for example,
mobile working or the use of applications while
driving.
Generalisability
The generalisability influences the application
potential. This can depend on individual
characteristics in products and services, but also on
individual approaches that cannot be summarised. A
heterogeneous database also falls into this category.
Volume
The number of applications influences the
application potential
Fig. II shows the assignment of the influencing factors to the
individual use cases and the experts' assessment of the
optimization potential regarding quality and speed, based on
a four-point Likert scale: None: No potential, Low: Low
potential, Medium: Medium potential, High: High potential.
Figure II: Influencing factors and optimization potential
The initial analysis shows that there is at least a low
optimization potential in all the use cases presented. To
verify the influencing factors and obtain concrete estimates
for the actual optimization potential, the experts were
surveyed in the period from November 1, 2020 to November
30, 2020 using a quantitative questionnaire. This procedure
is based on the Delphi method and serves to disclose the
interim results of the interviews in the expert group and to
obtain a quantification of the significance of the influencing
factors and the concrete optimization potential in the areas of
process speed, process quality and cost savings [49], [50].
Of the original 10 experts interviewed, 7 subsequently
participated in the quantitative survey. Due to the scope of
the question, the use cases were condensed for the
questionnaire. If functions are similar or can be combined
into an overall workflow, they were grouped accordingly.
Fig. III shows the summarization carried out.
Figure III: Summary level use-cases
3290

In the quantitative survey, the experts had the opportunity to
add or remove influencing factors extracted from the content
analysis and to rank them according to priority. The result of
the survey is presented in Table III as top 5 List. The column
count indicates the number of areas in which the influencing
factors were mentioned. The column score is the sum of the
position of the factor in the prioritisation of the experts. The
lower the score, the more important the influencing factor is.
To make the results comparable, the weighted score was
calculated by dividing the sum of the items by the number of
items.
TABLE III. TOP 5 WEIGHTED INFLUENCING FACTORS
Category
Areas
Count
Score
Weighted
score
Field of application
IIR
1
11
11.00
Volume
LM, FC,
PM, CS
4
54
13.50
Generalisability
LM, PM,
CS, CT
4
73
18.25
Dynamic interaction
IIR
1
19
19.00
Expertise
CT
1
20
20.00
The evaluation total showed that there is no influencing
factor that is relevant in all six use-case areas. However,
with volume, generalizability and complexity, some factors
are relevant in four areas. There is also overlap in the higher-
level areas, as the three influencing factors play a role in
lead management as part of sales & marketing, project
management, and customer service. Looking at the weighted
score shows that volume and generalizability play an
important role. These two factors show the second and third
best values. The influencing factor complexity, on the other
hand, occurs frequently but is only seen in 12th place overall
out of the 18 factors identified. The complexity of processes,
in general, does not have a major influence on the usefulness
of AI support. The highest weighted value has the
influencing factor application area, but this is only
considered relevant in the area of information input &
retrieval. The volume over a broad number of processes is
the most important influencing factor, as well as the fact that
generalisability plays a role in a valid determination of data
and the resulting models. Data enrichment, as well as the
predictability of events and the recognition of patterns, are
also in the upper half of the prioritisation. Since the basic
prerequisite for valid models for AI processes is of course
reliable data and actual correlations, the evaluation of the
prioritisation clearly shows that the experts are aware of this
issue and that the final result of this evaluation is also
covered by process models in the scientific literature [51],
[52].
B. Optimization potential
After considering the influencing factors, the analysis of
the optimization potential follows now. The experts were
asked to give an assessment of the impact of the respective
use cases in the areas of overall integration, quality
enhancement, process speed and cost savings. The
assessment was made from two perspectives, firstly for their
own company and secondly for other companies. The values
were queried as percentages so that they can be compared
and agreed upon across all areas.
Based on the collected data and to check whether there is
a dependency between the ratings of the process attributes
and the cost savings potential estimated by the experts, a
correlation matrix was formed. The correlation matrix
showed that quality and overall integration have only a
minor influence on the expected cost savings potential. Time
expenditure and total costs have a moderate influence on the
cost savings potential. There is a strong correlation between
the expected optimization potential in one's own company
and an even stronger correlation with the optimization
potential in other companies. One approach to identifying
relevant use cases affected by AI and IPA can be to identify
processes that are time-consuming and cost intensive. Based
on the experts' assessments of the optimization potential, Fig
IV shows the distribution in the individual areas.
Figure IV: Summary level use-cases
Further analysis of the graph is provided in the following
four sections.
1) Quality
It turns out that forecasting has the greatest influence on
quality. It can be assumed that data maintenance and also the
evaluation of potentials represent a challenge for the
companies surveyed; according to expert estimates, quality
can be increased by more than 42% through appropriate data
enrichment in combination with the evaluation of potentials
based on trained models. Through the features presented in
the use case controlling, the experts see a quality increase
potential of 35 %. Decisions can then be made based on an
enriched database. In addition, employees without expert
knowledge can gain important insights from overviews
through the explanation function, so that decisions can be
made on a broader basis. The quality of decisions can also
be improved by appropriate drilldowns, as data can be
scrutinized and important details retrieved. The experts also
see the potential for quality improvement of around 30% in
the other areas of lead management, information input &
retrieval, and project management and customer service. The
average value for all the properties examined is 33.81%.
2) Process speed
The next criterion to be considered is the potential for
3291

increasing process speed. With an improvement potential of
40%, the feature set Information Input & Retrieval is in the
first place. Voice interaction allows data to be entered into
the system more quickly and queries for data to be initialized
by voice. In second place are the customer service feature
sets. Here, the experts see an increase of more than 38%
through automation of ticket processing. On average across
all the use cases examined, an increase potential for process
throughput time of 35.5% was determined.
3) Cost savings potential
In the area of cost savings potential, the Information Input
& Retrieval feature set is also in first place with a potential
of over 37%. Customer Service also ranks second in this
category with a potential savings of 35%. On average, the
cost savings potential is over 31.11%.
4) Overall view
Looking at all functional areas together, it also appears that
Information Input & Retrieval is recognized as the most
useful feature set; drawing an average value across all
optimization potentials, this feature set receives 36.43%. In
second place in this approach is the Forecasting feature set
with 35%, closely followed by Customer Service with
34.44%. Lead management receives an overall score of
33.1%, controlling 31.67% and finally comes the functional
area of project management with 30%.
C. Combining influencing factors and optimization
potential
To obtain further indications for the identification of use
cases with corresponding optimization potential, a
comparison of the identified influencing factors and the
determined cost-saving potentials is carried out. Table IV
compares the influencing factors of the different areas with
the calculated mean value of the cost savings potential and
sorts them according to the highest cost savings potential.
TABLE IV. TOP 5 INFLUENCING FACTORS AND COST SAVING
POTENTIAL
Category
Areas
Cost saving
potential (%) –
Average – Mean
Field of application
& Dynamic
interaction
Information input &
retrieval
37.86
Volume
Lead-Management,
forecasting, project
management, customer
Service
30.54
Generalisability
Lead-Management, project
management, customer
Service,
controlling
30.24
Expertise
Controlling
26.67
But due to the high optimization potential of the IPA within
the processes for information input and query, the
corresponding influencing factors such as application area
and input form also play a correspondingly important role in
the overall consideration. Due to the mean value approach,
overarching factors such as volume or generalizability only
have a medium influence on the savings potential. The
approach of determining a suitable use case via the
influencing factors is therefore not necessarily practicable,
but the overview can give an indication of which factors
should be given particular consideration after the specific
selection of a use case.
V. CONCLUSION
Within the paper, it could be shown which influencing
factors are relevant in which use cases and which
optimization potential functions the use cases have. For
companies facing a decision on which business processes
should be optimized through the use of AI, the use cases
provided, as well as the combination of influencing factors
and the savings potential, provide an initial approach for
identifying suitable processes. Of course, a case-by-case
analysis must always be carried out under the respective
aspects of the company to make a target-oriented selection.
Through this approach, the research question can now be
answered conclusively. Overall, it was shown that the IPA
features were rated as the most useful. This finding is
aligning with other research projects. Studies show that
speech, with approximately 150 words per minute or 983
characters per minute, is significantly superior to traditional
keyboard input, with 31 words per minute or 203 characters
per minute [53]. In addition, the use of an IPA within an
ERP system can be used to respond to the weak points in
process flows in terms of usability, especially in navigation,
information overload and incorrect information delivery and
lack of system communicativeness [54]–[56]. By identifying
volume as the most important influencing factor, it was also
possible to explain why, according to studies [10]–[12], [57],
SMEs have not yet implemented AI in many cases. Without
a corresponding frequency of business processes,
investments are only profitable if the processes are complex
or time intensive. This is particularly evident in the area of
customer care. However, since an average cost saving of
over 31% can be achieved across all use cases, it also shows
that all process optimizations are generally worthwhile.
Therefore, it could be shown that the review of the most
important business processes is necessary and that the use of
AI and IPA features should be evaluated.
REFERENCES
[1] L. Deng, “Artificial Intelligence in the Rising Wave
of Deep Learning: The Historical Path and Future
Outlook [Perspectives],” IEEE Signal Process.
Mag., vol. 35, no. 1, pp. 180–177, Jan. 2018, doi:
10.1109/MSP.2017.2762725.
[2] D. Hecker, I. Döbel, U. Petersen, A. Rauschert, V.
Schmitz, and A. Voss, Zukunftsmarkt Künstliche
Intelligenz – Potenziale und Anwendungen.
Fraunhofer-Allianz Big Data, 2017.
[3] P. B. Brandtzaeg and A. Følstad, “Chatbots:
changing user needs and motivations,” Interactions,
vol. 25, no. 5, pp. 38–43, Aug. 2018, doi:
10.1145/3236669.
[4] C. Campbell, S. Sands, C. Ferraro, H.-Y. (Jody)
Tsao, and A. Mavrommatis, “From data to action:
3292

How marketers can leverage AI,” Bus. Horiz., vol.
63, no. 2, pp. 227–243, Mar. 2020, doi:
10.1016/j.bushor.2019.12.002.
[5] O. Henrich, “Chatbots auf dem Vormarsch: Der
künstlich-intelligente Buchhalter kommt,”
Wirtschaftsinformatik Manag., vol. 9, no. 6, pp. 72–
75, Dec. 2017, doi: 10.1007/s35764-017-0130-2.
[6] D. Kiefer and A. Ulmer, “Application of Artificial
Intelligence to optimize forecasting capability in
procurement CR-Category,” Wissenschaftliche
Vertiefungskonferenz, no. November, 2019, doi:
10.5281/zenodo.3539397.
[7] B. Böttcher, A.-L. Schwalm, and C. Velten,
“Machine Learning in deutschen Unternehmen -
Eine empirische Studie zu Betrieb und Anwendung
von Künstlicher Intelligenz,” 2018.
[8] PwC, “Künstliche Intelligenz in Unternehmen Mehr
künstliche Intelligenz wagen !,” 2019.
[9] M. Ulrich and D. Bachlechner, “Wirtschaftliche
Bewertung von KI in der Praxis – Status Quo,
methodische Ansätze und Handlungsempfehlungen,”
HMD Prax. der Wirtschaftsinformatik, vol. 57, no.
1, pp. 46–59, Feb. 2020, doi: 10.1365/s40702-019-
00576-9.
[10] M. Lundborg and C. Märkel, “Künstliche Intelligenz
im Mittelstand,” 2019.
[11] Mittelstand - Digital, “Künstliche intelligenz im
Mittelstand - Potenziale und Anwendungsbeispiele,”
2020.
[12] T. Wangermann, “KI in KMU,” Konrad Adenauer
Stift. Anal. Argumente, vol. 381, 2020.
[13] C. Szedlak, P. Poetters, and B. Leyendecker,
“Application of Artificial Intelligence In Small and
Medium-Sized Enterprises,” 2020.
[14] D. Hüsson, A. Holland, and R. A. Sánchez,
“Intelligent Personal Assistant in Business- Context:
Key-feature Evaluation for User Acceptance,” Bus.
Syst. Res., vol. 11, no. 3, pp. 147–166, 2020, doi:
https://doi.org/10.2478/bsrj-2020-0032.
[15] S. Mohapatra, “Human and computer interaction in
information system design for managing business,”
Inf. Syst. E-bus. Manag., no. 0123456789, Aug.
2020, doi: 10.1007/s10257-020-00475-3.
[16] B. Prieto, “Impacts of Artificial Intelligence on
Management of Large Complex Projects,” PM
World J., vol. VIII, no. V, 2019.
[17] P. Hofmann, J. Jöhnk, D. Protschky, and N. Urbach,
“Developing Purposeful AI Use Cases – A
Structured Method and Its Application in Project
Management,” in WI2020 Zentrale Tracks, vol. 0,
GITO Verlag, 2020, pp. 33–49.
[18] M. Tarafdar, C. M. Beath, and J. W. Ross, “Using
AI to Enhance Business Operations,” in MIT Sloan
Management Review, vol. 60, no. 4, 2019.
[19] E. B. Hansen and S. Bøgh, “Artificial intelligence
and internet of things in small and medium-sized
enterprises: A survey,” J. Manuf. Syst., no. October
2019, Aug. 2020, doi: 10.1016/j.jmsy.2020.08.009.
[20] M. Jain, P. Kumar, R. Kota, and S. N. Patel,
“Evaluating and informing the design of chatbots,”
DIS 2018 - Proc. 2018 Des. Interact. Syst. Conf., pp.
895–906, 2018, doi: 10.1145/3196709.3196735.
[21] Z. Saternus, “User preferences regarding smart
assistant for private- and work-related availability,”
in Proceedings of the 2019 ACM International Joint
Conference on Pervasive and Ubiquitous Computing
and Proceedings of the 2019 ACM International
Symposium on Wearable Computers -
UbiComp/ISWC ’19, 2019, pp. 1133–1138, doi:
10.1145/3341162.3350846.
[22] P. Soto-Acosta, S. Popa, and D. Palacios-Marqués,
“E-business, organizational innovation and firm
performance in manufacturing SMEs: an empirical
study in Spain,” Technol. Econ. Dev. Econ., vol. 22,
no. 6, pp. 885–904, Sep. 2015, doi:
10.3846/20294913.2015.1074126.
[23] R. Brüning, “Marketing und Vertrieb,” in
Digitalisierung und Industrie 4.0 -
Herausforderungen für den Mittelstand, W. Krüger,
A. Dreier, R. Merk, and B. Seel, Eds. Bielefeld:
Fachhochschule des Mittelstands, 2017, pp. 59–67.
[24] L. Teixeira, A. R. Xambre, J. Figueiredo, and H.
Alvelos, “Analysis and Design of a Project
Management Information System: Practical Case in
a Consulting Company,” Procedia Comput. Sci., vol.
100, pp. 171–178, 2016, doi:
10.1016/j.procs.2016.09.137.
[25] M. Daum, U. Holtgrewe, P. Schörpf, and M. Nocker,
“Call- und Service-Center: Konsolidierung der
Kundenservice-Dienstleistungen,” Work. Pap.
Forschungsförderung, vol. No. 077, 2018.
[26] B. L. Baker, C. J. Edwards, M. Mick, and S. E.
Conners, “Demand for Management Accounting
Information in Small Businesses: Judgment
Performance in Business Planning,” Int. J. Acad.
Bus. World, vol. 12, no. 1, pp. 93–102, 2018.
[27] P. Pisar and M. E. Mazo, “Controlling,
communication and corporate culture – the
opportunity for SMEs,” Econ. Sociol., vol. 13, no. 3,
pp. 113–132, Sep. 2020, doi: 10.14254/2071-
789X.2020/13-3/8.
[28] T. A. Chin, A. B. A. Hamid, A. Rasli, and R.
Baharun, “Adoption of Supply Chain Management
in SMEs,” Procedia - Soc. Behav. Sci., vol. 65, no.
ICIBSoS, pp. 614–619, Dec. 2012.
[29] Diamant Software, “Wer wollen wir sein? Der
kaufmännische Bereicherfindet sich neu.,” 2020.
[30] N. Syam and A. Sharma, “Waiting for a sales
renaissance in the fourth industrial revolution:
Machine learning and artificial intelligence in sales
research and practice,” Ind. Mark. Manag., vol. 69,
no. November 2017, pp. 135–146, Feb. 2018, doi:
10.1016/j.indmarman.2017.12.019.
[31] I. Lopatovska et al., “Talk to me: Exploring user
interactions with the Amazon Alexa,” J.
Librariansh. Inf. Sci., vol. 51, no. 4, pp. 984–997,
Dec. 2019, doi: 10.1177/0961000618759414.
[32] Z.-P. Fan, Y.-J. Che, and Z.-Y. Chen, “Product sales
3293

forecasting using online reviews and historical sales
data: A method combining the Bass model and
sentiment analysis,” J. Bus. Res., vol. 74, pp. 90–
100, May 2017, doi: 10.1016/j.jbusres.2017.01.010.
[33] S. Ren, C. Patrick Hui, and T. Jason Choi, “AI-
Based Fashion Sales Forecasting Methods in Big
Data Era,” 2018, pp. 9–26.
[34] T. Ammari, J. Kaye, J. Y. Tsai, and F. Bentley,
“Music, Search, and IoT: How people (really) use
voice assistants,” ACM Trans. Comput. Interact.,
vol. 26, no. 3, 2019, doi: 10.1145/3311956.
[35] D. Hüsson and A. Holland, “Intelligent Personal
Assistants in business processes: Evaluation of a
Prototype (V-IP-A),” in 32nd Bled eConference –
Humanizing Technology for a Sustainable Society,
2019, vol. 38, no. 1, pp. 1133–1145.
[36] D. Hüsson and A. Holland, “Intelligent Personal
Assistant and Reporting – Explaining Data to Users
through Speech Synthesis,” in Knowledge
Management in Digital Work Environments - State-
of-the-Art and Outlook Proceedings 10th
Conference Professional Knowledge Management,
P. Heisig, Ed. Potsdam: FHP Verlag, 2019, pp. 157–
161.
[37] C. Laroque, C.-A. Schumann, and C. Tittmann,
“Potenziale erschließen durch Künstliche Intelligenz
im Projektmanagement,” Proj. aktuell, vol. 3, pp.
39–41, 2019.
[38] H. K. Dam, T. Tran, J. Grundy, A. Ghose, and Y.
Kamei, “Towards Effective AI-Powered Agile
Project Management,” in 2019 IEEE/ACM 41st
International Conference on Software Engineering:
New Ideas and Emerging Results (ICSE-NIER), May
2019, pp. 41–44.
[39] D. Beneker and C. Gips, “Using clustering for
categorization of support tickets,” 2017.
[40] A. Mandal, N. Malhotra, S. Agarwal, A. Ray, and G.
Sridhara, “Cognitive System to Achieve Human-
Level Accuracy in Automated Assignment of
Helpdesk Email Tickets,” vol. 1, no. 2018, Springer
International Publishing, 2018, pp. 332–341.
[41] N. Madaan, G. Singh, A. Kumar, and G. B.
Dasgupta, “Neev: A cognitive support agent for
content improvement in hardware tickets,” in 2017
IFIP/IEEE Symposium on Integrated Network and
Service Management (IM), May 2017, pp. 239–246
[42] Q. Li and M. Vasarhelyi, “Developing a cognitive
assistant for the audit plan brainstorming session,”
Int. J. Digit. Account. Res., vol. 18, no. January, pp.
119–140, 2018, doi: 10.4192/1577-8517-v18_5.
[43] P. B. Brandtzaeg and A. Følstad, “Why People Use
Chatbots,” in Lecture Notes in Computer Science
(including subseries Lecture Notes in Artificial
Intelligence and Lecture Notes in Bioinformatics),
vol. 10673 LNCS, no. November, 2017, pp. 377–
392.
[44] D. Zuev, A. Kalistratov, and A. Zuev, “Machine
Learning in IT Service Management,” Procedia
Comput. Sci., vol. 145, pp. 675–679, 2018
[45] A. Maria Arcari, “Preventing crises and managing
turnaround processes in SMEs The role of economic
measurement tools,” Manag. Control, no. 3, pp.
131–155, Nov. 2018.
[46] M. Potančok, “Role of Data and Intuition in
Decision Making Processes,” J. Syst. Integr., vol. 10,
no. 3, pp. 31–34, 2019, doi: 10.20470/jsi.v10i3.377.
[47] P. Mayring and T. Fenzl, “Qualitative
Inhaltsanalyse,” in Handbuch Methoden der
empirischen Sozialforschung, Wiesbaden: Springer
Fachmedien Wiesbaden, 2014, pp. 543–556.
[48] P. Mayring, Qualitative Inhaltsanalyse: Grundlagen
und Techniken, 12th ed. Weinheim: Beltz, 2015.
[49] J. Steurer, “The Delphi method: an efficient
procedure to generate knowledge,” Skeletal Radiol.,
vol. 40, no. 8, pp. 959–961, Aug. 2011, doi:
10.1007/s00256-011-1145-z.
[50] S. McPherson, C. Reese, and M. C. Wendler,
“Methodology Update,” Nurs. Res., vol. 67, no. 5,
pp. 404–410, 2018.
[51] G. Martinetz, “Künstliche Intelligenz in
Unternehmensabläufe integrieren,”
Wirtschaftsinformatik Manag., vol. 12, no. 2, pp.
106–108, Apr. 2020.
[52] C. Rode-Schubert and P. Müller, “Welche
Fähigkeiten benötigt ein Unternehmen um
Künstliche Intelligenz nachhaltig erfolgreich
einzusetzen?,” in Transformationsvorhaben mit dem
Enterprise Transformation Cycle meistern, M. A.
Pfannstiel and P. F.-J. Steinhoff, Eds. Wiesbaden:
Springer Fachmedien Wiesbaden, 2020, pp. 143–
171.
[53] K. Sebold and G. Renner, “Usability of Input
Devices and Aids in the Field of Augmentative and
Alternative Communication - An Experimental
Study with People with and without Disabilities TT
- Usability von Eingabehilfsmitteln und
Ansteuerungsmöglichkeiten im Bereich der
Unterstüt,” Rehabilitation (Stuttg)., vol. 58, no. 5,
pp. 321–330, 2019.
[54] C. Lambeck, C. Fohrholz, C. Leyh, and R. Müller,
“Commonalities and contrasts: an Investigation of
ERP Usability in a Comparative User Study,” in
Proceedings of the 22nd European Conference on
Information Systems (ECIS 2014), 2014, pp. 1–15.
[55] C. Lambeck, R. Muller, C. Fohrholz, and C. Leyh,
“(Re-)Evaluating User Interface Aspects in ERP
Systems -- An Empirical User Study,” in 2014 47th
Hawaii International Conference on System
Sciences, Jan. 2014, pp. 396–405.
[56] W. Wong, V. Veneziano, and I. Mahmud, “Usability
of Enterprise Resource Planning software systems:
an evaluative analysis of the use of SAP in the
textile industry in Bangladesh,” Inf. Dev., vol. 32,
no. 4, pp. 1027–1041, Sep. 2016
[57] P. Kranzusch, A. Icks, B. Levering, and P. Pasing,
“Herausforderungen für den Mittelstand - Update
der Unternehmersicht 2019,” Inst. für
Mittelstandsforsch. Bonn, vol. No. 279, 2019.
3294