Content uploaded by Yang Yang
All content in this area was uploaded by Yang Yang on Apr 23, 2018
Content may be subject to copyright.
A better last-minute hotel deal via app?
Cross-channel price disparities between HotelTonight and OTAs
Yang Yang, Ph.D. *
1810 N.13th Street, Speakman Hall 304,
Philadelphia, PA 19122 USA
Xi Y. Leung, Ph.D.
University of North Texas
1155 Union Circle, #311100
Denton, TX 76203-5017 USA
* = corresponding author
Please cite as:
Yang, Y. and Leung, X. Y. (2018). A better last-minute hotel deal via app? Cross-channel price
disparities between HotelTonight and OTAs. Tourism Management. 68, 198-209.
A better last-minute hotel deal via app?
Cross-channel price disparities between HotelTonight and OTAs
Abstract: To better understand hotels’ cross-channel/platform pricing mechanisms, this study aims to
investigate various factors behind price discounts (as a type of price disparity) from a popular last-minute
hotel deal app, HotelTonight, compared to major online travel agency (OTA) websites. Using pricing data
collected from the HotelTonight app and OTA websites in six U.S. cities, we estimate several regression
models to examine price discounts. The results reveal that after controlling for other variables, price
discounts are largely shaped by online reputation metrics (i.e., relative review valence and volume on
TripAdvisor compared to on HotelTonight), complimentary access to services with high marginal
variable costs, and uncertainty in the room type offered. However, hotels’ market power does not explain
price disparities. Lastly, implications are discussed.
Keywords: last-minute deal; price disparity; online reputation; HotelTonight
Because of the perishable nature of hotel products and the high fixed costs of the hotel industry,
successful hotels always manipulate room prices in response to demand using sophisticated yield
management systems in order to maximize short-term revenue (Kimes, 1989). In addition to price
manipulation, hotels use a variety of distribution channels to sell their products to different market
segments (O’Connor, 2002). The advent of the Internet and the development of global distribution
systems has facilitated a paradigm shift in the hotel industry toward online distribution channels (Buhalis
& Law, 2008). In this new era, the proliferation of online distribution channels has also rendered hotel
rate structures transparent to travelers. Now, travelers are inclined to search online for the best deals or
lowest potential prices (Kimes, 2016; Schwartz, 2012). As a result, some degree of price disparity exists
across different distribution channels (Law, Chan, & Goh, 2007). It is therefore imperative for hotel
managers to carefully determine online pricing strategies given a better understanding of travelers’ online
deal-seeking behaviors (Marom & Seidmann, 2011).
One of the most important hotel pricing determinants is time, or advanced booking (Philips,
2005). Time-related pricing involves two contradictory assumptions of sellout risk and future discount
perception, leading to pricing practices like early-bird discounts and last-minute deals (Chen & Schwartz,
2013). Some studies on last-minute pricing have identified increases in hotel sales during the week before
a check-in date due to travelers’ heightened deal-seeking behaviors (Chen & Schwartz, 2013; Schwartz,
2012). Therefore, researchers suggest that last-minute hotel sales represent a substantial market segment
that should no longer be overlooked (Chen & Schwartz, 2013; Dacko, 2004). In general, this last-minute
booking market segment consists of tech-savvy consumers who are willing to compromise on factors like
hotel choice or room type to receive a better deal. Millennials constitute a large portion of this segment,
and mobile bookings dominate last-minute hotel bookings (Criteo, 2016). A study from Phocuswright
indicated that approximately 70% of mobile hotel bookings were for same-day or next-day check-ins
(Quinby, 2014). Another report by Criteo (2016) found that 60% of hotel bookings with same-day check-
ins were booked via smartphone, with another 7% completed via tablet.
Launched in January 2011, HotelTonight is a mobile hotel booking app that provides travelers
with last-minute discounted hotel rooms up to seven days in advance (HotelTonight, 2017; Leo, 2011).
As the first and leading app in hotel last-minute mobile booking, HotelTonight has spawned a number of
competitors, including the Asian-oriented HotelQuickly (launched in 2013), Roomlia (launched in 2014),
and One Night (launched in 2016) (Bishop, 2017; Huang, 2017). HotelTonight has received venture
funding totaling $117 million to date, leading to a $463-million valuation (O'Neill, 2017). As the
dominant last-minute hotel booking app in the Americas and Europe, HotelTonight serves more than
25,000 hotels in approximately 1,700 cities worldwide (Crook, 2017). HotelTonight differs from
traditional online travel agencies (OTAs) like Expedia or Priceline in two main respects: being mobile-
centric and dedicated to last-minute bookings (C.S.-W., 2013). First of all, HotelTonight is an app-only
hotel-booking tool that streamlines the booking process. Thus, consumers cannot book via the website as
they would through regular OTAs. Also, HotelTonight only sells last-minute rooms (up to seven days in
advance) with deep discounts off OTA-listed rates, whereas regular OTAs sell hotel rooms for any check-
in date (C.S.-W., 2013). What’s more, rather than listing every available hotel as regular OTAs do,
HotelTonight only allows a small number of listings in each city every day, so hotels compete against one
another to offer a good deal (Gaggioli, 2015). In sum, HotelTonight’s dedication to last-minute bookings
has generated new opportunities for hotel managers to fill rooms and maximize revenue at the last minute
(Makki, et al., 2016).
As a new and innovative online distribution channel, HotelTonight offers a unique last-minute
pricing mechanism that distinguishes itself from traditional OTAs. Well-developed OTA pricing
strategies may not work on HotelTonight; therefore, to take full advantage of last-minute sales, hotels
need to understand how such deals work in order to develop effective pricing strategies via HotelTonight.
In academia, HotelTonight, as an alternative distribution channel for hotels, has yet to garner significant
attention. The only previous study on HotelTonight looked at whether hotels’ performance was influenced
by using HotelTonight without a close examination of its unique last-minute pricing mechanism (Makki,
et al., 2016). To bridge this research gap, the present empirical study attempts to reveal the nature of last-
minute price discounts and their determining factors based on a multi-channel price comparison between
HotelTonight and traditional OTAs. Specifically, this study intends to address the following research
questions: 1) Is there a price disparity between HotelTonight and OTAs?; and 2) What are the factors
influencing this cross-channel price disparity (i.e., last-minute price discounts)? By answering these
research questions, this study intends to contribute to the hotel literature by shedding light on last-minute
pricing and, more importantly, understanding cross-channel price differentiation, which have been largely
untapped by previous literature. From a practitioner’s perspective, the findings of this study will help
facilitate hotel operators’ pricing decisions, revenue management, and online channel management.
2. Literature review
2.1. Hotel online distribution channels and pricing behavior
With the evolution of the Internet and various types of online distribution channels, Middleton
and Clarke (2001) predicted that the future hospitality market would never be dominated by a single
distribution channel. Therefore, hotels use an array of online and offline distribution channels to
maximize sales and market share while minimizing costs (Kang, Brewer, & Baloglu, 2007; Toh, Raven,
& DeKay, 2011). OTAs and third-party websites play an important role in hotels’ multi-channel
distribution strategy because they offer a vast selection of choices in a wide price range (Thakran &
Verma, 2013). Although hotels prefer to sell rooms through their own websites to reduce costs, OTAs
remain essential for hotel sales (Beritelli, Beritelli, Schegg, & Schegg, 2016) because of their accessibility
to potential consumers (Morosan & Jeong, 2008) and economies of scale associated with the large
number of hotels listed (Kim, Bojanic, & Warnick, 2009).
Various third-party websites (OTAs) use different business models, such as global distribution
system (GDS)-based travel agent models (Law, et al., 2007), merchant models (Leung, Guillet, & Law,
2014), opaque models (Anderson & Xie, 2014), and last-minute booking models (Jerath, Netessine, &
Veeraraghavan, 2010) to provide enticing deals to attract customers. In a study of hotel price fluctuations
in OTAs, Sun, Law, and Tse (2016) concluded that the last-minute booking mode was less frequently
adopted by OTAs, with most OTAs failing to keep their promise of a “best-rate guarantee.” In a case
study of Mediterranean hotels, Melis and Piga (2017) found that OTAs prefer uniform pricing to dynamic
pricing. OTAs’ and hotels’ optimal pricing strategies to maximize revenue have received considerable
academic attention. For example, Guo, Ling, Dong, and Liang (2013) and Ling, Guo, and Yang (2014)
explored an optimal pricing strategy through hotel–OTA cooperation based on a game model. Their
findings indicated that optimal pricing depends on hotel room capacity, occupancy rate, average room
rate, and the expected number of e-tourists. In addition, Guo, Ling, and Gao (2016) further analyzed
OTAs’ and hotels’ optimal pricing decisions when OTAs offered cash back for hotel bookings. Likewise,
Long and Shi (2017) studied optimal pricing strategies in the O2O model through cooperation between an
offline tour operator and an OTA based on a game model. The findings suggested that service level, unit
sale commission, service cost, and unit service compensation coefficient each influenced pricing
2.2. Hotel revenue management and dynamic pricing
Revenue management or yield management is the process of matching supply and demand by
allocating capacity to different customer segments at the right price in order to maximize revenue (Ivanov
& Zhechev, 2012; Upchurch, Ellis, & Seo, 2002). Revenue management practices have often been
embraced by industries with one or more of the following attributes: perishable inventory, limited
capacity, demand volatility, micro-segmentation of the market, advance booking availability, and high
fixed costs (Ivanov & Zhechev, 2012; Kimes, 1989). Because the hotel industry possesses all these
characteristics, revenue management techniques have been widely adopted to improve room revenue
(Kimes & Wirtz, 2003). More specifically, revenue management in the hotel industry focuses on using
information systems and pricing strategies to manipulate room rates in response to forecasted demand
(Kimes & Wirtz, 2003). Thanks to new technologies that make it easier to predict demand and adjust
prices accordingly (Elmaghraby & Keskinocak, 2003), dynamic pricing has enjoyed increasing popularity
in hotel revenue management (Ropero, 2011). Basically, dynamic pricing refers to the tactical practice of
determining optimal room rates contingent upon the day and time when a reservation is received (Abrate,
Fraquelli, & Viglia, 2012; Kannan & Kopalle, 2001).
Due to the perishability nature of hotel rooms, dynamic pricing on last-minute bookings plays a
vital role in hotel revenue management (Chen & Schwartz, 2013). Some hotels are intent on maximizing
occupancy by adopting last-minute deal pricing strategies or reducing prices closer to the date of stay
(Abrate, et al., 2012; Dacko, 2004). However, researchers have also warned that last-minute deals could
train customers to delay their purchases, which may ultimately compromise hotels’ long-term profitability
(Ovchinnikov & Milner, 2012). In fact, many hotels, especially those that are high-end, try to keep their
last-minute prices the same or even slightly higher than normal to protect long-term profitability and
brand equity (Abrate, et al., 2012; Dacko, 2004; Oses, Gerrikagoitia, & Alzua, 2016). Some studies have
also revealed that last-minute pricing strategies vary depending on seasonality (Dacko, 2004), guest
segmentation (Abrate, et al., 2012), stay day of the week (Abrate, et al., 2012; Oses, et al., 2016), and
hotel star rating (Abrate, et al., 2012; Oses, et al., 2016).
2.3. Multi-channel price disparity
For consumers who purchase travel products online, price is one of the most important motivators
behind their purchase decisions (Law & Wong, 2010; O’Connor, 2002). Because today’s travelers can
easily compare room rates across various online channels, cross-channel price parity has become a critical
issue for hoteliers and guests alike (Choi & Mattila, 2005; Gazzoli, Gon Kim, & Palakurthi, 2008). Kim,
Cho, Kim, and Shin (2014) demonstrated that online price parity exerts a positive impact on hotel
performance. However, most studies on rate parity have found room rates to be inconsistent across
distribution channels (Toh, et al., 2011). Researchers discovered that local travel agencies offer the lowest
rates for upscale and luxury hotels, whereas national OTAs tend to provide the best rates for mid-priced
hotels (Hui, Law, & Ye, 2009; Tso & Law, 2005). Surprisingly, hotels’ websites have been found to post
the most expensive prices (Hui, et al., 2009). Some scholars have also investigated room rate parity across
media. For example, Demirçiftçi, Cobanoglu, Beldona, and Cummings (2010) found no significant
difference between room rates from direct or indirect channels at an aggregate level. In the case of UK
hotels, Lim and Hall (2008) confirmed that hotel prices remained consistent across online and offline
distribution channels, and in the case of Macao hotels, Lee, Tang, and Fong (2016) observed room price
parity for three-star and four-star hotels only.
In the marketing field, market characteristics, retailer characteristics, and product characteristics
have been identified as key factors influencing multi-channel price differentiation (Chiu, Chu, & Wu,
2016; Eckert, 2011; Wolk & Ebling, 2010). Wolk and Ebling (2010) found that higher levels of online
competition and online reach foster greater price parity, whereas large companies with high revenue and
strong market power are more likely to engage in cross-channel price differentiation. Eckert (2011) also
identified that channel-based price differentiation is highest for service products, in terms of product type.
Chiu, et al. (2016) stated that high service quality with added benefits allows companies to increase price
dispersion. In the hospitality industry, researchers have also identified several factors that impact price
parity, such as the website business model (Tso & Law, 2005), hotel class (Hui, et al., 2009; O’Connor,
2003), hotel star rating (Law & Wong, 2010; Lim & Hall, 2008), and transaction booking time (Leung, et
al., 2014). For example, Lee, et al. (2016) found that high-class hotels tend to have greater price disparity
than low-class ones, and Toh, et al. (2011) indicated that small hotels are more likely to embrace price
parity across channels. Because hotels are generally likely to sell empty rooms at the last minute at low
prices to improve occupancy, studies have shown that channels offering last-minute booking also provide
the lowest rates (Law, et al., 2007; Leung, et al., 2014).
3. Model conceptualization and hypotheses development
Hotel pricing studies have been a popular research topic in revenue management since the early
1990s. Wang and Nicolau (2017) summarized five categories of consumer- and supplier-side factors that
influence hotel pricing: site-specific characteristics (e.g., hotel location); quality-signaling factors (e.g.,
hotel star rating, online rating, and chain affiliation); hotel services and amenities (e.g., availability of
complimentary breakfast and WiFi); property characteristics (e.g., an onsite restaurant and fitness center);
and external market factors (e.g., number of competitors and market accessibility). By comparing last-
minute prices on HotelTonight to the best available rates from major OTAs, this study evaluates cross-
channel price disparities by investigating last-minute price discounts. Three categories of multi-channel
price differentiation determinants have been identified through marketing research: retailer (hotel)
characteristics, product characteristics, and market characteristics (Chiu, et al., 2016; Eckert, 2011; Wolk
& Ebling, 2010). This study develops the following research hypotheses and proposed model based on
factors from these three categories and their influence on last-minute hotel room price discounts (see
(Please insert Figure 1 about here)
First of all, quality or reputation has always been an important retailer characteristic in
differentiating price (Abrate, et al., 2012). According to Zhao, Wang, Guo, and Law (2015), two major
online reputation metrics exist: (1) review valence, which is measured by the review rating reflecting
either positive or negative feedback about product attributes; and (2) review volume, measured by the
total amount of reviews reflecting a product’s online popularity. In an online distribution channel, both
review valence and review volume are considered important indicators of service quality and reputation
(Abrate, Capriello, & Fraquelli, 2011; de la Peña, Núñez-Serrano, Turrión, & Velázquez, 2016; Sparks &
Browning, 2011). Schamel (2012) empirically confirmed the significantly positive relationship between
hotel room rates and online ratings based on customer reviews from an online meta-booking engine.
Regarding review volume, de la Peña, et al. (2016) found that hotels receiving a higher-than-average
number of reviews are associated with higher room rates after controlling for other price determinants.
Based on past literature, better online reputation metrics lead to a price premium in room rates published
on OTAs. Once a hotel earns a decent reputation on TripAdvisor, the most popular online reputation
system used by many OTAs (Yoo, Sigala, & Gretzel, 2016), it is likely to leverage this advantage by
using OTA channels more often. Therefore, a hotel with a better relative TripAdvisor reputation becomes
reluctant to offer deeper discounts via HotelTonight, which has its own independent online reputation
system. Two hypotheses are thus proposed as follows:
H1a: A higher relative review valence on TripAdvisor leads to a lower last-minute price discount
H1b: A higher relative review volume on TripAdvisor leads to a lower last-minute price discount
Hotel amenities and services represent important product characteristics in multi-channel price
differentiation (Abrate, et al., 2012). For instance, hotel rates were found to be higher if hotels offered in-
room amenities such as mini-bars, televisions, and safes (Lee & Jang, 2012; Schamel, 2012). Moreover,
some hotel services were associated with higher room rates, such as valet parking or free parking, express
checkout, complimentary breakfast, and an onsite pool (Schamel, 2012; Wang & Nicolau, 2017; Yang,
Mueller, & Croes, 2016). Hotel services can be further categorized based on their variable costs. Some
services like complimentary parking and WiFi access are characterized by low variable costs, whereas
others such as complimentary breakfast are characterized by higher variable costs (Frey Jr & Clyman,
2017). We argue that services with higher variable costs are likely to influence price discounts on last-
minute distribution channels. On one hand, the perceived fairness principle of revenue management states
that customers believe a higher hotel price associated with increasing costs is fair (Kimes, 1994);
therefore, they are willing to pay a premium price for products with a higher cost. On the other hand,
hotels may consider cost factors when making pricing decisions (Collins & Parsa, 2006); to maintain a
reasonable level of profit margin, hotels tend to offer limited room rate discounts for reservations with
complimentary services that have high variable costs. The second hypothesis posits:
H2: Offering complimentary access to services with high marginal variable costs leads to a lower
last-minute price discount on HotelTonight.
Another product characteristic influencing multi-channel price differentiation in the hotel industry
is room type uncertainty while booking. Unlike bookings via traditional OTAs, last-minute bookings via
HotelTonight may not provide transparent room selection. Many hotels did not specify room type on the
app; rooms may be either one-bed or two-bed layouts. As most customers are risk-averse by nature
(Bailey, Olson, & Wonnacott, 1980), they are willing to pay less when confronted with this type of
uncertainty. From the hotel’s perspective, choosing not to reveal the room type provides greater flexibility
in inventory management (Vinod, 2004), especially at the last minute, so the hotel is willing to lower the
price on last-minute deals. A typical example of room type uncertainty comes from opaque selling
channels for hotel products, where certain characteristics of the product or service, including room type,
are hidden during the purchase process (Anderson & Xie, 2012). By offering a lower rate, opaque pricing
allows hotels to effectively differentiate pricing on those channels while maintaining price parity on
transparent price channels (Anderson, 2009). The third hypothesis therefore states:
H3: Uncertainty in room type when booking leads to a higher last-minute price discount on
The last determinant category, marketing characteristics, such as market structure and the
competition landscape also influence hotel multi-channel pricing (Becerra, Santaló, & Silva, 2013). For
example, hotel pricing has been shown to be negatively impacted by the number and proximity of
competitors (Balaguer & Pernías, 2013; Becerra, et al., 2013). Another important external factor
influencing hotel pricing is market power. Market power refers to hotels’ ability to influence prices above
a competitive level to earn greater profits, a phenomenon referred to as monopoly power (Nappu, Bansal,
& Saha, 2013). The concept of market power originated from the structure-conduct-performance
industrial organization paradigm, which reveals a positive relationship between the market concentration
(market power) and profitability of a firm (Bain, 1951). Generally, strong market power and brand power
decrease consumers’ price sensitivity, thereby empowering a company to charge higher prices (Wolk &
Ebling, 2010). In the context of the airline industry, Borenstein (1989) indicated that an airline with high
market power was able to increase airfares. Stavins (2001) further found that an airline’s price
discrimination decreased with higher market power, meaning that the airline offered fewer price
discounts. In the hotel industry, a positive relationship between market power and hotel profitability has
been well established (Pan, 2005; Tung, Lin, & Wang, 2010). We therefore argue that hotels possessing
strong market power tend to focus more on conventional OTAs than on last-minute channels to sell
inventory, and they are less likely to offer impressive last-minute deals given a lack of competition. It is
thus hypothesized that:
H4: A hotel’s stronger market power leads to lower last-minute price discounts on HotelTonight.
4. Research method
4.1. Data collection
We trained the coders to collect data manually from the HotelTonight mobile app in six major
U.S. lodging markets: New York, NY; Atlanta, GA; Chicago, IL; Houston, TX; Denver, CO; and Los
Angeles, CA. These cities were selected for two reasons. First, being located in different regions, these
sampled cities are geographically representative of the U.S. Second, according to Smith Travel Research,
they are among the top 25 U.S. lodging markets (Canina & Carvell, 2005). Although HotelTonight offers
reservations up to three days after the booking date, we examined only same-day reservations to evaluate
the last-minute nature of hotel booking. On each day, two independent coders collected hotel prices and
other relevant information from the app twice, once at 9am and again at 9pm local time. Because
HotelTonight often offers a ‘Geo Rate’ that can be lower than the already-discounted rate based on the
geo-location of mobile devices, coders were required to turn off the geo-locating function on their mobile
devices. The data collection period spanned 24 days, beginning on May 19 and ending June 11, 2017. For
each city, 15 to 16 hotels were usually listed in the app, and the hotel list could change at different times
over different days. We searched on Google and each city’s DMO website to identify upcoming events
during the research period and did not find any events that might influence the city’s hotel demand any
more than usual. As HotelTonight claims, the listed hotels are carefully selected using the app’s algorithm
to offer consumers the best deal of the day (HotelTonight, 2018). Occasionally, the app indicated that
rooms were sold out at certain hotels. For each hotel, the coder recorded its name, reputation (i.e.,
recommendation percentage and number of ratings), whether complimentary breakfast was provided, the
room type, booking price, and taxes and fees. The app overview is presented in Figure 2. Based on the
city and check-in/check-out dates specified, the app presents a list of a limited number of hotels. After
clicking the link to each hotel, the user can review many hotel-specific characteristics including its
address, amenities, and customer ratings (recommendation percentages) and reviews posted on
HotelTonight by previous users. If the user decides to book, more detailed information is presented about
the reservation, including the room type and total booking cost.
(Please insert Figure 2 about here)
We also incorporated two additional data sources for empirical analysis. First, our coders
searched the lowest available room rate on TripAdvisor for each hotel listed on HotelTonight. Serving as
a mega search engine, TripAdvisor provides a search function that lists hotel room rates from various
OTAs as well as from the hotel’s official website (Yoo, et al., 2016). Therefore, users can compare room
rates from various major OTAs such as Travelocity, Hotwire.com, Orbitz.com, Expedia.com,
Booking.com, and Hotels.com. Note that the room rates from hotel’s official website were excluded
because we consider OTA-listed rates only in this study. Also, the coder recorded the hotel’s online
reputation on TripAdvisor, including its overall customer rating and number of reviews. Second, we
collected operation data from the STR hotel census database for each hotel listed on HotelTonight. The
database covers around 98% of all U.S. hotel properties. To our knowledge, it is the largest of its kind and
is well-received by industry and academia (Kalnins, 2016). Some key hotel characteristics we obtained
included each hotel’s number of rooms, operation type, location type, and class. Also, with this database,
we were able to calculate the total number of hotel rooms in each zip code area that HotelTonight used.
After merging the HotelTonight price data with these two data sources, our final dataset consisted of
4.2. Model specification
To test the proposed research hypotheses outlined in the previous section, we specified a
regression model for last-minute price discounts from HotelTonight. The model was specified as follows:
=+ ++ (1)
where represents the price discount on the HotelTonight app compared to the lowest rates from major
OTAs for hotel i at day t, such that =
. Basically, the price discount is calculated
as the difference between the lowest OTA-listed room rate available (
) and the HotelTonight rate
) divided by the lowest OTA-listed room rate. This discount can also be depicted as a percentage,
showing the percentage higher or lower a HotelTonight rate is compared to the lowest available OTA-
listed rate. can be either positive or negative: a positive value indicates a lower hotel room rate on
HotelTonight versus OTAs, whereas a negative value suggests a higher rate on HotelTonight versus
OTAs. captures independent variables of interest that are used to test research hypotheses, and
captures additional independent variables used as control variables. Lastly, represents a set of dummies
indicating the day of the week, and is the normal error term in the regression that is independently and
normally distributed with a mean of zero and a given variance.
We defined several variables of interest, including Dif_rating, Dif_reviews, Dif_reviews,
Breakfast, Rm_type, and Mkt_power. Definitions of these variables are presented in Table 1. As suggested
by Hypotheses 1a and 1b, the estimated coefficients of Dif_rating and Dif_reviews were expected to be
negative. Furthermore, because a complimentary breakfast is characterized by a relatively high marginal
variable cost compared to complimentary parking and WiFi, the coefficient of Breakfast was expected to
be negative according to Hypothesis 2. Moreover, Hypothesis 3 predicts that the coefficient of Rm_type =
3 would be positive and larger than that of the other two levels, whereas Hypothesis 4 predicts a negative
coefficient for Mkt_power.
(Please insert Table 1 about here)
In addition to these independent variables of interest, we also included a set of control variables
in whose definitions are also presented in Table 1. Because hotels face greater pressure to fill unsold
rooms at later hours, we expected to see higher discounts at 9 pm and thus predicted a positive coefficient
of Booking_hour = 2. Hotel size and chain affiliation have been found to inform hotels’ pricing strategies
and price dispersion across different distribution channels (Toh, et al., 2011); therefore, we included
Rooms and Chain as control variables in the model. We used two additional categorical variables,
Location and Class, to capture a hotel’s location and class, respectively. Bull (1994) and Lee and Jang
(2011) confirmed the effects of geographic location relative to the city center and airport on hotel pricing.
Also, different hotel classes serve different customer segments and face different types of demand in
revenue management (Boyd & Kallesen, 2004). Note that we did not employ the hotel class classification
used in the HotelTonight app (including 'Luxe', 'Hip', 'Solid', 'Basic', 'Charming', 'Crashpad', and 'High
Roller') because the STR classification is considered reputable in both industry and academia (Hua &
Yang, 2017; Kalnins, 2016). Lastly, we incorporated a set of dummy variables to capture the day-of-the-
4.3. Data description
Table 2 presents the descriptive statistics of the continuous variables. First and foremost, the
dependent variable, , which represents the price discount offered via HotelTonight, had a mean value
of 0.116, suggesting that HotelTonight offers an average 11.6% discount compared to the lowest available
OTA-listed room rates. Its standard deviation was remarkably large, and both negative and positive
extreme values existed as indicated by the maximum and minimum values. We also investigated the
distribution of the dependent variable over sampled cities; Figure 3 shows the box plot of this variable in
different cities. We found that in general, the price discounts via HotelTonight were highest in New York
and Chicago and lowest in Atlanta. Moreover, we found a substantial level of variability in the discounts
in most cities as shown by the large distance between two adjacent lines.
(Please insert Table 2 about here)
(Please insert Figure 3 about here)
For the continuous independent variables, the mean of Dif_rating was -0.375, indicating that after
converting the HotelTonight rating into a five-point scale measure, the hotel’s TripAdvisor rating was
about 0.375 point lower than its HotelTonight rating. Moreover, we found that about 90% of our records
had a negative value for Dif_rating. These results suggest that hotels listed on HotelTonight usually
maintain a higher reputation on HotelTonight than on TripAdvisor. Another online reputation variable,
Dif_rating, had a mean value of 0.086, suggesting hotels have an average of 86 more reviews on
TripAdvisor compared to HotelTonight. This variable was also characterized by a large range as indicated
by its standard deviation and minimum and maximum numbers. Lastly, the mean values of Mkt_power
and Rooms suggested that guests tended to book a hotel via HotelTonight that accounts for an average of
15.8% of total rooms in the surveyed zip codes and an average of 266 guest rooms.
Table 3 presents the descriptive statistics of the categorical variables. We found that only a small
proportion, 10.78%, of price records included complimentary breakfast. Moreover, among the three room
types offered on HotelTonight, 76.07% were rooms selected by the hotel. Because of our balanced
sampling process, we had an approximate 50/50 division between morning and evening price records in
the sample. Furthermore, 63.91% of records in our sample were from chain-affiliated hotels. Regarding
hotel location, hotels in suburban and urban locations dominated. In particular, 68.06% of records came
from urban hotels. As for hotel class, high-end hotels dominated our sample, with 42.91% of records from
upper upscale hotels, 26.42% from upscale hotels, and 16.99% from luxury hotels. Lastly, during our 24-
day data collection period, we slightly over-sampled Friday, Saturday, and Sunday records. A matrix of
Pearson correlations was estimated to detect the multi-collinearity problem (see Appendix). Except for the
coefficient between dummies of the same categorical variable, pairwise coefficients were below 0.40,
showing an absence of multi-collinearity in our sample (Gujarati & Porter, 2010). We also calculated
variance inflation factor (VIF) values for all independent variables; the average VIF was 3.27.
(Please insert Table 3 about here)
5. Empirical results
We first estimated a model with control variables only; Model 1 in Table 4 presents these results.
The estimated coefficient of Booking_hour was neither significant nor consistent with our prediction,
suggesting that evening booking on HotelTonight for same-day check-in did not necessarily offer a better
deal. One possible explanation is that some popular hotels with better deals have already filled their
inventory before the evening, so different hotels are listed in the evening whose discounts may be pre-
determined by hotel managers when posting on HotelTonight. Moreover, we found the estimate of Rooms
to be negative and significant, suggesting that large hotels usually offered a lower discount via
HotelTonight. The magnitude of this estimate implies that a hotel with 10 more rooms is associated with a
price discount 0.114% lower on last-minute channels. The coefficient of another variable, Chain, was
estimated to be insignificant, and out of five dummies of Location, only the coefficient of Location = 5
was estimated to be significant and negative, suggesting that suburban hotels usually offered a lower
discount on HotelTonight compared to airport hotels (the reference group).
(Please insert Table 4 about here)
Regarding hotel class, the results of Model 1 show that compared to upper midscale hotels (the
reference group; Class = 3), midscale hotels (Class = 2) provided lower HotelTonight discounts whereas
hotels in other classes provided higher discounts. As indicated by the estimated coefficients, upper
upscale hotels (Class = 5) and luxury hotels (Class = 6) provided the highest last-minute discounts with
average rates that were 5.03% and 4.48% higher than those of midscale hotels. Figure 4 demonstrates the
predicted mean price discounts for hotels in different classes after controlling for other independent
variables. Interestingly, it shows that both very low-end (Class = 1, economy class) and very high-end
hotels (Class = 5, upper upscale class; Class = 6, luxury class) offered better discounts via HotelTonight.
In general, we found that after controlling for other factors, upper upscale hotels offered the highest
HotelTonight discount, which was about 13.1% lower than the lowest room rate available online. This
percentage was followed by the discount offered by luxury hotels (12.6%) and that offered by economy
hotels (12.5%). The comparatively impressive last-minute discounts offered by high-end hotels likely
represent efforts to protect brand equity (Xu & Chan, 2010) and maintain a reasonable level of perceived
price fairness for guests (Chung & Petrick, 2013); hence, these hotels appear more likely to lower their
room rates on less transparent booking channels, such as HotelTonight, instead of on transparent channels
like OTAs. For economy-class hotels, pricing mechanisms are among the few strategic tools they can
leverage in an absence of affluent resources and enticing features (Kalnins & Chung, 2004); thus, a heavy
price discount on less transparent booking channels might prove useful.
(Please insert Figure 4 about here)
Model 1 also presents the estimated coefficients of dummies indicating the day of the week.
Compared to the reference group (Monday), HotelTonight discounts were significantly higher on
Thursdays and Sundays and lower on Saturdays. More specifically, according to the estimated
coefficients, the last-minute discount rate was 2.34% higher on Sunday and 4.18% higher on Thursday
than on Monday. Figure 5 visualizes the predicted mean price discounts for hotels on different days of the
week after controlling for other independent variables. In general, we found that the hotel discounts via
HotelTonight were highest on nights adjacent to weekend nights. More specifically, the predicted mean
price discount was 15.0% for Thursday night and 13.2% for Sunday night. These discounts were lowest
on a typical weekend night, namely Saturday, when the predicted mean discount was only 8.1%. One
possible reason for this is that hotels experience higher demand on weekends and are therefore more
likely to achieve higher occupancy (Schamel, 2012), providing less incentive to offer significant
discounts via HotelTonight.
(Please insert Figure 5 about here)
Table 4 also presents the estimation results of models testing our proposed research hypotheses.
To minimize the impact of any multi-collinearity issues, we introduced each variable of interest
separately in addition to the specification of Model 1. In Model 2, we added Dif_rating, and its coefficient
was estimated to be negative and significant. This result corroborates Hypothesis 1a, and the coefficient
shows that a one-point increase in the rating gap between TripAdvisor and HotelTonight (on a five-point
scale) led to a 2.01% decrease in the price discount offered on HotelTonight. In Model 3, Dif_reviews was
added to test Hypothesis 1b. Its estimated coefficient was significant and negative, lending empirical
support to Hypothesis 1b. This result shows that a 100-count increase in the gap between TripAdvisor and
HotelTonight review totals led to a 0.13% decrease in price discounts via HotelTonight. In general, our
empirical results support Hypothesis 1, suggesting that price discounts on HotelTonight rely heavily on
the relative online reputation of TripAdvisor compared to HotelTonight.
In Model 4, the coefficient of Breakfast was estimated to be negative and significant, thus
supporting Hypothesis 2. Breakfast’s estimated coefficient indicates that the discount would be 3.74%
lower if hotels offered complimentary breakfast with booking. In Model 5, we added two dummies of
Rm_type by setting Rm_type = 2 as the reference group; the coefficient of Rm_type = 3 was estimated to
be statistically significant and positive. Hence, Hypothesis 3 is supported, and the result indicates a 10.6%
higher discount for an uncertain room type compared to a guaranteed double room. Another variable of
interest, Mkt_power, was introduced in Model 6, and its coefficient is not statistically significant.
Therefore, Hypothesis 4 is rejected, as we did not find any evidence supporting the role of market power
in shaping HotelTonight price discounts. One explanation for this result is that market power does not
explain the extent of price discrimination in the presence of strong consumer heterogeneity in brand
preferences (Holmes, 1989). Lastly, we included all independent variables in Model 7. The estimates of
the variables of interest changed very little; most notably, their signs and statistical significance did not
change, suggesting a reasonable level of robustness of our results with regard to model specification.
We ran an additional regression analysis based on Model 7 to further explain the results of
hypothesis testing; Table 5 presents these results. First, in Model 8, we considered the interaction effect of
relative review valence and review volume (Dif_rating * Dif_reviews). As suggested by Maslowska,
Malthouse, and Viswanathan (2017), review volume reflects review valence credibility; therefore, a large
review volume can strengthen the beneficial effect of positive valence and the detrimental effect of
negative valence. In this model, the coefficient of the interaction term was estimated to be negative and
significant, highlighting the significant interaction effect between relative TripAdvisor valence and
volume in shaping last-minute discounts. Second, as suggested by Hypothesis 2, offering complimentary
services with lower marginal variable costs should have less effect on last-minute price discounts.
Therefore, we incorporated dummy variables, indicating the availability of free parking (Parking = 1) and
complimentary WiFi access (WiFi = 1), into Model 9. The coefficients of these two additional variables
were estimated to be insignificant, providing further support for Hypothesis 2. Third, when developing
Hypothesis 3, we argued that hotels offer better last-minute deals for unspecified room types in order to
allow for more last-minute flexibility in inventory management. We thus introduced an interaction term
of uncertain room type (Rm_type = 3) and number of hotel rooms (rooms) in Model 10, and the term’s
coefficient was estimated to be negative albeit not significant at the 0.05 level. Lastly, we wanted to
further investigate Hypothesis 4, which was rejected by Model 6. Initially, we treated each single hotel
room as being equal irrespective of hotel class when constructing the market power measure. Therefore,
we introduced an interaction term of market power (Mkt_power) and hotel class (Class) in Model 11. The
coefficient of this interaction term was estimated to be negative and statistically significant, while the
main effect of market power was positive and statistically significant. This result suggests that only high-
end hotels, such as luxury hotels (Class = 6), are expected to consider market power when offering last-
minute discounts. In sum, although high-end hotels in general are likely to offer a better last-minute deal
(see Figure 4) than others, those with weaker market power tend to make the deal even better.
(Please insert Table 5 about here)
The issue of price disparity across multiple distribution channels has been an important topic in
hotel channel management since the rapid proliferation of online distribution options. With the advent of
mobile technology, price disparity has extended to online channels and mobile channels. Based on data
collected from a mobile last-minute booking channel (HotelTonight) and online channel (traditional
OTAs), this study presented an empirical analysis exploring cross-channel price disparities and their
underlying factors. The study investigated five factors influencing last-minute price discounts offered on
HotelTonight compared to OTAs using three channel-based price differentiation determinant categories in
marketing research. Review valence and volume, reflecting hotel reputation, were selected as retailer
(hotel) characteristics factors. Complimentary services with high marginal costs and room type
uncertainty in booking were selected as product characteristics factors. Market power was considered a
market characteristics factor.
Several findings were confirmed after regression analysis. First, the results showed that a higher
relative review valence and volume on TripAdvisor compared to HotelTonight led to lower last-minute
price discounts on HotelTonight, confirming previous marketing research showing that a positive seller
reputation enhances multi-channel price differentiation (Chiu, et al., 2016). Moreover, offering
complimentary access to services with high marginal variable costs, such as complimentary breakfast, led
to lower price discounts, whereas providing complimentary access to other services, like free parking and
free WiFi, did not influence discounts. Previous marketing research related to product characteristics
mainly focused on product types, such as durables, non-durables, resalable products, or service products
(Chiu, et al., 2016). Our study represents the first research effort to innovatively investigate the
relationship between products’ marginal costs and channel-based price differentiation. Furthermore, our
results indicated that uncertainty in room type when booking led to higher last-minute price discounts on
HotelTonight, particularly for smaller hotels where it is more feasible to offer greater discounts. These
findings may be explained by previous results suggesting that high product differentiation leads to
channel-based price premiums (Chiu, et al., 2016; Eckert, 2011). When the room type is uncertain, hotel
products become more homogeneous, ultimately resulting in less product differentiation. Therefore, the
hotel also loses the power to charge a price premium on last-minute booking channels. Last but not least,
market power was generally unable to explain price discounts on HotelTonight, contradicting previous
market research findings (Wolk & Ebling, 2010). However, we did find that market power shaped
discounts for luxury hotels. Regarding other factors used as control variables in the model, we found that
smaller hotels, higher class hotels, and hotels located in urban areas usually offered better HotelTonight
discounts, especially for last-minute check-ins on Thursdays and Sundays, confirming findings from Lee,
et al. (2016).
From a theoretical perspective, this study makes a pioneering research effort to collectively
examine, in a single empirical setting, two different dimensions of revenue management: cross-channel
price differentiation and last-minute pricing. The study contributed to the current literature on hotel
pricing in two major ways. First, unlike empirical studies focusing on room rate data from a single
distribution channel or platform (Yang, et al., 2016), we investigated price disparities across different
channels and platforms. Although some pioneering studies discovered cross-channel price disparities
among different online channels (Demirçiftçi, et al., 2010; Gazzoli, et al., 2008; Lim & Hall, 2008), this
study was the first to apply the marketing theory to unveil systematic underlying factors shaping price
disparities that have not been empirically evaluated in the literature. A better understanding of price
disparity helps clarify the nature of different hotel distribution channels and the advantages and
disadvantages associated with these channels for special hotels. Second, our study represents the first
research effort aimed at understanding price discounts from an emerging last-minute hotel booking app,
HotelTonight, which represents a new business model for hotel distribution channels in the information
technology era. We investigated the effects of some new factors on dynamic pricing of this channel, such
as room type uncertainty and booking hour on the day immediately prior to check-in. Last but not least,
this study substantiated the close relationship between revenue management and reputation management;
that is, tactical revenue management strategies depend heavily on reputation management.
7. Conclusion and Implications
These results provide several important practical implications for hotels. First, this study serves as
an eye-opener for hotel owners and managers who have not yet adopted HotelTonight as a distribution
channel. Compared to traditional OTAs, HotelTonight is still a relatively new product that not all U.S.
hotels use. This study presents a comprehensive picture of how HotelTonight works and its last-minute
pricing mechanism. The study results can thus be used by hotel owners and managers who want to learn
more about HotelTonight and decide whether to participate in this last-minute mobile distribution
channel. Second, for hotels who plan to or currently use HotelTonight, the empirical results from this
study can serve as benchmark parameters or best practices for hotels to propose pricing strategies on the
app. For example, hotels with better ratings on HotelTonight may be able to offer lower discounts to
maximize revenue. High-end hotels with less market power should offer better last-minute deals to
increase sales. The estimates of price discounts/premiums associated with room type uncertainty and the
provision of complimentary services can also be used for cost-benefit analysis once cost items are
available to formulate best practices for cross-channel/platform pricing. Last, this study highlights the
importance of reputation management on revenue management, especially considering that HotelTonight
has its own reputation system independent from TripAdvisor. Our results show that a better relative
HotelTonight reputation leads to lower HotelTonight discounts (i.e., a higher relative room rate on the
app). Therefore, HotelTonight is particularly promising as a distribution channel for hotels who suffer
from an unsatisfactory TripAdvisor reputation: having a better reputation on HotelTonight creates
opportunities to remain competitive.
The study findings also present practical implications for hotel distribution channels. On one
hand, for HotelTonight, the results revealed attributes of hotels that take greater advantage of last-minute
pricing strategies and that may be potential participants in the future. For example, this study found that
smaller, higher-class hotels located in urban areas usually offer better price discounts. In the future,
HotelTonight could target this type of hotel to increase the app’s ability to price discriminate from
traditional OTAs. On the other hand, for traditional OTAs like Expedia and Priceline, price disparities
between last-minute booking channels as revealed in this study could be interpreted as a niche market
suggestion. Especially due to the mobile nature of HotelTonight, it would be unusual for price disparity to
be evident to the general online public to the point that it would impair hotel brand equity. Traditional
OTAs could learn from this and create a mobile-only last-minute booking platform to supplement their
mainstream online booking channels.
Due to the high cost of manual data collection, our empirical sample only covered a relatively
short period of time and a limited number of hotels in a handful of cities. However, the focus of our study
was the cross-channel price disparity between last-minute booking apps and OTAs. Because the sample
hotels in this study are representative of the hotel population on last-minute booking apps, which is much
smaller than the population on OTAs, this study’s findings are meaningful and offer a comprehensive
picture of multi-channel price differentiation. Second, because of the scale difference in the rating
systems between TripAdvisor and HotelTonight, some compatibility issues may exist when calculating
rating differences. Third, due to difficulties with data collection, we did not analyze another innovative
pricing discrimination strategy called ‘Geo Rate’ from HotelTonight, which offers higher price discounts
for some hotels based on app users’ geographic location. Lastly, we analyzed how cross-channel/platform
pricing strategies are integrated with the general dynamic pricing strategy, and we suspect that a tactical
combination of the two pricing strategies can be more effective in hotel revenue management (e.g., using
last-minute deals to complement hotels’ general dynamic pricing strategy). Therefore, we recommend that
future research efforts investigate the interaction effect between cross- channel/platform pricing and
dynamic pricing using a more comprehensive dataset with a longer research period and more cities
Abrate, G., Capriello, A., & Fraquelli, G. (2011). When quality signals talk: Evidence from the Turin
hotel industry. Tourism Management, 32, 912-921.
Abrate, G., Fraquelli, G., & Viglia, G. (2012). Dynamic pricing strategies: Evidence from European
hotels. International Journal of Hospitality Management, 31, 160-168.
Anderson, C. K. (2009). Setting prices on Priceline. Interfaces, 39, 307-315.
Anderson, C. K., & Xie, X. (2012). A choice‐based dynamic programming approach for setting opaque
prices. Production and Operations Management, 21, 590-605.
Anderson, C. K., & Xie, X. (2014). Pricing and market segmentation using opaque selling mechanisms.
European Journal of Operational Research, 233, 263-272.
Bailey, M. J., Olson, M., & Wonnacott, P. (1980). The marginal utility of income does not increase:
Borrowing, lending, and Friedman-Savage gambles. The American Economic Review, 70, 372-
Bain, J. S. (1951). Relation of profit rate to industry concentration: American manufacturing, 1936–1940.
The Quarterly Journal of Economics, 65, 293-324.
Balaguer, J., & Pernías, J. C. (2013). Relationship between spatial agglomeration and hotel prices.
Evidence from business and tourism consumers. Tourism Management, 36, 391-400.
Becerra, M., Santaló, J., & Silva, R. (2013). Being better vs. being different: Differentiation, competition,
and pricing strategies in the Spanish hotel industry. Tourism Management, 34, 71-79.
Beritelli, P., Beritelli, P., Schegg, R., & Schegg, R. (2016). Maximizing online bookings through a multi-
channel-strategy: Effects of interdependencies and networks. International Journal of
Contemporary Hospitality Management, 28, 68-88.
Bishop, J. (2017). Three apps for getting the best last-minute hotel deals. Forbes. Retrieved January 22,
2018, from https://www.forbes.com/sites/bishopjordan/2017/03/09/last-minute-hotel-
Borenstein, S. (1989). Hubs and high fares: Dominance and market power in the US airline industry. The
RAND Journal of Economics, 344-365.
Boyd, E. A., & Kallesen, R. (2004). Practice papers: The science of revenue management when
passengers purchase the lowest available fare. Journal of Revenue and Pricing Management, 3,
Buhalis, D., & Law, R. (2008). Progress in information technology and tourism management: 20 years on
and 10 years after the Internet—The state of eTourism research. Tourism Management, 29, 609-
Bull, A. O. (1994). Pricing a motel's location. International Journal of Contemporary Hospitality
Management, 6, 10-15.
C.S.-W. (2013). Hotel-booking apps: Touch here for a bed. Retrieved February 4, 2013, from
Canina, L., & Carvell, S. (2005). Lodging demand for urban hotels in major metropolitan markets.
Journal of Hospitality & Tourism Research, 29, 291-311.
Chen, C.-C., & Schwartz, Z. (2013). On revenue management and last minute booking dynamics.
International Journal of Contemporary Hospitality Management, 25, 7-22.
Chiu, P.-H., Chu, K.-K., & Wu, Y. (2016). Exploring the effects of major characteristics on price
differentiation in multi-channel retailers. International Journal of Organizational Innovation
(Online), 9, 124-141.
Choi, S., & Mattila, A. S. (2005). Impact of information on customer fairness perceptions of hotel
revenue management. Cornell Hotel and Restaurant Administration Quarterly, 46, 444-451.
Chung, J. Y., & Petrick, J. F. (2013). Price fairness of airline ancillary fees: An attributional approach.
Journal of Travel Research, 52, 168-181.
Collins, M., & Parsa, H. G. (2006). Pricing strategies to maximize revenues in the lodging industry.
International Journal of Hospitality Management, 25, 91-107.
Criteo. (2016). Travel flash report 2016: Get ready for a mobile summer. Retrieved January 22, 2018,
Crook, J. (2017). HotelTonight to expand booking window to 100 days. Retrieved September 7, 2017,
Dacko, S. G. (2004). Marketing strategies for last-minute travel and tourism: Profitability and revenue
management implications. Journal of Travel & Tourism Marketing, 16, 7-20.
de la Peña, M. R., Núñez-Serrano, J. A., Turrión, J., & Velázquez, F. J. (2016). Are innovations relevant
for consumers in the hospitality industry? A hedonic approach for Cuban hotels. Tourism
Management, 55, 184-196.
Demirçiftçi, T., Cobanoglu, C., Beldona, S., & Cummings, P. R. (2010). Room rate parity analysis across
different hotel distribution channels in the US. Journal of Hospitality Marketing & Management,
Eckert, C. (2011). Different channel - different price? Investigating the practice of multi-channel price
differentiation. GfK Marketing Intelligence Review, 3, 50-53.
Elmaghraby, W., & Keskinocak, P. (2003). Dynamic pricing in the presence of inventory considerations:
Research overview, current practices, and future directions. Management Science, 49, 1287-1309.
Frey Jr, S. C., & Clyman, D. R. (2017). Voyager Inn International. Darden Business Publishing Cases, 1-
Gaggioli, A. (2015). How to reach last minute guests with HotelTonight. Cloudbeds. Retrieved January
22, 2018, from https://www.cloudbeds.com/articles/how-to-reach-last-minute-guests-with-
Gazzoli, G., Gon Kim, W., & Palakurthi, R. (2008). Online distribution strategies and competition: Are
the global hotel companies getting it right? International Journal of Contemporary Hospitality
Management, 20, 375-387.
Gujarati, D. N., & Porter, D. C. (2010). Essentials of Econometrics. New York: McGraw-Hill/Irwin.
Guo, X., Ling, L., Dong, Y., & Liang, L. (2013). Cooperation contract in tourism supply chains: The
optimal pricing strategy of hotels for cooperative third party strategic websites. Annals of Tourism
Research, 41, 20-41.
Guo, X., Ling, L., & Gao, Z. (2016). Optimal pricing strategy for hotels when online travel agencies use
customer cash backs: A game-theoretic approach. Journal of Revenue and Pricing Management,
Holmes, T. J. (1989). The effects of third-degree price discrimination in oligopoly. The American
Economic Review, 79, 244-250.
HotelTonight. (2017). About. Retrieved September 17, 2017, from https://www.hoteltonight.com/about
HotelTonight. (2018). Frequently asked questions. Retrieved January 22, 2018, from
Hua, N., & Yang, Y. (2017). Systematic effects of crime on hotel operating performance. Tourism
Management, 60, 257-269.
Huang, N. (2017). The latest in hotel booking and travel apps. Travel Tripper. Retrieved January 22,
2018, from http://www.traveltripper.com/blog/the-latest-in-hotel-booking-and-travel-apps/
Hui, S., Law, R., & Ye, Q. (2009). An analysis of the best online rates for hotels in China. Journal of
Hospitality Marketing & Management, 18, 829-844.
Ivanov, S., & Zhechev, V. (2012). Hotel revenue management–A critical literature review. Tourism: An
International Interdisciplinary Journal, 60, 175-197.
Jerath, K., Netessine, S., & Veeraraghavan, S. K. (2010). Revenue management with strategic customers:
Last-minute selling and opaque selling. Management Science, 56, 430-448.
Kalnins, A. (2016). Pricing variation within dual-distribution chains: The different implications of
externalities and signaling for high- and low-quality brands. Management Science, 63, 139-152.
Kalnins, A., & Chung, W. (2004). Resource-seeking agglomeration: A study of market entry in the
lodging industry. Strategic Management Journal, 25, 689-699.
Kang, B., Brewer, K. P., & Baloglu, S. (2007). Profitability and survivability of hotel distribution
channels: An industry perspective. Journal of Travel & Tourism Marketing, 22, 37-50.
Kannan, P. K., & Kopalle, P. K. (2001). Dynamic pricing on the Internet: Importance and implications for
consumer behavior. International Journal of Electronic Commerce, 5, 63-83.
Kim, J., Bojanic, D. C., & Warnick, R. B. (2009). Price bundling and travel product pricing practices used
by online channels of distribution. Journal of Travel Research, 47, 403-412.
Kim, W. G., Cho, M., Kim, D., & Shin, G.-C. (2014). The effect of price dispersion on hotel
performance. Tourism Economics, 20, 1159-1179.
Kimes, S. E. (1989). The basics of yield management. The Cornell Hotel and Restaurant Administration
Quarterly, 30, 14-19.
Kimes, S. E. (1994). Perceived fairness of yield management. The Cornell Hotel and Restaurant
Administration Quarterly, 35, 22-29.
Kimes, S. E. (2016). The evolution of hotel revenue management. Journal of Revenue and Pricing
Management, 15, 247-251.
Kimes, S. E., & Wirtz, J. (2003). Has revenue management become acceptable? Findings from an
international study on the perceived fairness of rate fences. Journal of Service Research, 6, 125-
Law, R., Chan, I., & Goh, C. (2007). Where to find the lowest hotel room rates on the internet? The case
of Hong Kong. International Journal of Contemporary Hospitality Management, 19, 495-506.
Law, R., & Wong, R. (2010). Analysing room rates and terms and conditions for the online booking of
hotel rooms. Asia Pacific Journal of Tourism Research, 15, 43-56.
Lee, P. C. B., Tang, H., & Fong, S. W. S. (2016). Price parity, channel conflict, and hotel rooms in
Macao. Tourism Economics, 22, 1431-1439.
Lee, S. K., & Jang, S. (2011). Room rates of U.S. airport hotels: Examining the dual effects of
proximities. Journal of Travel Research, 50, 186-197.
Lee, S. K., & Jang, S. (2012). Premium or discount in hotel room rates? The dual effects of a central
downtown location. Cornell Hospitality Quarterly, 53, 165-173.
Leo, J. (2011). Web Buzz: Hotel Tonight app alerts travelers to same-day deals. In Los Angeles Times.
Leung, R., Guillet, B. D., & Law, R. (2014). The channel that offers the lowest online room rates: A case
study of hotels in Hong Kong. International Journal of Hospitality & Tourism Administration,
Lim, W. M., & Hall, M. J. (2008). Pricing consistency across direct and indirect distribution channels in
South West UK hotels. Journal of Vacation Marketing, 14, 331-344.
Ling, L., Guo, X., & Yang, C. (2014). Opening the online marketplace: An examination of hotel pricing
and travel agency on-line distribution of rooms. Tourism Management, 45, 234-243.
Long, Y., & Shi, P. (2017). Pricing strategies of tour operator and online travel agency based on
cooperation to achieve O2O model. Tourism Management, 62, 302-311.
Makki, A. M., Makki, A. M., Singh, D., Singh, D., Ozturk, A. B., & Ozturk, A. B. (2016). HotelTonight
usage and hotel profitability. Journal of Hospitality and Tourism Technology, 7, 313-327.
Marom, O., & Seidmann, A. (2011). Using “last-minute” sales for vertical differentiation on the Internet.
Decision Support Systems, 51, 894-903.
Maslowska, E., Malthouse, E. C., & Viswanathan, V. (2017). Do customer reviews drive purchase
decisions? The moderating roles of review exposure and price. Decision Support Systems, 98, 1-
Melis, G., & Piga, C. A. (2017). Are all online hotel prices created dynamic? An empirical assessment.
International Journal of Hospitality Management, 67, 163-173.
Middleton, V., & Clarke, J. (2001). Marketing in Travel and Tourism. London: Butterworth Heinemann.
Morosan, C., & Jeong, M. (2008). Users’ perceptions of two types of hotel reservation Web sites.
International Journal of Hospitality Management, 27, 284-292.
Nappu, M. B., Bansal, R. C., & Saha, T. K. (2013). Market power implication on congested power
system: A case study of financial withheld strategy. International Journal of Electrical Power &
Energy Systems, 47, 408-415.
O'Neill, S. (2017). HotelTonight raises another $37 million in funding. Skift. Retrieved January 22,
2018, from https://skift.com/2017/03/22/hoteltonight-37-million-funding-series-e/
O’Connor, P. (2002). An empirical analysis of hotel chain online pricing strategies. Information
Technology & Tourism, 5, 65-72.
O’Connor, P. (2003). On-line pricing: An analysis of hotel-company practices. Cornell Hospitality
Quarterly, 44, 88-96.
Oses, N., Gerrikagoitia, J. K., & Alzua, A. (2016). Evidence of hotels’ dynamic pricing patterns on an
internet distribution channel: The case study of the Basque Country’s hotels in 2013–2014.
Information Technology & Tourism, 15, 365-394.
Ovchinnikov, A., & Milner, J. M. (2012). Revenue management with end‐of‐period discounts in the
presence of customer learning. Production and Operations Management, 21, 69-84.
Pan, C.-M. (2005). Market structure and profitability in the international tourist hotel industry. Tourism
Management, 26, 845-850.
Philips, L. R. (2005). Pricing and Revenue Optimization. Stanford, CA: Stanford University Press.
Quinby, D. (2014). Just how large is last minute on mobile? (With apologies to HotelTonight)
Phocuswright. Retrieved January 22, 2018, from http://www.phocuswright.com/Travel-
Ropero, M. A. (2011). Dynamic pricing policies of hotel establishments in an online travel agency.
Tourism Economics, 17, 1087-1102.
Schamel, G. (2012). Weekend vs. midweek stays: Modelling hotel room rates in a small market.
International Journal of Hospitality Management, 31, 1113-1118.
Schwartz, Z. (2012). “Probability blindness” and last-minute booking of hotel rooms: The case of
Bayesian updating. International Journal of Tourism Sciences, 12, 63-77.
Sparks, B. A., & Browning, V. (2011). The impact of online reviews on hotel booking intentions and
perception of trust. Tourism Management, 32, 1310-1323.
Stavins, J. (2001). Price discrimination in the airline market: The effect of market concentration. The
Review of Economics and Statistics, 83, 200-202.
Sun, S., Law, R., & Tse, T. (2016). Exploring price fluctuations across different online travel agencies: A
case study of room reservations in an upscale hotel in Hong Kong. Journal of Vacation
Marketing, 22, 167-178.
Thakran, K., & Verma, R. (2013). The emergence of hybrid online distribution channels in travel, tourism
and hospitality. Cornell Hospitality Quarterly, 54, 240-247.
Toh, R. S., Raven, P., & DeKay, F. (2011). Selling rooms: Hotels vs. third-party websites. Cornell
Hospitality Quarterly, 52, 181-189.
Tso, A., & Law, R. (2005). Analysing the online pricing practices of hotels in Hong Kong. International
Journal of Hospitality Management, 24, 301-307.
Tung, G.-S., Lin, C.-Y., & Wang, C.-Y. (2010). The market structure, conduct and performance paradigm
re-applied to the international tourist hotel industry. African Journal of Business Management, 4,
Upchurch, R. S., Ellis, T., & Seo, J. (2002). Revenue management underpinnings: An exploratory review.
International Journal of Hospitality Management, 21, 67-83.
Vinod, B. (2004). Unlocking the value of revenue management in the hotel industry. Journal of Revenue
and Pricing Management, 3, 178-190.
Wang, D., & Nicolau, J. L. (2017). Price determinants of sharing economy based accommodation rental:
A study of listings from 33 cities on Airbnb. com. International Journal of Hospitality
Management, 62, 120-131.
Wolk, A., & Ebling, C. (2010). Multi-channel price differentiation: An empirical investigation of
existence and causes. International Journal of Research in Marketing, 27, 142-150.
Xu, B. J., & Chan, A. (2010). A conceptual framework of hotel experience and customer-based brand
equity: Some research questions and implications. International Journal of Contemporary
Hospitality Management, 22, 174-193.
Yang, Y., Mueller, N. J., & Croes, R. R. (2016). Market accessibility and hotel prices in the Caribbean:
The moderating effect of quality-signaling factors. Tourism Management, 56, 40-51.
Yoo, K.-H., Sigala, M., & Gretzel, U. (2016). Exploring TripAdvisor. In R. Egger, I. Gula & D. Walcher
(Eds.), Open Tourism: Open Innovation, Crowdsourcing and Co-Creation Challenging the
Tourism Industry (pp. 239-255). Berlin, Heidelberg: Springer Berlin Heidelberg.
Zhao, X., Wang, L., Guo, X., & Law, R. (2015). The influence of online reviews to online hotel booking
intentions. International Journal of Contemporary Hospitality Management, 27, 1343-1364.
Table 1. Descriptive statistics of continuous variables.
Price discount on the HotelTonight app compared to the lowest rates from major OTAs for
Independent Variables of Major Interest
Relative TripAdvisor review valence compared to that of HotelTonight, calculated as the
difference between a hotel’s TripAdvisor rating and its HotelTonight rating (in %) divided by
Relative TripAdvisor review volume compared to that of HotelTonight, calculated as the
difference between a hotel’s TripAdvisor review number (in thousands) and its HotelTonight
review number (in thousands)
A dummy variable indicating whether complimentary breakfast is included in the HotelTonight
rate. Breakfast = 1 if complimentary breakfast is included, and Breakfast = 0 if not.
A categorical variable capturing the room type booked via HotelTonight. Three room types
are used on HotelTonight: rooms with 1 bed (Rm_type = 1), rooms with 2 beds (Rm_type =
2), and rooms selected by the hotel (Rm_type = 3).
Percentage of a hotel property’s rooms relative to the total number of hotel rooms in the zip
code area of the hotel.
A dummy variable indicating whether the price discount was calculated based on the rates
collected at 9 am (Booking_hour = 1) or 9 pm (Booking_hour = 2).
Number of rooms in the hotel (in thousands)
A dummy variable indicating whether the hotel was affiliated with a hotel chain. Chain = 1 if
the hotel was affiliated with a hotel chain, and Chain = 0 if not.
A categorical variable indicating hotel location type by STR: 1= airport, 2 = interstate, 3 =
resort, 4 = small metro/town, 5 = suburban, and 6 = urban
A categorical variable indicating hotel class by STR: 1 = economy, 2 = midscale, 3 = upper
midscale, 4 = upscale, 5 = upper upscale, and 6 = luxury.
A categorical variable indicating the day of week.
Table 2. Descriptive statistics of continuous variables.
Table 3. Descriptive statistics of categorical variables.
Breakfast = 0 (breakfast excluded) †
Breakfast = 1 (breakfast included)
Rm_type = 1 (room with 1 bed) †
Rm_type = 2 (room with 2 beds)
Rm_type = 3 (room selected by hotel)
Booking_hour = 0 (9am) †
Booking_hour = 1 (9pm)
Chain = 0
Chain = 1
Location = 1 (Airport) †
Location = 2 (Interstate)
Location = 3 (Resort)
Location = 4 (Small Metro/Town)
Location = 5 (Suburban)
Location = 6 (Urban)
Class = 1 (Economy)
Class = 2 (Midscale)
Class = 3 (Upper Midscale) †
Class = 4 (Upscale)
Class = 5 (Upper Upscale)
Class = 6 (Luxury)
Day = Sunday
Day = Monday †
Day = Tuesday
Day = Wednesday
Day = Thursday
Day = Friday
Day = Saturday
(Note: † indicates the reference category of categorical variables used for further regression analysis.)
Table 4. Estimation results of regression models for hypothesis testing.
Breakfast = 1
Rm_type = 1
Rm_type = 3
Chain = 1
Location = 2
Location = 3
Location = 4
Location = 5
Location = 6
Class = 1
Class = 2
Class = 4
Class = 5
Class = 6
Day = Sunday
Day = Friday
(Notes: *** indicates significance at the 0.01 level, ** indicates significance at the 0.05 level, and * indicates
significance at the 0.10 level. Robust standard errors are presented in parentheses.)
Table 5. Estimation results of additional regression models.
Dif_rating * Dif_reviews
Breakfast = 1
Parking = 1
WiFi = 1
Rm_type = 1
Rm_type = 3
Rm_type = 3 * rooms
Mkt_power * Class
Booking_hour = 1
Chain = 1
Location = 2
Location = 3
Location = 4
Location = 5
Location = 6
Class = 1
Class = 2
Class = 4
Class = 5
Class = 6
(Notes: *** indicates significance at the 0.01 level, ** indicates significance at the 0.05 level, and * indicates
significance at the 0.10 level. Robust standard errors are presented in parentheses.)
Figure 2. Screenshots of HotelTonight app.
Figure 3. Box plot of price discounts over sampled cities.
Predicted price discount from HotelTonight
Figure 4. Predicted price discounts on HotelTonight for different hotel classes.
Predicted price discount from HotelTonight
Figure 5. Predicted price discounts on HotelTonight for days of the week.
Table A1. Correlation matrix of independent variables.
st = 1
e = 2
e = 3
hour = 1
n = 2
n = 3
n = 4
n = 5
n = 6
Breakfast = 1
Rm_type = 2
Rm_type = 3
r = 1
Chain = 1
Location = 2
Location = 3
Location = 4
Location = 5
Location = 6
Class = 1
Class = 2
Class = 4
Class = 5
Class = 6
(Note: The correlation coefficients with day of the week variables are not presented for brevity; their correlation coefficients with other
variables are below 0.10.)