Pricing in a Social World: How consumers use ratings, reviews and price when choosing a hotel

With the advent of the Online Travel Agents (OTA’s), prices became transparent, and hotels were forced to pay close attention to how they were priced relative to the competition in the market.  Now with the growing popularity of online ratings and reviews, consumers have additional information to use to evaluate the value of your hotel relative to your competitors’ properties.  In order to continue to make profit-maximizing price and positioning decisions, hotel managers must understand how consumers use this new information with price to make a purchase decision.

As a follow up to our first study, when we found that there is a strong relationship between user-generated content (UGC), or ratings and reviews, and quality and value perceptions of hotel room purchases, Breffni Noone, Associate Professor, The Pennsylvania State University and I wanted to explore a bit further how consumers trade-off these attributes with price.  We designed a choice modeling experiment where we asked consumers to select the hotel they would buy from among a choice of three, with varying levels of attributes.  By following the participant’s choice patterns, the value they place on each attribute, and each level of each attribute can be statistically derived.  As well, the likelihood that they would pick a hotel with a specific combination of attributes can be identified.

The study design

This was a scenario-based, online study.  We recruited a representative sample of the U.S. population via an online survey recruitment company, ensuring that participants had traveled for leisure in the past, and that they had made the booking themselves, online.

We told our participants that they were taking a vacation with a friend, and were looking for a four star hotel in a city center. We provided a selection of three hotels that met their quality and location criteria, and they were asked to indicate which they would buy.  They repeated this exercise three times, and then we asked them to tell us what they were thinking about when they were making their choices.

We varied the price, the name of the hotel, the aggregate rating, the TripAdvisor Rank, the sentiment of the review, the content of the review and the language of the review for each hotel.  Table 1 shows the attributes and levels that we tested.

Table 1: Attributes and Levels tested

Our first study demonstrated the power of the reviews in consumers’ assessment of the quality and value of the hotel purchase, so we wanted to take the opportunity to learn more about how elements of the reviews influenced consumer decision making.  In addition to the valence of the reviews (positive or negative), we tested whether what the reviewers talked about (content) and how they talked about it (language) had an influence on our participants’ choice behavior.

The results

Results showed that the review valence (positive or negative) had the most significant impact on choice behavior, followed by price, then aggregate rating, then TripAdvisor rank.  Known versus unknown brand was marginally significant, with consumers showing a slight preference for the known brand.  The content and language was not a significant influencer of choice.

We think the reason that review content and language were not significant is probably because consumers equally value both the service and physical property of the hotel (this was validated in the open-ended responses we collected).  Further, whether the review was positive or negative appeared so important that it is likely that the impact of the language style was overshadowed.

Figure 1 shows the utility value of each attribute level in the study.  The utility is the relative value the consumers place on each change in level of the attribute.  In this type of analysis, the value of the utilities themselves are less important than the direction of the impact and the magnitude relative to the other metrics.

The bars with an asterisk represent significant utilities.  The red bars represent the negative impact of negative reviews and of raising the price from low to mid, and then from mid to high.  The blue bars represent the positive impact on choice of raising ratings and TripAdvisor rank, as well as the positive impact of a known brand over an unknown brand.

Figure 1: Utility value of each attribute level

There are two findings to note in particular on this chart.  First, you can clearly see the strength of the impact of those negative reviews.  It is the largest bar on the chart, even greater than raising price. Further, the relative positive impact of ratings, rank and brand is small as compared to those negative reviews.   Note that the TripAdvisor rankings have a smaller impact than the ratings.  Secondly, when you break out the individual impacts of the levels of ratings and rankings on choice, you will notice that consumers only notice a difference when comparing hotels with mid-range to those with high-range values.  They do not value a mid-range as compared with a low-range value.  This finding adds a nuance to the recent study from the Cornell Center for Hospitality Research, which found an 11.2% increase in pricing power for each point increase in a ratings metric.  Our study suggests that hotels will only see this benefit if they raise their ratings from a mid-level score to a high score.  There will be no benefit from a lower score to mid-level score movement.

Impact of negative reviews

Choice modeling allows for a calculation of the overall value consumers place on a combination of attributes, which means we can evaluate the relative impacts of changing attribute levels on the whole picture of the consumers’ likelihood to choose.  Once again, the actual value of the number is less important than the values relative to each other.

Not surprisingly the combination of attributes that maximize a consumer’s likelihood to choose is positive reviews, low price, high TA rank, high rating and known brand.  This results in a baseline overall utility of 1.95.

Positive + $195 + High Rank + 4.8 + Known Brand = 1.95

Notice the drop in overall utility when you change only the price to the highest price level.

Positive + $295 + High Rank + 4.8 + Known Brand = 0.46

Raising the price has a relatively large impact on overall utility, even when everything else is held equal. Clearly consumers prefer to pay the lowest price they can.  Now observe what happens when you hold all of the values equal, but change the reviews from positive to negative.  The utility value drops to practically zero.

 Negative + $195 + High Rank + 4.8 + Known Brand = 0.01

Even the positive impact of a lower price does not outweigh the negative impact of the negative reviews.

Key Takeaways

This study was designed to evaluate how consumers make tradeoffs between price and non-price attributes of a hotel when making a purchase decision.  There are four major takeaways from this study for managers:

  1. Reviews and price are the most important influencers of choice. While consumers did pay attention to aggregate ratings, TripAdvisor rank and to a lesser extent, brand, positive reviews contributed the most to consumer choice behavior followed by lower price.
  2. Negative reviews remove you from the choice set. Period.  Lower price or higher ratings do not overcome the impact of negative reviews.  Consumers simply will not choose a hotel with negative reviews.  Hotels that are in this unfortunate situation should focus energies on improving their reputation.
  3. Consumers prefer to pay a lower price. While consumers would go for a higher-priced hotel when the reviews and ratings were better than the alternatives, all things being equal, they will look for the lowest price.  Hotels need to understand their position relative to their competition both on reputation and on price in order to take advantage of any pricing power associated with positive UGC.
  4. Consumers only notice high ratings and rankings.  Our results showed that consumers only notice ratings and rankings when they are high as compared to other choices.  Consumers do not place any value on the comparison between low and mid-level ratings and rankings.

The bottom line is that driving revenue and share in the hospitality industry is no longer just about competing on price.  Consumers are clearly turning to user-generated content to inform their purchase decisions, in particular, reviews. This means that hoteliers must not only keep an eye on how they are priced relative to the market, but also on how they are positioned in terms of their reputation.

tags: Hospitality Analytics, Revenue Management and Price Optimization, Social Media

11 Comments

  1. Posted October 25, 2013 at 12:14 pm | Permalink

    Great research and a wonderful article Kelly. It may be a bit technical for some hoteliers, but your Key Takeaways make it quite clear what higher end hotels need to do to drive more revenue. At the very least they have to be in the consideration set and negative reviews will keep them from getting there. Thanks!

    • Kelly McGuire Kelly McGuire
      Posted October 29, 2013 at 10:23 am | Permalink

      Glad you liked the blog, Madigan. I hope that hoteliers can find some use in the study results. The pricing decision for hoteliers certainly isn't getting any easier! Thanks!

  2. Posted October 30, 2013 at 11:00 am | Permalink

    Congratulations on this great research! I think it is comparably harder to produce valuable first hand data for criteria like the impact on reviews and their influence on consumer behavior since it is not purely numeric. And I would say it is even harder to communicate the findings in an understandable manner. But as Madigan puts it, the Key Takeaways make it really easy to understand what this research is about and what hoteliers should care about!

    • Kelly McGuire Kelly McGuire
      Posted October 31, 2013 at 8:47 am | Permalink

      Thanks, Willem. Our goal in conducting this research was to contribute both to academic theory as well as industry practice. It is really important to us that the results are understandable and actionable. Glad to hear that we achieved that!

  3. Posted October 30, 2013 at 6:58 pm | Permalink

    Great article kelly, it covers the research you spoke about at the ROC conf this year too. Just one question, would you agree that the research findings culd be limited or the degree of relevance may duffer on the region and source market. Was any such data also tracked as part if this study ?

    • Kelly McGuire Kelly McGuire
      Posted October 31, 2013 at 8:56 am | Permalink

      Hi Yogeesh,
      Very good question. Our participants in this research were a representative sample (distributed by age, gender, education, income) of the US population. Behavior did not seem to change by demographic group within the US (which was somewhat surprising to us). I agree that the effects may be different for groups outside of the US, however. This would probably depend on the degree of online review activity in that market, as well as online shopping behavior. In our study the majority of the participants use the internet the majority of the time to book rooms (they had to have booked a leisure trip themselves online in the last year to qualify for the study). If online booking isn't as popular in a particular market, effects might change also.
      Kelly

  4. Posted October 31, 2013 at 9:15 am | Permalink

    Would love to get copy of your study

  5. Posted December 4, 2013 at 4:09 pm | Permalink

    This is a very interesting study. Were participants able to see just one review, or were they presented with several reviews much like Trip Advisor and other sites do.

    I would guess you'd get very different results if you showed people only one negative review versus several positive reviews with one negative review mixed in.

    • Kelly McGuire Kelly McGuire
      Posted December 4, 2013 at 4:54 pm | Permalink

      Hi Jeff,
      Good question, and I suspect you are probably right about that effect. In this study, we used a set of five reviews that were all the same valence. We did that because we needed the consistency for the statstics on the individual attribute (i.e. how much did consumers value positive reviews over negative, and as compared to other attributes). In our first study, described here: http://blogs.sas.com/content/hospitality/2013/06/21/pricing-in-a-social-world/, we used a set of 10 reviews, 8 were positive with 2 negative mixed in and vice versa. We had more flexibility towards realism in that study, because we were looking at the influence of reviews in general, not the values associated with the specific level of the attribute (i.e. positive).
      Kelly

  6. Eduardo P. Erazo
    Posted July 9, 2014 at 12:54 pm | Permalink

    Kelly,

    Would it be possible to get a copy of your article or a citation so I can track it down? This seems like great research and I'd be very interested in reading about your methodology, specifically-

    Did you run any metrics for multicollinearity or autocorrelation, such as VIF?
    Also, did you try a stepwise inclusion regression to see changes in R^2 in order to determine best fit?

    Thanks and looking forward to hearing from you,
    Eduardo

    • Kelly McGuire Kelly McGuire
      Posted July 15, 2014 at 10:20 am | Permalink

      Hi Eduardo,

      Thanks for the comments! The research methodology we used was Discrete Choice Analysis, which tracks customer choice patterns and then predicts the value that they place on the different attributes tested. This is not a regression-based modeling methodology, so we did not need to use any of the methods you suggest. The research is published in The Journal of Revenue and Pricing Management. http://www.palgrave-journals.com/rpm/journal/v12/n5/full/rpm201313a.html

      Cheers!
      Kelly

10 Trackbacks

  1. [...] liberamente tratta Pricing in a Social World: How consumers use ratings, reviews and price when choosing a hotel, di Kelly [...]

  2. [...] and The Pennsylvania State University published a new research called “Pricing in a social world: The influence of non-price information on hotel choice.” They present some interesting findings, but the bottomline is that reviews trump price as [...]

  3. [...] do matter. A report by SAS noted that positive reviews (less so TripAdvisor-based rank and brand), followed by lower price [...]

  4. By Marketing Turístico para hoteles y destinos on October 31, 2013 at 2:04 am

    [...] el estudio publicado en el blog de SAS puedes extraer una serie de conclusiones interesantes, aquí las comento [...]

  5. [...] studio, condotto dall’azienda SAS e dalla Pennsylvania State University, non è che un approfondimento [...]

  6. [...] even a lower price won’t help your bottom line if you have negative online reviews.  (Click here to see the details of the how the study was [...]

  7. [...] For more details about this research and the findings, read Kelly's blog post, "Pricing in a Social World: How consumers use ratings, reviews and price when choosing a hotel." [...]

  8. [...] Pricing in a Social World: How consumers use ratings, reviews and price when choosing a hotel, discusses how consumers make trade-offs between price and other attributes of a hotel, such as ratings or online reputation, when making a purchase decision. [...]

  9. By Measuring Productivity & Success in an Automated Era on February 12, 2014 at 2:57 am

    [...] was particularly intrigued, therefore, by a point in the recent Noone and McGuire study which indicated that the score is not as important as the actual reviews and comments – [...]

  10. […] the ability to capture, comprehend and act on guest or patron feedback is critical. Previously we learned from Kelly that user generated content from social review sites influences the purchase process for hotel […]

Post a Comment

Your email is never published nor shared. Required fields are marked *

*
*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <p> <pre lang="" line="" escaped=""> <q cite=""> <strike> <strong>