The Billboard Effect is dead, says a study of hotels listed on OTAs
A new study on how US travelers book hotels online contradicts earlier research on the topic. It suggests that a much-touted “billboard effect” for hotels listed on online travel agency (OTA) websites may no longer exist.
The study of about 50,000 online travel shoppers in 2014 found that users who browsed hotel sites tended to ultimately purchase their hotel stays via intermediaries, such as Expedia, Booking.com, and TripAdvisor, on average.
The behavior was not true in reverse. Internet users who began their browsing on OTAs and similar intermediary sites tended to purchase hotel stays through those intermediaries, too.
The study was teased during a panel talk at the Revenue Strategy Summit held on Wednesday in Washington, D.C.
Different study, different sponsors, different results
The data was analyzed by P. K. Kannan, chair of the department of marketing at the Robert H. Smith School of Business at the University of Maryland.
Changed consumer behavior?
The study’s results suggest a shift in consumer behavior since 2009, when Chris Anderson, an associate professor at Cornell University, published a study about Expedia and hotels in the JHM Hotels.
His study found that the appearance of a hotel chain on the first page of results on Expedia coincided with a rise in reservations through JMH Hotels’ own websites. The direct booking uplifted ranged between 7.5% and 26% uplift.
His follow-up April 2011 study of 1,720 reservations at InterContinental Hotel brands found a similar behavior. For every commissionable hotel reservation that came through Expedia, there were between three and nine reservations at an IHG-related website that were influenced by the hotel having been listed on Expedia, on average.
The studies were touted by Expedia and others as proof of a so-called “billboard effect”, where OTAs pull double duty as a kind of search engine marketing tool for hotels’ own websites.
Bye-bye, billboard effect?
Unlike the Cornell work, which used data from Expedia and hotel chains directly, this study used clickstream data from a market research firm and it was analyzed via attribution modeling.
Yet like the Cornell research, this study did not include mobile device users.
Kannan declined to share details of the study with Tnooz prior to the report’s publication this winter.
What follows is a transcript of what Kannan told the attendees of the conference. Note: When he says “hotel sites” he means “well known US hotel chain sites”, such as Marriott.com or BestWestern.com.”
“Here’s a quick snapshot of what we found in a preliminary analysis of data.
We are looking at how people are searching online in their search and booking process. What we have is data about their clickstream behavior from a syndicated market research firm.
We looked at two different years, 2012 and 2014, to see how behavior is changing.
We did a path analysis. We track how long users visit sites, how many pages they visit at each site, and whether they complete a transaction via that site, such as an intermediary like Expedia. If they booked a hotel, we know what brand they booked.
The market research firm gives us session data as well as the transaction data.
Once we know there was a conversion, we merge that particular point with the transaction data, which gives us a lot more information about what they did, such as which brand hotel they booked a stay with and if they also booked a car or a flight.
We have done an individual-level analysis of visitors to find out how loyal they are to various domains.
We track each individual user and see whether, if they visit a hotel site, what are the chances they’ll book at the hotel site, and if they visit a hotel site, what are the chances they’d book at an intermediary site.
We’re able to map how loyal consumers are to particular domains. We’ve created three broad categories: Those who exclusively visit hotel sites, those who exclusively visit intermediary sites, and those who visit both domains.
Between 2012 and 2014, the number of people who exclusively visit hotel sites has come down from 12% to 10%. Whereas people who visit exclusively intermediary sites have gone up from 48% to 60%. People who go to both types of sites have dropped from 40% to 30%.
We then drilled down on the data about people who visit both domains and checked to see what kind of behavior they exhibited.
We tried to figure out how many booked at hotel chains versus those who booked at intermediary sites. What we find is that, between 2012 and 2014, the number has increased for hotel chains from 7.9% to 10.8%. The number for intermediaries has dropped from 7.7% to 6%.
There is not an appreciable change in the percentage of people who browse at both types of domains that then book at both types of domains.
This is not good news for hotels, despite first impressions.
Note that we found that the segment of shoppers who go to both types of sites has dropped from 40% to 30% in the past two years.
People who in 2012 were willing to visit both domains are in 2014 opting disproportionately to use intermediaries exclusively.
We also looked for search behavior. How many visits happen at intermediary sites and how many visits happen at hotel sites? 82% end up at an intermediary shop for browsing, while hotel sites get 18%.
Here is another look at the data. What happens between two consecutive visits? If you went to a hotel site, where do you go later? Do you go back or go somewhere else?
Some people repeat visit hotel sites. That demographic has increased from 57.4% to 62.9% over two years.
The people who leave hotel sites and go to intermediary sites later was 42.6% and dropped 37.1% (the complements of the previous numbers).
At first that seems like good news for hotels.
But the repeat visits for intermediary sites is fairly high: 90%. If you go to an intermediary site, there’s only a 10% chance your next visit will be a hotel site.
We tried to look at each individual’s string of path data and relate it against it against conversions.
The metric we used for this is called Generalized Transfer Entropy (GTE). What this technique does is let you, at the individual level and in a statistically significant way, indicate whether a user is more likely to go to point B if he or she went to point A.
In 2012, users who browsed hotel sites had a significant directional tendency of transacting at those hotel chain sites.
In 2014, this changed. Users who went to a hotel site in 2014 were more likely to actually convert at intermediary sites.
Meanwhile, intermediaries retained their popularity.
In 2012, users who went to intermediaries tended to book at intermediaries and were very unlikely to book at a hotel site instead. This held true in 2014….
To say that differently, in 2012, people who browse to hotel sites have a significant directional impact to actually transact at the hotel sites.
If you went to an intermediary site, you were very unlikely to go to hotel site to transact. It’s a negative impact on you transacting elsewhere. If you book some other travel product at an intermediary, such as a rental car, it has a negative impact on you booking a hotel at a hotel site.
In 2014, if you went to a hotel site, you’re also more likely to go book at an intermediary.
We have a lot more data to go through, such as at the chain-level and intermediary-by-intermediary.”
The data is expected to be released sometime between late 2015 and February 2016, said Cindy Estis Green, CEO at hospitality analytics consultancy Kalibri Labs and a co-organizer of the conference. Results from the study will be incorporated into the consultancy’s annual Distribution Channel Analysis report.
THE CORNELL REPORT: Does the Expedia billboard effect still exist for hotels?
NB: Image courtesy seanoneill/flickr
Sean O’Neill is Editor-in-Chief of Tnooz.
Before joining us, Sean was the future of travel columnist at BBC Travel, senior editor of BudgetTravel.com, and an associate editor at Kiplinger’s. He now lives in New Jersey, after a four-year stint in London. Follow him on Twitter.