The billboard effect: does an OTA listing bring direct bookings?
The billboard effect suggests OTA listings drive direct bookings, but data is limited. Learn what Cornell studies found and how to test this on your own hotel data.
Contents
Key takeaways
- The billboard effect counts only the reservations a hotel wins through its other channels while it is visible on a booking site, never the bookings made on that site itself.
- In Chris K. Anderson's Cornell Hospitality Report (Vol. 9 No. 16, October 2009), four JHM hotels spent 40 days listed on Expedia and 40 days not; reservations through their other channels were 7.5% to 26% higher while listed, with the independent hotel at 26%.
- The 2011 follow-up (Vol. 11 No. 8, April 2011) used comScore panel data on 1,720 reservations on IHG brand websites and estimated that each Expedia reservation was matched by three to nine brand-website reservations influenced by the listing.
- Both studies rest on US data from 2008 to 2010 and a single booking site, so they show a mechanism worth testing, not a percentage a hotel in Türkiye can budget.
- A hotel can test the idea itself by comparing its direct channels in periods when its listing is on and off, over weeks rather than days, with the measures written down beforehand.
The billboard effect is the idea that a booking-site listing sells a hotel beyond the bookings made on that site: travelers see the hotel there, then book direct. A 2009 Cornell study by Chris K. Anderson found reservations through four US hotels' other channels 7.5% to 26% higher while they were listed on Expedia; a 2011 follow-up measured the browsing that precedes such bookings.
Every argument about booking sites eventually reaches the same sentence: we have to be on it, it is our billboard. The sentence is said with confidence and almost never with a number attached to it. This article attaches the numbers that exist, explains what they were actually measured on, and gives you a way to test the idea on your own hotel's data.
What does the billboard effect mean?
It is the extra business a hotel gets through its other channels because travelers saw it on a booking site. The listing advertises the hotel even when the booking happens somewhere else.
The mechanism is simple. A booking site is a busy shelf. A traveler who has never heard of the property sees its name, photos, price and reviews there, and later types the hotel's name into a search engine, calls the front desk or opens the hotel's own website. The name comes from the roadside analogy: a billboard does not complete a sale, it makes a name familiar. A booking site differs in one respect, because it charges a commission on the sales it makes rather than a fee for the display.
The measurement excludes the site's own sales. Bookings made on the booking site are the site doing its job as a seller; the billboard effect is what happens outside it, in the hotel's direct channels and in other channels. The effect can show up in two places: more direct bookings, or the same number of direct bookings converting better because the guest already trusts the name.
It is not a discount, not a guarantee and not a substitute for a direct channel that works. It is a reason a listing may be worth more than its own booking line suggests.
Key pointThree terms used below
An OTA, or online travel agency, is a booking site that sells hotel rooms for a commission. A direct channel is the hotel's own website, phone, email or walk-in. The billboard effect is the reservations through other channels that exist because the hotel was visible on a booking site.
What did the Cornell studies actually measure?
Two reports by Chris K. Anderson at Cornell measured the idea twice: once by switching a hotel's listing on and off, and once by following what travelers did before they booked on a hotel brand's own website.
The first, "The Billboard Effect: Online Travel Agent Impact on Non-OTA Reservation Volume" (Cornell Hospitality Report Vol. 9 No. 16, October 2009), describes a pseudo-experiment with Expedia and JHM Hotels from early October to December 2008. Four hotels, one independent and three branded, were cycled on and off Expedia in blocks of seven to eleven days, so each spent 40 days displayed and 40 days not displayed.
Leaving out bookings made on Expedia itself, average daily reservations through the hotels' other channels were 7.5% to 26% higher while the hotel was listed: Branded Hotel 1 at 7.5%, Branded Hotel 2 at 9.1%, Branded Hotel 3 at 14.1% and the Independent Hotel at 26%. The author calls the independent hotel's data the cleanest. He also names his own interference: for Branded Hotels 1 and 2, seven other hotels of the same chain family within 18 miles stayed listed on Expedia throughout, so their figures are the most subject to interference. Branded Hotel 3, with three such hotels within 20 miles, he calls a little cleaner.
The second, "Search, OTAs, and Online Booking: An Expanded Analysis of the Billboard Effect" (Vol. 11 No. 8, April 2011), used comScore panel data on 1,720 reservations made on InterContinental Hotels Group brand websites in June, July and August of 2008, 2009 and 2010, plus 122 IHG reservations the same panel made through Expedia. Almost 75% of the people who booked on the brand website had visited an OTA first, almost 83% had run a search on Google, Yahoo or Bing, and 61.8% had visited Expedia. The report concluded that each Expedia reservation was matched by between three and nine brand-website reservations influenced by the OTA listing, and calls this conservative because telephone bookings were not counted. Both reports are available from the Center for Hospitality Research at Cornell University.
| Report | Data and sample | Headline result | Main limit |
|---|---|---|---|
| The Billboard Effect, Vol. 9 No. 16, Oct 2009 | Four JHM hotels cycled on and off Expedia, Oct to Dec 2008; 40 days listed, 40 not | Other-channel reservations 7.5% to 26% higher while listed | Four hotels, one site, US market, 2008 data |
| Search, OTAs, and Online Booking, Vol. 11 No. 8, Apr 2011 | comScore panel, 1,720 IHG brand-site reservations, 2008 to 2010 summers | Each Expedia booking matched by three to nine influenced brand-site bookings | Measures influence, not cause; no phone bookings |
Read together, the two reports describe the same path from two ends: the first counts the extra reservations, the second counts the browsing that precedes them.
Why do the results not travel automatically to your hotel?
Because the evidence comes from four US hotels and one chain's brand websites, using data from 2008 to 2010, and the size of the effect depends on things only your own numbers can show.
The data is old in browsing terms. Travelers in 2008 to 2010 reached a hotel through a desktop search and a booking site; mobile booking, metasearch and free booking links have changed the path since, and the studies cannot describe it. The sample is also small and the market is one: four hotels in one test, one booking site, one country. There is no equivalent published measurement for Türkiye behind this article, so any Turkish figure quoted at you is someone else's number wearing a local costume.
The effect is measured on volume, not on profit. Every extra direct booking sits next to the commission paid on the listing's own bookings, and the net result depends on the mix between the two. Property type changes the arithmetic too. A resort with a strong name and repeat guests, a city hotel selling to corporate accounts and a 20-room pension in a coastal town are seen by different travelers in different ways. It is plausible that an unknown independent property gains the most name recognition, and the 2009 independent hotel showed the largest figure, but that is one hotel, not a rule.
Two further cautions. The 2011 study measures influence, not cause: a traveler who visits an OTA and then books on the brand website may have intended to book direct all along. And the direction of the effect can reverse. A hotel that is visible on a booking site at a lower price than its own website teaches the traveler to book on the site, and the billboard sends the booking back to the billboard.
Watch outDo not turn 26% into a forecast
The figure comes from a 2009 Cornell Hospitality Report on four US hotels listed on one booking site, using 2008 data. Quoting it as an expected uplift for a hotel in Türkiye is a misuse of the study. Use it as a reason to run your own test, never as a budget line.
How can a hotel test the billboard effect on its own data?
Compare your direct channels in periods when your listing is visible and when it is not, over weeks rather than days, and write the rules down before you start.
Choose a change you control: closing availability for a defined set of dates, pausing the listing, or opening a new one. Record the exact dates and times, because the comparison depends on them. Fix the measures in advance: sessions on your own website, searches for your hotel's name, direct phone and email enquiries, direct bookings, and bookings through other channels, all with the booking site's own bookings excluded.
Pair the periods. Set on-periods against off-periods of similar length and against the same weeks a year earlier; day of week, school holidays and local events move direct volume far more than a listing does. Respect the length: the 2009 test used blocks of seven to eleven days and 40 days on each side per hotel. A two-day pause measures the weather, not the billboard. Write down what else changed inside each period, a campaign, a rate change, a festival, a road closure, so the result can be read honestly later.
Set the threshold before you look at the data: what change in direct volume would make you keep the listing, and what would make you reconsider it. Finally, measure total contribution, not only direct volume. Dropping a listing also removes the bookings it would have made, so the honest comparison is the whole channel mix over the period. If your online bookings are already flat across channels, start with a channel-by-channel diagnosis before running any on-and-off test.
Checklist
- Exact dates and times the listing changes
- The direct measures to compare, with the booking site's own bookings excluded
- The paired off-period and the same weeks last year
- Campaigns, rate changes and events inside each period
- Minimum length: weeks, not days
- The change that would make you keep or reconsider the listing
What should the billboard effect change in your channel mix?
It makes a listing a marketing decision as well as a sales one, so judge it on the contribution of the whole mix rather than on the commission line alone.
Compare channels by what the hotel keeps, not by the price shown to the guest. A listing that costs commission but sends bookable travelers to your own site can be cheaper overall than a listing that only sells on its own account. The effect is also an argument for a complete, accurate listing: current photos, correct rates, answered reviews. A weak listing advertises the hotel badly, and the traveler who arrives by name search meets the same weak impression.
Keep the direct price at least as good as the listing price. If the traveler's name search lands on a worse price, the billboard sends the booking back to the booking site and the effect disappears into commission. Watch concentration as well as cost: when a large share of room nights arrives through one booking site, the question is not only dependence but whether your own channels can convert the attention that listing creates.
Fix the direct channel before you credit or blame the listing. A website that is slow, unclear on availability or hard to book on wastes the travelers the listing sends. Then rebalance in steps and measure each step, because changing several channels in the same month makes it impossible to say which change produced the result.
TipOne line for the next season
Judge each listing on the total contribution of the channel mix over the season, including the direct bookings it plausibly influenced, not on the commission line alone.
What to do next
Pick one listing, write down the dates it will be on and off, fix the direct measures you will compare, and run the test over weeks. Then judge the listing on the total contribution of the channel mix, and rebalance only after the numbers are in front of you.
Frequently asked questions
Does the billboard effect apply to small hotels and pensions?
The 2009 Cornell report's independent hotel showed the largest figure, 26%, and the author called its data the cleanest of the four. That is one hotel in one US market, so it supports the idea that an unknown property may gain the most name recognition, but it does not prove it for a pension in Türkiye. The only honest answer for a specific property is its own on-and-off comparison.
How long does a billboard effect take to appear?
The published work gives no timing rule. The 2009 test used blocks of seven to eleven days and 40 days on each side per hotel, which suggests weeks, not days. Judge a listing over a full season, and never on a single weekend, because day of week and events dominate short windows.
Should I list on more booking sites because of the billboard effect?
No. Each listing adds commission, rate and availability maintenance, and another place where prices can fall out of line. The effect is a reason to keep a listing complete and consistent, not a reason to multiply listings. Change one channel at a time so you can still read the result.
Can I put the billboard effect in my budget?
Not as a percentage. The published figures come from four US hotels and one chain's brand websites, on 2008 to 2010 data, and were never a forecast for any other property. Budget on bookings you can observe, and treat the billboard effect as a reason to run a test, not as income.
Does the billboard effect still hold with metasearch and AI assistants in the path?
Neither study measured metasearch or AI assistants, so there is no measured answer in the sources behind this article. The underlying mechanism, a traveler seeing the hotel somewhere and booking somewhere else, has not changed in principle. Whether it still produces the same size of effect is exactly what a hotel's own on-and-off test is for.