TripAdvisor beta tests Just for You personalized hotel recommendationsNewsBy Sean O'Neil | May 1, 2014Share This article was originally published on It's no secret that user-review giant TripAdvisor has been beta-testing a "Just for You" service to customize its hotel search results with personalization input from the user.But the details have been sketchy.In November at the PhoCusWright conference, CEO Steve Kaufer discussed it in a presentation.In February, he repeated the highlights in a quarterly earnings call: In 2014, we plan to develop a more customized and personalized experience for every user on this site. We started beta testing our Just For You personalization initiative in Q4 and believe we are just scratching the surface in terms of providing a more targeted travel planning experience as we seek to delight our users with a personal guide when they are on their trip.Share this quote In additional comments to CNBC, Kaufer said: It's about finding the best particular recommendation for your particular trip. Right now, when you come to TripAdvisor, and you look at a page, you're seeing great recommendations. But it's still a little bit one-size-fits-all. Why not take the intense amount of data that we have about travelers all around the world and really help deliver a recommendation just for you as an individual, and a different one if you come back a year later and we know a little bit more about you.Share this quote Tnooz has since dug up a bit of color on the project.A spokesperson says the company expects "just for you" to roll out later this year. The idea behind the feature is basically to provide more contextual and personalized hotel recommendations, based on what the traveler has liked in the past and what type of trips they have told us about on TripAdvisor.Share this quote The feature is currently being tested with a small segment of the community. When this Tnooz reporter logs in as a member of tripadvisor.co.uk, he has the beta test view.Users in the beta test see a "Just for you" box as one of their ways of filtering hotel search results, alongside generic filtering categories such as "business" and "romantic."Clicking the "just for you" button produces a new set of results, not sorted by lowest price or highest rating but instead by some other unspecified algorithm.The list is introduced with the phrase "Similar to hotels you've viewed" and then, in each property box, "Similar to hotels you've researched".Those statements make it sound as if "just for you" picks are merely similar hotels to the ones a user has viewed on past visits to the site -- which would not be an impressive example of "personalization."A spokesperson responds: Because we're still in beta, I really can't expand on the detail at this stage. But it is more sophisticated than what you describe.Share this quoteTrying the beta-test version, I did a search for hotels in Brussels, and clicked the "Just for you" recommendations. The results put Sofitel properties at the top of the list.Coincidentally, I had viewed several Sofitel hotel pages last May when reporting the story of an Accor executive posting anonymous TripAdvisor reviews of the properties.I've barely used TripAdvisor since (getting recommendations from elsewhere). So the fact that I had looked at Sofitel pages 10 months ago may explain why I am now seeing Sofitel as "just for you" recommendations. I've never stayed at a Sofitel, so it couldn't be based on any transactional data.I did another search of Pittsburgh hotels. None of the top five "just for you recommendations" included the Wyndham brand (though there is a Wyndham in town). That brand was the only US chain I've bothered to write a positive review about ever on TripAdvisor.In that Pittsburgh case, there appears to be no direct connection between my having written a positive review of a brand and the likelihood a sister property of that brand might show up at the top of rankings when doing a search.On a search for Vienna hotels, the "just for you" button produced a list with the number one choice K+K Hotel Maria Theresa (replacing the generic (non-personalised) number one Hotel Sacher. The number two personalized choice is the Hotel Rathaus Wein und Design. The third was another darned Sofitel.The inclusion of Hotel Rathaus Wein und Design is telling. I have years ago looked at the TripAdvisor page for that property. I did not book it through TripAdvisor's pay-per-commission links, though -- booking directly with the property's own website instead.Being recommended a hotel I've already stayed at twice isn't very personalized -- but TripAdvisor doesn't know I've stayed there twice, hence the flaw in the personalization.One side note: Hotel Wein und Design has topped TripAdvisor's rankings as the most highly rated by users, which might instead explain why it topped the "just for you" recommendations, too.Sample size of oneThe service is still in beta-testing. But the personalization feature does not, in my anecdotal tests, appear to be anything sophisticated.Work to be doneIt's unclear how soon "just for you" recommendations might be pushed by e-mail to users, rather than waiting for users to perform a search. Either way, the recommendations will have to be good if people are to be persuaded to book rooms.A recent TripAdvisor recommendation suggested hotels in Kiev, but given geopolitical circumstances, that may not be the best idea right now.TripAdvisor does have the data to do betterAt a panel talk on personalization in London this week, Nathan Clapton, VP of mobile partnerships, noted that TripAdvisor is broadly committed to adding personalization features to its service.Some of the user data it has collected has been from its legacy of its Cities I’ve Visited Facebook app.The idea of bragging about where you've visited has been imported into TripAdvisor's site itself: Users are regularly invited to rate places they've visited. As a user, you can see what data you supplied by clicking on Member Profile and then selecting "contributions."Clapton pointed out that this data on places visited and liked could be useful in honing the site's recommendations -- eventually.