Recommendations in Travel: Getting the Data!

In the travel industry personalized recommendations is arguably the most sought after feature. But also one of the most difficult to properly implement. If we knew the types of things you liked to do and places you liked to go we could show you the best deals to the places we know you love. Instead we show everyone about the same deals to the most popular places.

But you aren’t everyone. You don’t like tourist traps. You like adventures!

Well, online travel agencies (OTA) don’t know that. OTAs treat everyone the same because they can’t get enough data from you before you leave and go somewhere else to search for deals. Trust me. If they knew what you liked they would give you the best deals imaginable to the places they think you want to go. Then get you to pay them to get you there.

That’s the crux of the problem. Not enough data. OTAs can’t tell who you are so they can’t show you deals for places you’ll like, they just guess blindly.

Companies like Google, Facebook, Linkedin, Pinterest, and other sites you spend intimate time with have all this data. They know everything about you. And from this information they can map your personality almost down to a T to show you the most relevant ads and make you most likely to buy something.

Sites with sparse data like Amazon try something called collaborative filtering. They look at their millions of users and find other users with similar search categories and purchases as you among other attributes. You show similar viewing and purchasing patterns as these other couple thousand users. This allows Amazon to be reasonably certain that you might be interested in these items as well.

Although Amazon doesn’t have as much data on you as many other sites you frequent. It does have a little data from a lot of other users and it leverages that to try and identify what you are most likely to buy.

Travel companies have this data too, but have largely been slow to implement it in any meaningful way. A typical traveler will not impulse book but will rather look at many, many other sites comparing deals to shortlist the best ones for them and then book.

Business travelers will just come on the site. Find the flight they want and book.

Not too many interactions to find out about the people using your website to book.

So. How can you get the data you need?

There are many ways to overcome this problem. Lots of silver bullet solutions come to mind like Chrome Extensions, cool new apps that ask for your preferences up front, a Travel version of Pinterest to see what users like, but often the best solutions are the least sexy ones.

Just ask your customer. Nearly all travel companies have long load times on their websites as they “find you the best deals.” Just plunk a couple simple questions. Which of these do you like beaches, backpacking, food, etc? People are waiting for up to 30 seconds in most cases.

That’s 30 seconds of data the OTAs could be collecting to find out what kind of person this customer is and help her find great deals later on!

That data could be used to filter the most relevant hotels for the user on the front end before the user even sees the results.

Just a relatively small tweak that doesn’t even need to be implemented on the backend could help OTAs increase their conversion and help their customers find exactly the hotel and trip they are looking for.

Below are a couple of my quick mockups on how the data could be collected during the wait times.

Getting customer preferences on Expedia during load screen
Getting customer preferences on Priceline during load screen

Product Manager and designer writing about ideas. Living and working in SF. See more of my projects at www.alexcreates.me

Product Manager and designer writing about ideas. Living and working in SF. See more of my projects at www.alexcreates.me