When we visit sites like Amazon, Netflix, or YouTube, we take for granted the fact these services have been exploring and learning our behavior for years. They know us well - they make relevant recommendations that are unique to the user, drawing us back to their sites time and time again. The big players in travel (think Expedia Group, Booking.com, TripAdvisor, etc.) are no different. They have already been using personalization for years and achieving impressive results. Today, travelers are accustomed to online booking, on-demand virtual assistants, and constant access via devices and applications. Whether they’re booking hotels, looking for flights, or browsing unique experiences, they not only expect to be able to browse from their phones they also expect the website (or application) they’re using to provide options that fit the specific criteria they’re looking for.
Whether you’re traveling for pleasure, business, or a combo of the two, travel is becoming more intelligent, diversified and customized. Machine learning and personalization can give travelers on-demand services, customization of unique experiences, and insights (like predictive pricing or delayed forecasting) providing the customer with the unique, tailored experience they desire.
By using personalization methods companies are able to make more relevant recommendations to consumers, improving their experience, typically resulting in increased conversions. We’ve found one of the best ways to do this is by creating a personalized feed. To implement an intelligent feed that displays relevant flights, hotels, trips, or packages to viewers, we recommend these 5 steps.
1. Capture Explicit User Preferences
This is the basis of doing any kind of personalized feed. The service needs to ask the user for the specific criteria they’re looking for in their trip. Obvious criteria or filters include location, price, specific travel dates (or a flex search), activities, and/or experiences.. Most travel sites today have the ability to filter based on these options. Something to consider is presenting the user with a “deal breaker / must have” toggle. Having this toggle, or some kind of sliding scale on how strong of a preference it is, will be fantastic input for your personalization engine later on.
2. Track Implicit User Behavior
As your users interact with the platform, be sure to track their actions and record those analytics. Some behaviors will indicate stronger preferences than others. For instance, sharing a destination or activity with a spouse or friend is a much stronger indicator than viewing some pictures or looking at reviews and moving on.
3. Compare to Other Users
As the models of user preferences are developed and evolve, you will find users who are similar to each other. Once you’ve identified these similar users, you can leverage their actions, flights, destinations, hotel or activities they’re viewing to add options for their counterparts.
4. Rinse, Repeat
The personalization engine evolves over time and requires constant input as well as tweaking. As more and more data is fed into the system, new insights will be gained and profiles will morph. This feedback loop of both positively reinforcing and negatively detracting behaviors will help the model tune itself and become better at making recommendations for each user to view.
5. Bringing It All Together
Once steps 1-4 are in place, the travel platform will be well positioned to capture a market of users that expect to be able to self-serve. This platform will present the most relevant destinations, flights, activities or rental properties to each user, being sure not to show content that is irrelevant.
Now what?
Users of intelligent, highly personalized feeds are more likely to remain engaged, spread the word, and ultimately find their dream vacation. There are many travel companies out there, so providing your clients and prospects with the best user experience is crucial. To gain a broader understanding of the benefits of feed technology and how you can implement them contact us with your questions!
