Get Personal!
Sanjay Vakil – TripAdvisor
Presentation (PDF)
Instant Personalization: Homepage
Cities from around the world friends have been to
Reviews that friends have written
Other pages display relevant content
10m/mo (US)
Decrease bounce rate
Increase page views
Increase in commerce
– <1% opt-out
– Few complaints in forums
Initial Thoughts
Excitement over being IP partner
40m reviews
Engaging interface
Property reviews
How many reviews can we actually show?
– Privacy Policy
– Are your friends members of TripAdvisor?
– How many of your friends have written reviews?
Equation #1
Your Users(Content)/Internet is approx 0
Hard UGC Numbers
Of 100 users, 10 will become members, 1 will generate content
Facebook users have about 200 friends
Chances are 1 or 2 have actually written a review
Personalization Pyramid
Property/Reviews -> Geo Reviews -> Reviews -> ? -> Clickstream
Equation #2
Engagement is proportional to the Content x Relevance
As you move further down the pyramid the value of the content decreases
What next?
Get More Content
Stuff that is relevant to what people want to see
Cities I’ve Visited – Facebook app built in 2007
25M users
60 pins per user
1.4B pins
+ “profile” pins(see friends on the map)
Better CIV numbers
About 5% of FB profiles
Show frieds that have been to cities, good conversation starter with limited content
Equation #3
Weak content x your friends = good content
Lessons
Don’t optimize for the top of the pyramid
Figure out how to stretch your data
Find an ethical source of additional data
Questions
What does TA show when no data is available?
Never have dead space, use an ad to share content
Broaden content up the chain but not too far
The invite only program is it still invite only?
No idea, it is a FB program. Two roads, know someone or be a pretty serious property that they want to work with.
Connected Impressions 