Taboola's EngageRank video recommendation technology has been officially adopted by the NYTimes.com, following a successful 8-month trial. NYTimes.com has implemented Taboola's recommendations in a section called "Other Videos You May Like" as thumbnails below the main player window and when the video ends. Taboola's CEO and founder Adam Singolda told me that based on A/B testing vs. other recommendations technologies, Taboola was found to drive 250% higher video views. Last week I met up with Adam and Lior Golan, who runs product and technology at the company and was in from Israel where he's based.
EngageRank delivers recommendations using algorithms combining both behavioral and contextual data about users' engagement with videos they watch. Specific inputs include the current video viewed, session length, geographic location and prior behavior. The system also learns from which recommended videos users clicked on and how long they engaged with them, helping improve future recommendations. As Lior explained, the key is that EngageRank constantly reviews the inputs, creating a feedback loop so that each user's behavior helps improve the subsequent recommendations.
Delivering more video views is of course critical to the business model success of Taboola's customers. As I've written before about Taboola, the company's willingness to be paid strictly on a success basis is a big differentiator. Taboola either takes 20% of the inventory of recommended videos, which it uses to insert ads, or it takes a revenue share for customers who want to retain full advertising control.
In either case though, Taboola's compensation is based only on the incremental viewership its recommendations have driven against a baseline alternative recommendation system of the customer's choosing. Taboola mandates its customers allocate 1-5% of their videos to use the alternate recommendations system both to create the baseline, and also to provide insight into how well any possible competitor's technology may be performing.
Taboola can also make video recommendations based on text pages with its Text2Video product, and can distribute recommended video to 3rd parties with its Affiliate2Video product. By maintaining its superior recommendations vs. others, Taboola is in a strong position to help distribute, drive viewership and increase monetization for any site using either its own video or video from 3rd parties. Since video is becoming table stakes for all web sites, that's an increasingly valuable service to be able to provide. Taboola's customers include Bloomberg, CNN, Demand Media, Revision3, Kiplingers and others.
What do you think? Post a comment now (no sign-in required).