Adam Singolda is CEO of Taboola, a video recommendation and advertising company.
AdExchanger.com: It seems more and more content recommendation companies are getting into ads. Why does this make sense?
AS: Advertisements that are suggested as content with high intent and context have long been the most productive and highest yielding online. Examples are Google Adwords (search ads), infomercials, and most recently, promoted tweets on Twitter .
As publishers already have navigation real estate on their content pages (such as ‘popular items’, ‘more from the same category’, etc), leveraging that space to display personalized items as well as sponsored recommendations makes sense for publishers and users alike. Publishers aren’t required to allocate additional real estate on their pages, and users can easily find the contextually relevant ads and engage with them within the context of the player.
What problem is Taboola solving?
As video inventory is becoming more abundant, navigating through video libraries is getting harder -which is a difficult technology problem to solve .
We find that on average, most viewers watch only one video per session when visiting Websites that offer videos.
Taboola solves this problem. Different users have different needs and in different times.
As viewers watch more videos, the playlists they are served with become much more relevant to them.
Likewise, publishers win because they get to serve more videos and yield more ad dollars.
Everybody wins.
Where do you get the datasets that are used to provide recommendations and ad targeting? Do videos need to be completely transcribed, for example, to understand what might be the best message?
Taboola’s dataset is a mix of semantic/contextual data and behavioral/engagement data.
Taboola’s behavioral dataset is generated by “watching” how viewers behave and interact while watching videos across our network and then analyzed into a rating score that we call EngageRank which determines whether the user liked the video or not.
Our contextual dataset is gathered by ingesting our video publishers’ RSS/MRSS feeds, as well as by automatically crawling their public video pages to obtain video meta-data and contextual data. We also support the leading Online Video Providers (OVPs) such as Brightcove, ThePlatform, Ooyala, Kaltura and others. There is no need to transcribe the videos.
What’s your target market in terms of potential clients?
Taboola has 3 products.
- Video2Video – a product for video publishers. Video2Video generates a set of recommendations or a playlist of relevant videos for every video in a publisher’s library.
- Text2Video – a product for video and content publishers – websites with a large amount of article content pages and a video library. Text2Video delivers a playlist of videos from the publisher’s library matched to each and every content/article page on the publisher Website. This product goes beyond text, leveraging Taboola’s behavioral knowledge of which types of videos would yield high EngageRank, engaging each viewer for the longest session.
- Affiliate2Video – To be fully rolled out this winter, this is a platform for both video publishers/producers as well as content Websites that don’t have a video library and would like to include videos across their Website. Leveraging the core Taboola contextual and behavioral technologies powering Text2Video, Affiliate2Video empowers video publishers to build their own Video Distribution Network (VDN) by providing them with a way to sign-up and deliver their video library, ads, and video player, “wrapped” by the Taboola technology, to 3rd party blogs and sites that don’t have videos. Taboola will automatically match every article on a 3rd party site with a suggested video playlist.
How about performance – what are you seeing?
For publishers using Taboola on their article pages (Text2Video), and video pages/player (Video2Video) we’re seeing 100%-300% uplift in videos per session. This is tested in constant A/B testing to help our customers quantify Taboola’s value over time. Taboola’s robust A/B system automatically establishes two or more testing buckets upon launch. One bucket (the baseline) serves recommendations using a publishers’ existing / internal systems, (such as “popular video from the same category”) or even a 3rd party recommendation engine vendor. The other bucket is Taboola’s. Our customers get a daily graph showing the uplift in video views between the buckets, which translates into quantifiable lift in CPM revenue for our customers.
Please discuss how pricing works.
We have two pricing models.
- Ad Supported – Customers can use our product’s offering without a license fee, and keep 100% of the uplift in pre-roll revenue they generate using Taboola. Taboola serves sponsored recommendations along side the video recommendation playlist presented on an article page or video page/player.
- Share from lift – Our customers share revenue from the pre-roll CPM revenue with us generated from the uplift Taboola yields in video views, calculated using the A/B testing dashboard
How many employees do you have today? How are you funded? Any plans here?
Taboola has 30+ employees with headquarters in Manhattan, and an R&D center in Tel-Aviv, Israel. We’ve raised $6M from Evergreen VC and private investors. As part of that fund raise, Mr.. Bob Cohn (former Sequoia Capital partner, and Founder & CEO at Octel) joined as Chairman of the board.
Any plans to expand into other digital channels such as mobile or email?
In early 2011 Taboola will launch an API which will allow customers as well as 3rd party technology vendors to develop Apps using Taboola’s Video2Video, Text2Video and VDN technologies.
A year from now, what milestones would you like Taboola to have accomplished?
To be the standard platform for any blog and site on the web, to offer video playlists and video recommendations to users across all platforms, and to become the glue between dispersed video libraries and content on the Web.
For users – our goal at Taboola is to redefine the way people discover videos they like.
Follow Taboola (@taboola) and AdExchanger.com (@adexchanger) on Twitter.
