Interesting article by Colin Nagy on AdWeek today: Panning for Gold With Search Engines: Smarter, social recommendations are long overdue. Colin argues, very successfully I think, that search engines are overloaded with content and not doing the best job of filtering for quality. As he puts it:
“With the proliferation of content over the past few years, we’ve created too many rivers and not enough dams”
Couldn’t agree more. That’s why part of what we’re trying to do at Outbrain is to algorithmically identify what content is interesting to different audience segments within different contexts. We have learned through trial and error that filtering for content quality is a very different task than filtering for relevancy. So for example, when we recommend content we’ll score articles not just on whether we think they are related to the article you just read, or whether they have lots of inbound links, but whether historically people who have clicked into it have read through it fully, spent time on the destination page and kept exploring more content after clicking, etc. After all, why would we want to recommend content to you if 90% of people who read it bounce?
So to continue Colin’s metaphor, our approach is to siphon as much water as we can from various content streams in order to learn quickly which sources are fresh and pure, and which are polluted. Then we quickly build a reservoir around the good stuff and pour it into the cups of readers around the web.