Facebook Incorporates Liked Web Content Into Search Results
A little over a month ago, we reported this story “Facebook In Early Stage Search Engine Tests?” and detailed some of the changes we were noticing. Well, today, based on another tip from bloggers at All Facebook, we can confirm that Facebook is indeed displaying ‘liked’ news content in it’s search result drop down. In fact, further to what All Facebook reported, they are also displaying content shared by your network of friends that matches the keywords you are sharing.
Like All Facebook, perhaps we should have suspected something was up, because yesterday a new type of update was displaying in the activity stream. Did you see it? I noticed that when i shared the stunning Arcade Fire HTML 5 video and hilarious UpCeption (An Up & Inception mashup video), as it got further shared beyond my network, a distinctly new social update displayed.
Rather than saying:
Friend X shared LINK with Friend Y
The new update says:
Friend X and Friend Y BOTH shared Link
It’s distinct because it highlights when both people share the same link. Whereas the first scenario tells the story of single interactions between friends, this new scenario tells the story of the interaction of single content plays among your network.
Sound familiar? It should be. If there was an activity stream for the Google Crawler calculating PageRank, arguably it would say something very similar. DomainX.com and DomainY.com both linked to SearchEngineWatch.com.
How is Facebook Displaying and Ranking Web Content?
These seem to be the factors that influence ranking so far. In descending order of importance.
Also, crucially there are currently two types of search engine results pages (SERPs):
The trend seems to be that the 3 ranking factors above, replicate within the post by friends and post by everyone section. Keywords definitely play a role in determining ‘what’ the content is. Ranking is determined by directness of the relationship to you. Where there is no direct relationship, ‘likes’ becomes the strongest signal. The formula reads something like this:
(Keyword relevance of shared content within your immediate friend network) divided by (ranked in order of who shared it) x (total number of likes)
For example, the screenshot below prioritises my active friends over less active ones, and in turn ranks that segment based on total ‘likes’. Both are active in the sense that we interact on Facebook, but one is in more regular contact than the other.
Also it is worth noting that it is not happening just on news content. It’s happenining on all web content. Furthermore, where the like button is not integrated on the destination page, in-activity stream likes are used to rank the web content. Vimeo does not have Facebook Like Buttons, but lots of people liked the link, so UpCeption, ranks for Inception in this instance, because it is most relevant to my network.
However, arguably, that is the wrong way around. Searching for stuff I already had seen defeats the purpose of web search as a content discovery mechanism – which must ultimately be Facebook’s goal. Ironically, they may have to downplay proximity to achieve the lofty goals of rivalling Google.
So, the big question remains as to whether users will start searching Facebook posts as a way of discovering new content. Try some searches for ‘funny’, ‘brilliant’ or ‘hilarious’ within the ‘posts by friends’ or ‘posts by everyone’ section to find out for yourself whether this could be a useful feature.