Facebook introduced “Graph Search” at today’s press event. The idea isn’t new. Google has constantly been trying to display contextual search results with products like Google Now and Google Plus (Socially) integrated results. On top of that, users now demand exact results : things like Wolfram Alpha. Simply put, Graph Search is a search engine that searches “you and your connections” for the exact answers you want. The idea at the heart of this Facebook product : What ‘Apple’ may mean to an urban geek may not necessarily resemble the understanding of a farmer. How should the context be evaluated? This is where Facebook, with its massive user-base and activity data, comes in. The relevance of any “Graph Search Result” is cross-checked with your data – your likes, friends and their friends, location visits, tags and every public activity that Facebook can track. The conventional page-link search has outlived itself and probably, what Facebook calls, “Graph Search” is the future of discovering new data. To get a better idea, lets have a look at a few queries and their possible results in this new search engine –
- You’ve gone to San Francisco and you “checked in” to some place on Facebook (A way of notifying Facebook about your location). You search for “Restaurant’s nearby.” Pages of restaurants in SF near your current geographic location will appear and their ranking will purely be based upon your and your friend’s likes and visits. This could give Yelp a run for its money.
- You search for “Games friends from my school play” or “Music people in my city listen to” or “Friends who play table-tennis and work at my company.” Based solely on the information made “public”, your results will be delivered. Results will contain pages, profiles or photos.
- The Photo search is interesting. Text is manageable but how can you sort images? Graph search’s important use could be finding images by applying different parameters which make sense in social contexts. For example, this query : Photos of my marriage that my friends liked, shared and commented on. You cannot get them without actually checking every uploaded picture and tabulating the number of likes and shares. With one search, you have the best pictures (assuming that you consider your friend’s choices to be good).
- For queries like “The answer to life, universe and everything”, which I am not sure your friends would be aware of, Facebook will pull up results from Bing. (Incase you don’t know, the answer is 42.) It is a strategic partnership that will help Bing pull up its market share and at the same time, provide results to users, right on Facebook.
Search is a biiiiiiig market. Apple is chasing it with Siri, Google is exploiting whatever it can, Yahoo almost missed it and now, Facebook enters it (with the almost dead Bing) to model an engine completely around the user. Coming to the monetization part, Zuckerberg says “This could potentially be a business over time.” Hell yes, with businesses dying to take the top spots in such search results, it is a lot of money. The reason – results are ultrapersonal and carry with them, a recommendation from connections which is worth more than any creative advertisement. The conversion rates shall definitely be higher than those of the existing search results. People have already started asking stuff like “How can I improve my brands visibility in these results.” We have already had enough of SEO (Search Engine Optimization) and we might soon have GSE (Graph Search Optimization) techniques coming up. I’m sure, engineering a product like this would have been huge challenge. Understanding and processing the requests made in colloquial language is a problem. Scaling down the likes, comments, shares, visits, activity, etc to simple numbers to allow ranking is another tricky task. Right now, the product is in Beta and open to limited testing. You can join the waiting list to try out your luck.