What Graph Search needs to do
Facebook’s Graph Search is a revolutionary new form of searching and finding information about ourselves, friends and the world.
The problem is it’s too much work.
Many have come away from trying Facebook’s new take on search unimpressed. They wonder how they’ll ever use it in their daily life. They don’t get satisfactory results when they do try a query.
This is largely because Graph Search puts the onus on users to ask clever questions if they want to get better answers. To a degree this is true with any search engine: using certain keywords and operators will help narrow down the results to be more relevant. But for most queries, users can find what they’re looking for even without advanced search features. With Graph Search, the value of the product is hardly apparent until users add more qualifiers, like the specific audience they want to search among or content type they’re looking for.
Google doesn’t require people to search for “highest ranked website about the NFL” or “article about the sequester by popular news sites.” Someone can type “NFL” or “sequester” and likely find what they want.
Now, Graph Search isn’t positioned as a replacement for Google. As CEO Mark Zuckerberg said at the product launch, web search is about asking open-ended queries to return links that might have the answer to a question you might have. Graph Search is about precise queries and exact matches.
But in effect, this means users must come up with their own algorithms for what they’re looking for, whether it’s “movies my friends like,” “movies my friends of friends like,” “movies liked by people who like movies I like” or “movies liked by NYU Film students.” That can be difficult for people who don’t naturally think in precise queries.
More commonly, people have questions like, “where should I take my girlfriend to dinner?”, “What TV series should I watch next?” or “Who would go with me to this concert?” These are often the very questions we turn to our friends for already, but Graph Search can’t yet answer them if users type them in this way.
Of course, Facebook has the data to offer relevant suggestions in each of these instances, but it leaves it up to users to decide how to find it. Users could search “restaurants nearby liked by people in a relationship,” “TV shows liked by people who like shows I like” or “friends who like music I like and live nearby.” However, most users might not be able to come up with these ideas on their own, or even if they did, they might have trouble wording their query in the right way for Facebook’s system to understand it.
People would find more value in Graph Search if they could pose their general questions and have Facebook return a range of results. For instance, if users search for books to read, Facebook could show a section of books their friends like, a section of books that users with similar tastes like, or books that are popular among users in the same demographic or profession. From there, users could dive into the sections they’re most interested in. Or Facebook could give users more suggested queries to try after they’ve conducted a search. So if users search “books my friends like,” Facebook might say, “Also see what your friends of friends have read” or “See what other people like you like.”
Much of the traditional search experience is about visiting sites that lead to other sites and uncovering things in the process, especially when someone is researching products or vacation plans and looking for recommendations. Facebook’s approach of offering an exact match or no results at all doesn’t lead users down this path of discovery. The challenge for the company is to either change how people search or make its product more familiar and intuitive.
With more user education, better natural language processing and a rethinking of how to display results, Facebook could eventually make Graph Search a more functional and commonly used tool for recommendations. Until then, it’s a powerful product inhibited by users’ own understanding and creativity.
Facebook shares its ‘favorite ways’ people use Graph Search
How Graph Search can help users see the world through different lenses
Graph Search: a recommendation engine only Facebook could power