Facebook Is No Longer Flat: On Graph Search

A sample Graph Search result for the query “friends photos before 1999.”

By now the news is sweeping across the blogosophere and into the mainstream press: Facebook is doing Search!

Well, not so fast. Facebook is not doing search, at least not search Google-style. However, the world’s largest social network has radically re-engineered its native search experience, and the result is not only much, much better, it’s also changed my mind about the company’s long term future.

Yesterday, Tom Stocky, Facebook Product Management Director, and Lars Rasmussen, Engineering Director, gave me a sneak peek of today’s much anticipated announcement (it’s gonna be a phone! A new Newsfeed! A big acquisition!). So as to not bury the lead, Facebook has built what it’s calling “Graph Search,” a solidly conceived structured-search service which leverages the company’s massive trove of personal data in any number of new ways (some obvious, some nuanced, and some glaring omissions). But before I get to the details, I want to write about why this matters so much.

Prior to seeing the new search, I was not certain Facebook would ever live up to the hype it has accrued over its short life. It’s a good service, but it’s flat – over time, it struck me, people would tire of tending to it. They set up their social graph, toss a few sheep, poke some pals (or not), “like” this or that (often off-domain), waste hours on Farmville, and then…engagement drops slowly over time. I’m also not a fan of Facebook’s domain-specific approach to the world, as many of you know. Facebook’s new search doesn’t address Facebook’s walled garden mentality (yet), but it nails the first issue. Once this search product is rolled out to all of its members, Facebook will no longer be flat.

This is a big deal on many fronts. First and foremost, Facebook has an engagement problem, particularly in markets (like the US) where its use has become ubiquitous and many of its original users are two, three or more years into the “Facebook habit.” While the company doesn’t talk about this issue, I am confident it’s real (in private conversations with people at Facebook, it’s called the “set it up and forget it” problem). If people do not constantly feed Facebook with engagement, its value attenuates over time. As the service slows in overall growth, engagement with its current base becomes critical. New connections are the lifeblood of a service like Facebook. Without a steady stream of meaningful Likes, Friend Requests, declared Interests, and such, the platform would wither.

Put another way, Facebook needed a service that layered a fresh blanket of value over its core topography. Graph Search is it.

Zuckerberg’s Engagement

One sign of how important this new search is? According to the folks I spoke to yesterday, Facebook’s mercurial founder and CEO Mark Zuckerberg calls Search the “third pillar” of the company’s service, elevating it to the level of Newsfeed and Timeline, the two most important new features since Facebook’s launch (Open Graph is probably up there as well, but it’s true value remains locked up until there is mortar connecting it all, which Search could well be).

A team of engineers and product folk have been working on Graph Search for more than a year, and Zuckerberg has been engaged with them the entire time. The team has been in “lockdown” – a exclusive state of focus on one product so as to ship it as quickly as possible – for the past 34 days. Lockdown is a time honored and rather prestigious occurence inside Facebook, dating back to Zuckerberg’s original Facemash dorm room programming outburst. During the Search team lockdown, Stocky told me, he and Rasmussen got plenty of 2 AM emails and unexpected late night visits from the CEO.

In other words, this is A Really Big Deal for the company.

Why? Well, a quick tour of the product will explain.

What Is It? 

Graph Search subsumes Facebook’s previous search offering, which was extremely weak and focused mainly on the use case of navigation (finding people and pages).  The new service takes full advantage of the face that Facebook is, at its core, a massive structured database of tagged entities. The initial beta “indexes” four main types of these entities: People, Photos, Places, and Interests. Over time, I am told, Facebook will expand its index to include all Facebook posts and even the Open Graph – which means the “rest of the web.”

But for now, users can search across four main categories, using a slick set of intuitive verbs (“lives,” “like,” “work,” etc.), nouns (“San Francisco,” “Indian,” “restaurants,” “friends” etc.), prepositions (“before,” “with,” “in”) and pronouns (“who,” that,” etc.). This makes for a richly structured set of results: “Friends of friends who live in San Francisco and like Indian restaurants,” for example. Or “Friends who have been to Ireland,” or “Photos of friends before 1990.” Once you get the hang of it, the possible pivot points are endless, and the results are quite intriguing.

Stocky and Rasmussen, both ex-Googlers, walked me through a few intriguing use cases, one of which harkens back to one of Facebook’s original use cases – dating – and another which looks forward and presents a threat to LinkedIn’s current strength: Recruiting.

Let’s say you’re single, and you’re interested only in dating engineers who are also friends of your friends. With Graph Search, it’s ridiculously easy to find “friends of friends” who are also engineers. (And single, of course). You can look at their pictures, profiles, interests, and then ask for an introduction from whichever of your pals happens to be connected to one who looks like a good prospect (you could also just “poke” the guy if you wanted to…). Want only C++ programmers, or Indian C++ programmers, or  Indian C++ programmers under 35 years of age? Done.

Or, let’s say you work at, I dunno, Google. And you want to recruit product management talent from, say, Facebook. Again, the best way to get to that talent is probably a friend. So why not do a search for “friends of friends who work at Facebook and are product managers”? Why not, indeed.

One can imagine such functionality will create a lot of new engagement on the service. And not just from people “friending” prospective beaus or hires. Recall that when Google burst onto the scene, it prompted a dramatic response from owners of web pages, who immediately began rewiring their sites to be optimized for search. Similarly, Facebook’s Graph Search will incent Facebook users to “dress” themselves in better meta-data, so as to be properly represented in all those new structured results. People will start to update their profiles with more dates, photo tags, relationship statuses, and, and, and…you get the picture. No one wants to be left out of a consideration set, after all.

Facebook Gets More Weather

Last year I wrote a post titled “Facebook Is Now Making Its Own Weather.” The focus was on Facebook’s Newsfeed, and how an economy of value was now in place to game Facebook’s “edgerank” algorithm, which determines what stories show up in a person’s feed. With Graph Search, I expect a similar ecosystem will emerge. All of a sudden, two things will be true that previously were not: Facebook users will be using search, a lot, creating liquidity in Facebook “SERPS.” And secondly, there will be significant perceived value in being included in those search results, both for individuals (I want to be considered for that job at Google!) and for companies/brands (I want to message to anyone looking for a job!).

While Graph Search is in very early beta, I don’t think I’m going out on a limb by predicting that it won’t be long before Facebook integrates a product that lets marketers purchase ads in these new search results. It already has a similar product, which is by default included in suggested searches (the “auto completed” queries suggested to a user as they enter terms). At the moment, however, paid listings are not included in search results. They will be. Which means, of course, the rise of a native SEO/SEM ecosystem inside Facebook. Add in Open Graph search across the web, and presto…Google’s got some serious potential competition. (Well, not exactly presto. Incorporating Open Graph is going to take some serious chops and time. But still…).

Even without incorporation of Open Graph or Posts, Graph Search is going to change the game for brands and people on the Facebook service. As I watched Stocky and Rasmussen put their product through its paces, I couldn’t help but wonder how much new traffic the product will drive around the Facebook Platform. Will Facebook be watching “conversions” – clickthroughs from search results to profiles and pages? Of course they will! Will Facebook report those referrals to individuals and brands, much as Google Analytics does for webpage today? Not yet…but wait for it. It’ll come….

 What’s Missing: Sharing Results

I’ve already noted that Graph Search does not index content (posts) or the Open Graph, though I’m told that’s coming. But the big miss, from my point of view, is the inability to share search results.

Share search results? Who’d want to do that? Well, in web search, very few of us. That’s because with rare exception, open web search is not an inherently social action – it’s private and it’s ephemeral. But inside the walls of Facebook, it’s definitionally so. In fact, I’d argue that every single “result page” in Graph Search is a “media object” in its own right. If you search for “pictures of friends before 1990,” for example, you get the equivalent of a Pinterest board of your friends’ childhood shots. Wouldn’t you like to post that on your timeline so your pals can see it? Better yet, wouldn’t you like to export it to Pinterest or Tumblr? Of course you would (but, alas, I don’t expect Facebook will allow it, under cover of “protecting user privacy.” More on that in a second.)

Or take another example. Say you have a pal in Southern California who is despondent after being dumped by her boyfriend. You do a quick Graph Search for “single friends of friends under 30 who work in Los Angeles.” The results look pretty promising. Don’t you want to shoot them over to your pal with the subject line “Don’t despair, there are plenty of fish in the sea!” Of  course you do.

I mean, just a query like “Photos I Like” is a huge feature win for Facebook. And who wouldn’t want to post a montage of “Photos I Like” to their timeline? (Or, ahem, their personal blog?!)

For now, you can’t share the results of your searches with anyone else, and that’s a bug that should be a feature. When I brought up the issue, I was told that the privacy implications of sharing searches were extremely complicated. Because of past missteps and current scrutiny, Facebook is going to tread cautiously here (privacy was a central theme in Graph Search’s launch). I certainly understand why, but while those issues are sorted, I expect there are going to be a lot of screen shots of Graph search results being shared around the web.

Bigger privacy issues will likely arise around what might be called the Randi Zuckerberg principle – as in “Oh shit, I didn’t realize I’d show up in that circumstance!” Graph Search is going to expose all manner of privacy controls as super important, and send millions back to Facebook’s sometimes-confusing dashboards, so as to appropriately re-tool settings such that nothing untwoard shows up in this important new functionality.

And to me, this is a Very Good Thing. A couple of years ago, I wrote a post titled  The Rise of Digital Plumage in which I predicted that we’d all become habituated to “dressing” ourselves in structured data, so as to best present ourselves to the world at any given time. Graph Search is another important tool in our ever-growing digital wardrobe, one that motivates us to understand and manage the implications of our ever-expanding digital footprint.

Facebook just posted an announcement about its new search here.  The initial beta will roll out slowly, folks will have to ask to join a waitlist to get the service. I’ll be updating this post as the news is discussed and digested….

53 thoughts on “Facebook Is No Longer Flat: On Graph Search”

    1. I’m not sure this one won’t reverse that trend, to be honest. You all know I’m a critic of the service, but I see this as having potential for creating a positive use cycle, so to speak.

      1. I actually think people will begin liking and sharing less info about themselves when they realize how exposed their data and gestures will become, and how much marketers will be preying on that data.

  1. I do still wonder about the ability to monetize this. For instance, what value are ads in a SERP that’s predicated on things your friends have done? If I’m searching for good restaurants my friends have been to, I’m probably not clicking on a sponsored link that also appears in that space…

    1. You will if that restaurant is surfaced as liked by tons of “friends of friends” or, by people who live in your neighborhood, or by folks who live near the restaurant, or…you get the picture?

  2. The question that comes to mind is, “Will my mom use it?” She has an iPad, uses Facebook, plays Words With Friends, and is incredibly tech-challenged. She represents the majority of people I see on Facebook. You know the ones that post, “sorry, couldn’t figure out how to link to this.”

    So, will she use Graph Search. No, not a chance.

    When Google rolls out new search features, it’s often in answer to behavior users are already exhibiting. How many users currently are searching for weather on Facebook? How about “engineers who are a friend of a friend that like restaurants at or near the Canadian border”?

      1. I don’t think that Zuck is so concerned with moms here… This is a multi-generational and multidimensional transformation…

    1. Given that very few people know how to post to another person’s timeline, my guess is it won’t see widespread use unless we’re somehow prompted to use it.

      If it continues to be a shiny object long enough to get lots of refinement and UX love, then maybe it will someday become part of the average user’s experience. We marketers and community managers will surely use the heck out of it, though. 🙂

    2. I’m sure they will start auto-exposing to your mom in no time, so she’ll be using it without even knowing she is. The same way they’ll sell ads, they can also find popular queries for your mom’s demographic and start pushing things she’s interested in without her even asking.

    3. “When Google rolls out new search features, it’s often in answer to behavior users are already exhibiting.”

      Which is true, but unfortunate. Because it limits Google to, well, behaviors that users are already exhibiting. Which behaviors are defined and constrained by the tools that they currently have. Which means that you’ll never see a behavior, even if there is a user information need, if there is currently no way for the user to express that behavior. Classic catch-22.

      I have no idea if this Facebook graph search will catch on or not. But hey, good on them for trying. For not letting themselves be caught in the “existing behavior” trap.

    4. Wrong question. Facebook is trying to ‘eat’ LinkedIn, dating sites, restaurant recommendation sites, etc. If your mom isn’t even on Facebook, (mine sure isn’t) it doesn’t matter if she likes graph search. This is Facebook’s best chance to take it to Google in the ad business.

      1. I’m sure I’ll do a few graph search for fun every now and then, but I don’t want my preferences to be available for search. Facebook keeps pushing the idea of “like” means recommendation, I rarely like brand pages anymore. I’ve liked a lot of pages for coupons and other crap in the past, they’re NOT recommendations and I certainly don’t want people to see my likes in a structured way and interpret the data as my preferences…..ugh…I hate facebook

    1. Nope, you’re not going to upload that borderline photo or make the non-PC comment or express emotion or anything that be used by anyone to your detriment. You’ll be buttoned up LinkedIn style, self-editing, using Facebook persona optimization tools… and in not so long you’ll get tired of the neutered existence and you’ll move to a site that allows you to be you. I love the concept but I think the productization is way too uncontrolled. They are playing with fire.

  3. It is definitely an important tool added to Facebook. Still, the search results will be dependant on what people decide to share on Facebook. For example, LinkedIn users have a big incentive to include company names, titles, skills, location, etc.. because they realize that recruiters are potentially scanning their profiles. Facebook on the other hand scares some people from sharing too much due to its ever changing privacy policy. So, Facebook will need to incent users to share more of their personal data and the popularity of the search might be a good way to start this process.

  4. An excellent and incisive analysis John. I have been experiencing severely decreased engagement with FB myself, and actually contemplating terminating my account. But this certainly gives it a new significance and might convince me into staying on the network, if at all, to derive value from the Graph Search feature. Not a Google killer, but will certainly hurt Google, as people will take some of their searches to FB.

    And as for all those who think that the potential privacy ramifications of Graph Search will cause people to hunker down and set everything to private on FB; remember, only the smart, savvy and paranoid would actually remember to secure all their online activity. There will still be plenty of ‘normal’ people out there who do not value their privacy all that much. And they will keep adding volumes of data to FB which you can search and benefit from. Its not necessarily a two-way street.

    1. Thanks for this. THe role Microsoft is playing I will admit, I do not understand well. I’ve reached out to them.

  5. Really nice analysis! Enjoy reading your blog and in-depth thinking.
    I suppose my big question about the Graph Search (aside from the name!) is how it works or could work on mobile. I can imagine quite a few use cases on mobile (find nearest restaurant that my friends have been to and like etc) that are more prevalant on-the-go or on a mobile device. There’s several different services like FourSquare, Yelp etc that have a user base (albeit much smaller one) in solving these user needs.
    Also, separately, it feels like Graph Search should start to index listings data as well … like maps, local businesses etc to be truly useful. From my personal experience, Facebook’s listing data set is full of duplicates, incorrect listings and can do with a lot of moderation and fixing.

  6. Perhaps I am missing something here, but organizing information around FB friends seems like the wrong organizing principle? I methodically organized the people I follow on Twitter, and that’s an interest graph that I would like to search against. The interest graph of my friends on FB is just a little too Bud Light for me.

    1. This is the essence of Facebook’s persistent problem, its Achilles Heel. A core orientation around communication between friends and family (primary) almost by definition constrains its ability to monetize. People are simply not in a space cognitively where they are receptive to marketing. Interest Graph > Friends/Family Graph.

  7. Thank you, for writing this post, just an awesome job of presenting the value of the semantic web, opps I mean “graph search” 🙂 It’s going to be fun to watch the ripples this causes.

    1. No one has to be on Facebook, or to feed it data. That’s the thing we’re all realizing, and it will give us great power.

  8. Semantic web on FB island. Which is fine. Except I’m not sure the rich data that is on the island matches the use cases being widely touted?

    FB has rich data for friends and photos. For employment/jobs, places, business/service reviews…not so much…not outside of the top x% of the user pyramid. For the rest of the pyramid, FB data is suspect, e.g. skewed because they “liked” some place or business that they didn’t really like because the place gave them an incentive to do so, and meanwhile never liked the place that they really do like since there was no incentive to do so….

    If above is accurate (?), then isn’t FB back to the engagement issue you outlined so well (other than the people/businesses that will be motivated to extend their respective FB graphs; a (relatively) small group because only a small group will perceive immediate benefit for this investment of their time)?

    1. yes, they need to jump to another level of value. I think they have enough levers to pull to try to start the flywheel…if they get it right

  9. So perhaps graph search is a potential answer to what I was griping about in the comments on another of your posts, two weeks ago? That of web search being fundamentally uninteresting these days because it only gives common, expected, “zeitgeistian” answers.

    With graph search, I see much more potential for the questions and interactions to be exploratory and uncommon.

    To me, the issue hinges on whether Facebook plans to pagerank-itize the graph search algorithm. Are they going to filter, reduce, narrow, shallow answers to graph queries toward the expected, popular, common results? Or are they going to mix, expand, widen and deepen answers to graph queries toward the surprising, unexpected, serendipitous, comprehensive?

    Pagerank has the effect of reducing to the popular. Will graph search expand to the surprising? With graph search, that potential is there. But will they use it? Or will they, perhaps under pressure from advertising business models, fall back on the reductive pagerank-like approach? That’s the big question I’ll be following.

    1. Excellent questions JG. Will they publish the core algos? Will they be responsible in how the work with the ecosystem?

  10. I like the idea of Graph Search to look up photos, people that I went to a school, etc. But I can’t see myself using it to find a business or restaurant. I don’t recommend or “like” businesses on Facebook, and I don’t think many my friends do either, so how would using my friends as a search tool help me?

    1. Good question, and one that graph search will either spin up demand for, or fade…and I know Facebook wants this to work, so expect they will put gaph search in front of you quite a bit, encouraging you to use it and also to take actions that feed it. rather like Google has with Google+

Leave a Reply to johnbattelle Cancel reply

Your email address will not be published. Required fields are marked *