[Ed. note: This is the first of what will be an ongoing interview series called "5 Questions With...," in which we'll publish brief interviews with interesting and important online marketing newsmakers. Depending on the topic, the interviews may be published on Search Engine Land or Marketing Land and, depending on the topic and interview subject, we may occasionally ask more than five questions.]
Facebook made a splash this week when it announced Graph Search, an upgrade to its long-neglected search tool. It’s very much a beta product. Facebook emphasized that the product will expand to include searches for status updates and other text content, not just “Likes” and check-ins and photos and the limited data it searches now.
But astute Search Engine Land readers will remember that there’s already a search engine that’s tackling Facebook search — and doing it differently than Facebook itself. Last September, I profiled YourTrove, a social search engine that crawls Facebook content and provides a searchable index that includes the text in status updates, photo captions and more. You can’t get that in Facebook’s Graph Search – not yet.
We reached out this week via email to Jesse Emery, Co-founder and Chief Identity Officer at YourTrove, for our first “5 Questions With…” interview, to get his thoughts on Graph Search, how it’s different from YourTrove, and how it might impact YourTrove’s plans.
5 Questions With … Jesse Emery, YourTrove, On Facebook’s Graph Search
Matt McGee: The Graph Search that Facebook announced this week is different — at least in its current form — from what Trove is doing. Can you explain how?
Jesse Emery: In the two services current forms, there are two big differences. The first is that Graph Search is querying against very explicit user actions — such as Likes, Tags, and Check-Ins. So when Facebook talks about a query such as “Friends who like Star Wars” they literally mean that the search will return your friends who clicked the “Like” button on the Star Wars page. YourTrove, on the other hand, does text search through posts and other meta data, such as photo captions.
The other, and probably bigger, difference is that YourTrove is much more focused on user content, rather than say, people or places the way Graph Search is focused. While Graph Search is also returning photos (presumably via photo owners and tags), YourTrove will return photos based on the captions and comments. So, while YourTrove can’t (currently) do “Friends who like Star Wars,” it can search for “Star Wars photos” and probably return much better results than a Graph Search for Star Wars photos can at the moment.
In the long-term the biggest difference is that YourTrove was designed from the ground up to search content from a lot more services than just Facebook, so that a user could search, for example Facebook, Tumblr and Pinterest content simultaneously.
MM: You mentioned on Twitter that YourTrove began by using structured queries like this, but you guys weren’t satisfied with the results. Why? Was it not comprehensive enough for what you had in mind?
JE: What I was getting at with that tweet is that we used to store the content YourTrove ingested in a very structured way. This made it really easy to do explicit, filter-like queries like “Show me Matt’s photos” (a very Graph Search-like query, no?). But YourTrove is content focused and we’re more interested in enabling users to do searches like “Show me photos of Matt’s dog” or even just “Matt’s dog” and return photos, videos, and statuses about Matt’s dog. To do those kinds of queries, in any kind of performant way, we had to un-structure a lot of that data.
Note that this isn’t to say that we can’t do “Show me Matt’s photos,” in fact that’s pretty trivial from our system’s standpoint. We just haven’t focused on exposing and refining that from a UI/UX standpoint the way Facebook has with Graph Search.
MM: What are your thoughts on Graph Search — strengths? Weaknesses?
JE: I am on the waiting list like most people, so I have only been able to do the demo searches. I think the most obvious strength is that Graph Search is a vast improvement over the current Facebook search, which left a lot to be desired. It’s also a huge step for Facebook in terms of interest and location oriented discovery.
The most obvious weakness is that it’s not searching or returning things like posts, comments, and links.
Additionally, I am worried that it’s going to make property owners (i.e., Page admins) even more aggressive about trying to get users to Like them. There are already a lot of annoying practices in that arena and Graph Search just further incentivizes trying to get Likes.
MM: How do you think the average Facebook user will react to Graph Search?
JE: I think, on the whole, most people will find it to be an improvement. Obviously, since it’s Facebook, there will be the expected gripes and people surprised by their privacy settings, but it’s just so much better than the current search that I think most people will view it positively.
MM: Will Graph Search impact the development of Trove?
JE: In the short term, not at all. For one thing, Graph Search won’t initially be available through the API, but additionally, right now Graph Search and YourTrove almost perfectly complement each other. Both of them are pretty good at the other’s weak spots.
MM: Give us an update on Trove and how things have progressed since I spoke with you guys back in September.
JE: We’ve been working on several things.
The first is getting users into YourTrove faster. We haven’t been nearly as fast at getting people off the wait list and into the system as we would like and we’ve been working on both software and hardware improvements in that arena to speed things up.
We’ve also been beta testing Tumblr as an additional service that you can add to YourTrove and search content across both Facebook and Tumblr.
Many people have been asking for us to get more frequent content updates from Facebook and that functionality is implemented and in the testing phase now as well.