It’s no secret that Facebook’s existing search engine is bad; that drum has been beaten to death. But you don’t have to wait for Facebook to make it better; there’s a pretty darn good Facebook search engine out there now.
It’s called Trove, and it’s officially launching on Tuesday. But you can connect your Facebook account and get started now; the site had a soft launch about two weeks ago. I’ve been testing it, and it’s easily the best Facebook search experience I’ve seen. There are a few missing features that I’ll mention below, but Trove’s potential is obvious.
Seth Blank, Trove’s founder and CEO, says the service isn’t a replacement for Facebook search. He’s sure Facebook will eventually have its own great search experience. But the Trove team, which totals three full-timers plus contract help, is aiming bigger: at social search in general.
“Social search is like an iceberg,” Blank said on a call this week. “Most of it you don’t see at all, and even what you do see disappears too quickly. We want to bring together everything from everywhere.”
How Trove Works
After you’ve connected your Facebook account, Trove scans the content associated with your account — your own updates, links and photos, plus those posted by your friends — to build its searchable index. When it’s done, you end up with a way to search back through all that content (well, at least what’s allowed by Facebook’s API) to find links, images, names, and more. Here’s a look at a search that I did this week for “food” — I’ll blacken the names and avatars associated with each post for the sake of privacy (and more on that topic in a bit).
Trove presents search results in reverse chronological order, which wasn’t how the team originally programmed it. At first, Trove defaulted to showing results based on relevance, but repeated feedback from early users was that the display was “broken,” Blank says, so Trove switched to sorting by date.
What I found most impressive about Trove is that it’s not based on pure text matching. If you look closely at the screenshot above, you’ll see that Trove found content that doesn’t match the search term “food,” but does include related terms like “eating” and “restaurants.”
“We’re building a contextual graph based on how people talk,” Blank says. “Most social search, if you give it the word ‘cat,’ will show you things like ‘cats’ and ‘caterpillars.’ We can serve content that doesn’t match, but is still relevant.”
Trove provides a number of filters at the top of the results page that let you limit searches to photos, links, check-ins and so forth.
One notable thing that Trove is missing, though, is the ability to search only through your own content — or, for that matter, only through one friend’s content. Blank says it’s easy to do and is on Trove’s radar, but they haven’t developed a great interface for it yet. It’s a must-add feature.
That’s not all it’s missing: Trove’s to-do list includes things like supporting search through Facebook Groups and, more importantly to most users, updating each user’s scan/index more often. During the pre-launch phase, Trove hasn’t been updating users’ scans at all; Blank says that will be changing soon, so that Trove can show recent Facebook content.
Any social search service comes with privacy concerns built in. Trove is no exception, especially since it’s launching as a search engine for Facebook — the social network that comes with more built-in privacy baggage than any other.
When you connect a service, only you have permission to see what you have connected, you are not granting your friends or any unauthorized third parties the ability to see your data.
Trove is also planning a tool for users that change their mind after connecting their Facebook (or other) account. It’ll let users delete their account and remove their indexed data. The tool doesn’t exist at the moment, so the company is doing it manually when asked. (See the Trove FAQ for more about that.)
The Money Question
Trove already has funding and Blank says the company is in the process of raising more now. So, like many startups, Trove doesn’t have to worry about making money on day one.
“We’ll figure out monetization later,” he says. “Our focus for at least the first year is to create an amazing experience for users and to constantly make our search results better.”
Trove, Blank says, has several money-making options that it could eventually pursue, from the obvious option of putting ads on the site to things like syndicating its technology for private uses.
As I’ve mentioned above, there are a few obvious features that Trove is lacking, like the ability to search just through your own content and/or just through a friend’s content. And setting up regular rescanning and indexing is a must. Both of those, Blank says, are on Trove’s to-do list.
He’s also focused on trying to improve the onboarding process. Due to Facebook’s API limits, it takes at least several hours — and sometimes a full day — from the time a user signs up to the time his/her content indexing scan is complete. A very limited search is available almost immediately, but Trove’s real value isn’t obvious until that scan is finished and all the content is searchable. Blank knows that Trove needs to improve in this area and make the site more immediately beneficial.
Trove only does Facebook search now, but Blank says the plan is to expand in coming months to include other social networks like Tumblr, Flickr, Instagram and hopefully Twitter, depending on how its recent API issues play out.
If Trove nails its to-do list and adds additional social networks, an already very good Facebook search engine has the potential to be an excellent social search engine.