Last week I met with Giuseppe D’Antonio, the CEO of Milan-based social service Circleme — a recommendations engine that is currently in beta — to find out what his service was up to.
It turned out that the basic thrust was simple: just tell Circleme about the things that you like and it can help you find more things to enjoy. Users can find films to watch, bands to listen to, restaurants to eat in and much more.
A simple idea, yes, but you’d be forgiven for thinking that it also sounded very familiar. After all, it seems like everyone and their dog wants to branch out into recommendations right now, whether it’s Foursquare trying to tip you off to places you’d like, TripAdvisor helping you find a hotel, or Yelp trying to corner the local services market.
Oh, and then there’s Delicious, which is reworking itself into a web recommendations service, and hot new apps like Pinterest and Amen. Oh, and then there’s a couple of little sites called Facebook and Google+ that are going head-to-head in their battle to capture social signals.
You get the idea: this is a crowded space.
D’Antonio thinks Circleme is different, however, for a few reasons.
First, he says, it’s big. Rather than coming from a niche, it’s trying to cover lots of ground from the very start. The things you can like are divided into a range of categories — movies, music, places, goods and so on — and there are currently 650,000 individual items in the database, with as many as 12 million waiting in the wings. It doesn’t exist in a vacuum either, pulling in your data from Facebook, Foursquare, Netflix and Goodreads (with more to come). That means most new users won’t have to face a standing start.
Second, he argues, the site isn’t just a database of stuff with recommendations built on top. It’s actually grown out of some strong technology — in particular, a semantic engine called Cascaad, which we wrote about when it launched in 2009. Cascaad makes it easy for the system to connect and rank new items that users add to the site, and Circleme is a reworking of this technology for a broader consumer audience.
Third, he points out, Circleme is well-designed — something that’s really important when you are trying to encourage new users in. He’s right: the service is straightforward to use, intuitive and quite beautiful in places. And while new features being added to the beta all the time — today, for example, the site is adding an “activity board” for users to see what’s happening — the team seems to be working hard to retain a sense of elegance wherever possible.
Of course, there’s design and there’s design. But it’s done well, which can make a big difference: take Path 2, for example, is slowly inveigling it’s way into my life because it’s a pleasure to use — and much simpler than using all of the different individual services out there that it connects me to.
Users seem to be responding, says D’Antonio. The stats are good, and early usage is growing fast — both in terms of the number of users and the amount that they are using the site.
“Of course it is early to make any conclusions on these numbers,” he admits. “But we keep track of these every day, and every time we do a new release we make sure that the numbers look better than the day before. I believe that if we can keep these numbers growing, as we grow our community, we have a good chance to do well in this space.”
But even though Circleme seems to be getting many of these things right, it’s got a big task on its hands: becoming a major player in the recommendation game won’t be easy. In fact, its three strengths could also be its weaknesses.
It’s starting off broad, but many great recommendation services start off (and remain) in a niche that means they can focus on being the best at their individual type of activity: say Last.fm for music. There’s a graveyard full of startups who have tried to go too big too soon, and even those who have succeeded to some degree — such as Hunch, which was bought by eBay recently — have had to radically alter themselves along the way to keep going.
Great interfaces, too, are all well and good — but they don’t solve all problems. Right now Circleme’s got two pinch points: getting data in and then getting it out. To bring data in you can piggyback on other services, or you can enter items individually, or you can download a bookmarklet that lets you like a web page or thing — but experience says it needs more. At the other end, being a great recommendation service means delivering tips to people precisely when they need them most, and that suggests to me that we need to watch carefully to see how it expands into mobile.
And finally, the service’s success will ultimately rest on two things: the accuracy of its technology, and the ability to capture lots of activity and data. Circleme has already got the technology in Cascaad, but it’s making a bet that the act of liking is enough in and of itself. And that’s where competition is at its fiercest because the most successful recommendation platforms are the ones that start out doing something else entirely.
For example: these days, the place I go to find links that I might be interested in is Twitter. Lots of people find Facebook is now their serendipity engine. Convincing people to keep all of their interests and passions in one place is hard — but if you have a way to capture lots of user time and activity, you can start doing things with it.
Could Circleme make a decent stab at taking on this market? It’s certainly got a chance, though it’s going to have to fight hard to rise above its rivals.
Still, somebody has to get recommendations right sometime, don’t they?
If you want to take a look at Circleme’s private beta, there are 500 signups available for GigaOM readers.