Microsoft Roots, Social Media Chops: Spindle’s Take on Local Search
We’re now six full years into the smartphone revolution, and software developers of all stripes still haven’t figured out one of its most promising equations: How to put consumers and local businesses together at the right time, in the right place.
And oh boy, have they been trying. From giants like Google and Facebook to the newest of startups, there’s an overload of apps that pump out feeds of nearby meals, discounts, and diversions. But no clear winner.
Why’s this combination so difficult to solve? Pat Kinsel, the CEO and co-founder of Boston-based Spindle, thinks it’s about quality. And at Spindle, that means finding a smarter way to sift and sort social media content.
“Almost all other services are focused on place recommendations,” Kinsel says. “What they end up with is this sort of absolute sense of ‘this place is better than that place.’ So it’s really a Yelp alternative. And that’s great, but I find that that’s pretty stale.”
All of the applications that Spindle jockeys with for attention are focused on roughly the same behavior: A smartphone owner is looking for things to do around them, whether that’s a restaurant to try for dinner, a museum or theater for entertainment, or a cool new bar for drinks. And the apps all do basically the same thing. The big difference is where they get their information (and how good that data is), and how much effort they put into sifting it to make the best stuff stand out.
Why are there so many? Well, there’s no clear winner yet—and this is such an obvious function of smartphone-equipped users, that someone will eventually make an app that does it well. The main reason is money—specifically small and medium-sized businesses that are often local or regional in scope.
If you can turn the smartphone into an honest-to-goodness useful tool for leading consumers into a business and making the cash register ring, you’ve just built a huge new advertising network and unlocked one of the Holy Grails of business and marketing that haven’t yet been shaken up by the Internet generation.
As someone who recently moved to a new city, I personally find myself stuck in this mode a lot—but there’s nothing that has quite threaded the needle perfectly for my own needs, despite a ton of startups trying to make it happen.
Foursquare gets close, but mostly because it has a lot of data from its core function, allowing people to “check in” to places they visit. The city-search feature of Foursquare, called “Explore,” is something the well-funded company has started to emphasize a lot more in recent months.
I’ve found that its results are not always on point, since you’re in the position of trusting a lot of other people’s preferences. And the advertising Foursquare sells on its map of nearby places can mess with the interface, annoyingly and repeatedly snapping your location to some unwanted business that paid to get to the top of the pile. It’s imperfect, but I still use it all the time because I haven’t really found anything better.
Bigger software companies have been hungry to crack this nut too. Facebook and Google have tried several test projects, both rolling out new attempts in late 2012.
Facebook’s newest foray into the “local discovery” arena is called Nearby, a feature inside the main Facebook mobile app that does all of the standard find-a-place stuff. Facebook has a huge amount of detailed user data—at least for users who “like” and “check in” to places they visit through Facebook—which it uses to sort nearby results to find the best recommendations.
Google’s most recent venture into local discovery is an app called Field Trip, which was released in the fall. Again, all of the standard stuff is present, with Google doing what the search behemoth does best, ranking and filtering results for relevance. Google relies on a bunch of different sources, including the Food Network and (Google-owned) Zagat, to determine quality.
Spindle gets its data from the two big names of social networking: Twitter and Facebook. Its feed tries to harness the information that businesses are putting out about themselves on social media to give consumers a sense of what’s on sale, who’s performing, or whether there are any open bookings that need to be filled.
To make sure it’s not just a firehose of social-media postings, Spindle built an algorithm that pushes items up or down in its ranking based on some subjective measures. The goal, Kinsel says, is to whittle the flood of information down into things that are timely and interesting—the kind of stuff that would be more likely to make someone walk in the door of a local business.
Relying on just two sources of data can have its pitfalls. If you’re in an area where businesses aren’t exactly hip to the whole social-media game, the feed can look pretty sparse—especially if you’ve got the search radius tuned down to less than a mile, which is pretty typical for what someone on a wintertime stroll for lunch would want to see, for instance.
Spindle’s bet seems to be that social sources will continue to grow and be adopted by business, solving any scarcity problems—“If you look at the social content being shared, it’s like absurd,” Kinsel says.
“If you were to open up other apps … they’re not telling you what is happening at this place at this particular moment,” Kinsel says. “There are other apps that try to ingest content from these sources, but we don’t think they have this notion of timeliness or relevance.”
So, for instance, if a nearby coffee shop posted to its Facebook page that it was going to have a special tasting of fancy imported beans today at 2 pm, Spindle would make sure I saw that before the event happened, but not after. Or, if the tasting was going to happen over the weekend, Spindle’s algorithm would store that bit of information for a few days until it became ripe.
To select for “interestingness,” as Kinsel puts it, Spindle tries to apply several different kinds of filters. “So we say things like, we think coffee shops are interesting in the morning and not in the afternoon unless they actually use the word ‘afternoon,’” he says. “We think that bars are particularly interesting when they talk about live music.”
That all gets computed when someone pops open the app and starts looking for things around them. Stuff that isn’t timely or seems uninteresting to a Spindle user gets suppressed by the algorithm, and messages that would more likely apply to a person walking around with their phone open at lunchtime, for instance, move to the surface. Users can also give feedback that helps tune the system even further (the app is currently live in Boston, San Francisco, and New York).
It’s the same kind of scoring exercise that search engines do all the time. And it’s no mistake that the Spindle crew spent time building an algorithm like this.
Before starting the company, members of Spindle’s founding team worked at Microsoft, both in the Boston area and back at the company’s home base near Seattle. They met as part of Microsoft’s internal innovation labs projects, which went under a few names, including Startup Labs (headed by current TechStars Boston director Reed Sturtevant) and FUSE Labs (under former Microsoft software chief Ray Ozzie).
One of those projects was a version of “social search” for Microsoft’s Bing search engine. Bing, as you may recall, has long had a special relationship with Facebook, getting access to a level of data from the social networking leader that others (especially the common enemy, Google) can’t get its hands on. Microsoft, of course, also has invested in Facebook. The Spindle crew also worked on Docs.com, a project that sought to make Microsoft’s Office software more easily usable and sharable online.
During that work with Facebook, founder and CEO Mark Zuckerberg made a big impression with his exhortation about “reimagining something from the ground up to be social,” Kinsel says. But that had mixed results when it came to reinventing Web search—no matter how much you littered a sidebar on the page with social-media results, or blended “likes” and shared posts into the main feed of search results.
“A lot of people, you’d put it in front of them, and their response would be: ‘OK, great. This is search,’” Kinsel says. “And we’d say, ‘No, no, no, but it’s different!’ And they’d go, ‘OK, but it’s search.’”
“So when we left, we said we really wanted to come up with a fundamentally different way of ranking the content, and also a fundamentally different experience. So it’s really, for us, this combination of: What are all of the signals that your phone provides, and how do we translate that into a meaningful query?” Kinsel says. “We know where you are, we know the time of day, we learn more about your behavior over time. How do we turn that into an interesting query?”
Answering that question well takes time. So, in contrast to a lot of the “lean startup” approaches championed by many technical entrepreneurs these days—build a minimum product, get it out asap, and start experimenting and tweaking in the real world—Spindle took a decidedly deliberative approach to building the core technology that helps it ingest and re-order social data.
“We didn’t ship anything for almost a year and a half after starting the company. And we built a couple of things and said, ‘You know, not interesting enough,’ and kept going. And that’s also where we really made the investment in these tools,” Kinsel says.
As a result, if Spindle finds enough interest to keep adding feeds or new applications (like sports, for instance) that show off what’s happening in a location, the startup thinks it has the core technology to make building the new application much easier.
“That’s why we say we’re trying to build the discovery engine for the social Web. Yes, we really believe strongly in this app. But on the technology side, we’re trying to do much more,” Kinsel says.
Spindle announced in November that it had raised $2.3 million in total financing, from Polaris Ventures, Greylock Partners, Lerer Ventures, SV Angel, Atlas Venture, and Broad Beach Ventures. Also listed as investors are some connections from Microsoft: Ray Ozzie, Raman Narayanan, and Project 11, the fund founded by ex-Microsofties and current Boston TechStars honchos Reed Sturtevant and Katie Rae.
The startup released an early version of its app last August, and a more ready-for-primetime update in November. Kinsel wouldn’t say how many users the app has today, but he did allow that so far, “User growth is not that impressive. What we’ve been focusing on is repeat usage of our user base,” which Kinsel says shows people are coming back three or more times a day to check out what’s new.
Look for Spindle to add more investment cash as it tries to build out its user base in new cities, adding to the current markets of Boston, San Francisco, and New York. With the big dogs of technology and a horde of other upstarts all lining up to find an opening in this market, the race is starting to get interesting.