Ness, the restaurant recommendations app that uses social data combined with machine-learning techniques to offer personalized suggestions, has released the next major version of its application today, now focused on what it’s calling “instant recommendations.” In the earlier version, Ness relied on user-initiated searches and a setup wizard that asked users to select their favorite cuisines and rate restaurants. Today, users are instead taken directly to the app’s homescreen for immediate recommendations based on time of day, location and popularity.
When new users first launch the revamped app, they’re shown only a couple of explanatory screens before seeing the list of popular places nearby. Going forward, training the app to learn your tastes by rating venues or liking cuisines is now an optional feature, accessible through the “Personalize” setting.
While this onboarding process obviously makes Ness easier for newcomers testing the app out for the first time, it also allows Ness’ previous users a more “lean-back” type of experience, as well. Now they don’t have to launch the app then kick off searches; Ness will just know what they like and make suggestions.
Ness co-founder and CEO Corey Reese tells us that even if users never dive into the “Personalize” section, the app is now able to improve its recommendations in time. It also lets users rate places from restaurants’ cards themselves, slowly building up a database of user likes and dislikes. However, Ness can implicitly learn a user’s tastes, too, says Reese.
“Let’s say you don’t tap on Vietnamese places ever – Ness will pick up on that. It will start showing you fewer Vietnamese places,” Reese explains. “But let’s say that every time you see a Japanese restaurant, you tap on that, and take a look at dinner time, it will start showing you more Japanese restaurants at dinner time.”
It’s a simple enough concept, but it has taken the company two years of building to get to this point. The company has spent a large part of that time combining the data it licenses from various vendors and matching it up with data normalized across social services, including Facebook, Foursquare, Instagram and OpenTable.
Ness also now uses social data, among other things, to explain why it made the recommendations it did. For example, it might tell you that several of your friends have tried this sushi place. Or it might be more of a matter of two places being similar (e.g. you liked X so you might like Y). Reese says there are more than 40 different types of unique explanations provided to users in the new app.
Ness 2.0 adds a couple of other features, as well, including the ability to add places to lists, share places via email or text, and, in this same screen, the ability to swipe to call, get directions or reserve the table via OpenTable.
With its ability to offer fast, personalized recommendations by cuisine, proximity, popularity and time of day, as well as that informed by social data, Ness 2.0 is starting to look more like Alfred, the local recommendations app Google acquired in late 2011. (That app hasn’t been updated on Android since then, though the iOS version was updated last year to sync data via Google to different devices).
Also like Alfred, Ness is preparing to expand into other verticals beyond just food. This year, the company will use its same technology to begin powering recommendations for nightlife spots, as well as activities and events. Other verticals may follow.
The company is also weeks away from a new web version of its service, which will build out Ness 2.0’s list feature even further.
“You’ll be able to more easily pull things into lists, share them and syndicate them across your social networks,” Reese notes. The web version will also be built with a responsive design, he says, allowing it to work on Android, where Ness doesn’t offer a native experience today. The hope is to also go live on Android before the year is out.
Ness raised a $15 million Series B round last summer, and has now grown from 3 million user-provided ratings at that time to 4.5 million today. Reese compares this to Yelp’s 7 million+, saying he feels the app is on a good path. The company has also seen more than 300,000 downloads, over 3.5 million user searches, and 30 million+ recommendations served to date. Its restaurant database today includes more than 500,000 venues across the U.S.
The new iOS application is available for download here in the Apple App Store.
Comment