The trendy Williamsburg restaurant boasted five stars on Yelp, but Ashot Gabrelyanov found it grungy and unremarkable. When he got home, he scrolled through the reviews, trying to see what so many people liked about the place. Then he understood: the restaurant offered unlimited mimosas for $12.
That jarring experience led Gabrelyanov to invest $150,000 in developing Borsch, a food recommendation app that could be described as the anti-Yelp. Instead of subjective reviews, the app offers photos, presented in the curated style of Instagram. You can browse nearby restaurants by dish – “burger,” for instance, brings up tantalizing images of available patties – or by mealtime.
The app’s debut comes as visual platforms threaten to overtake review-based competitors in the food space. Instagram and Pinterest have eroded Facebook’s dominance among food brands, according to L2’s Digital IQ Index: Food, and deals like OpenTable’s $10 million acquisition of Foodspotting are becoming increasingly commonplace. Gabrelyanov pointed to Black Tap, the New York burger joint that gained notoriety not through reviews, but on Instagram, where its elaborate milkshakes grew so popular that getting one required hours of waiting.
“Pictures can promote your restaurant better than reviews,” he said.
While Yelp touts its openness – anyone can be a critic – that also makes it difficult to understand what a restaurant is actually like. The author might hate broccoli, or with fake reviews becoming a growing problem, they might not even be real. While Yelp does feature photos, browsing through them can be time-consuming.
“You start with the idea that you want to eat something, then you type it in, see a list of restaurants, click on them, click on the pictures,” Gabrelyanov told me. “We help people find the best meals around in one click.”
The app is currently available in 10 restaurant-dense cities including New York, San Francisco, London, and Berlin. In addition to selling advertising, Borsch aims to partner with delivery companies like Postmates, charging a fee for each transaction.
Many of the images displayed come from user uploads, which the app then recognizes and categorizes appropriately. After analyzing over 10 million pictures, Borsch can recognize over 300 dishes and identify the 15 most popular dishes with 98% accuracy, although it’s occasionally stymied by more avant-garde fare.
“One time a London chef made a dessert shaped like a Lego block,” the founder said. “A human couldn’t even recognize that.”