New England clam chowder at Seattle's Pike Place Chowder is Yelp's most popular dish, as ranked by machine learning technology. / <a href='https://www.facebook.com/pikeplacechowder/photos/a.10152136284856164.1073741826.233601046163/10155263988006164/?type=3&theater'>Pike Place Chowder</a> New England clam chowder at Seattle's Pike Place Chowder is Yelp's most popular dish, as ranked by machine learning technology. / <a href='https://www.facebook.com/pikeplacechowder/photos/a.10152136284856164.1073741826.233601046163/10155263988006164/?type=3&theater'>Pike Place Chowder</a>
Tech

Yelp Adds Popular Dishes Feature to Help You Make Up Your Mind

When it comes to dining at restaurants, we eat with our eyes before we eat with our mouths. Over the last decade, we’ve become conditioned to check ratings and reviews before choosing a restaurant, and photos have become a huge player in the game.

According to a recent survey conducted by TripAdvisor, 94 percent of U.S. respondents reported basing a dining decision off of an online review and 60 percent have been influenced by restaurant photos.

Yelp just unveiled a new feature in its iOS and Android mobile apps: popular dishes. Menu items are ranked as most popular by current Yelp reviews, photographs, and mentions, and displayed in order with their name and price. “We know that people love pictures and looking through photos of food for figuring out what to order,” said Yelp product manager Matt Geddie. “They’d also read the reviews and review highlights and try to put together what they should order. And we thought while kind of exploring that content individaully is still important, there might be a way we can make the use case a little easier by kind of pulling together all the photos and reviews about each dish and telling users which ones are most popular.”

It’s a small feature addition for Yelp, but one that signifies the growing importance of a good food photo when choosing a restaurant and dish to order. In fact, Yelp places a lot of weight on the photos accompanying its reviews. “We use a lot of machine learning to rank photos for how beautiful we think they are. We have this quality machine learning model that takes a lot of factors into account — like the resolution of the camera and colors and brightness. We also have a lot of other machine learning that tells us what’s in the photo. We use a combination of those signals to rank photos for this feature as well as for the business,” said Geddie.

Photos are a huge part of any restaurant decision. OpenTable acquired a small startup called Foodspotting in 2013 that encouraged diners to post photos and recommend specific restaurant dishes — not just restaurants. The functionality has since been rolled into OpenTable’s own system. 

Restaurants don’t have a say in which dishes are deemed most popular just as they don’t have a say in how customers review their business. Though according to Geddie, the company is working on some ways for consumers and business owners to notify Yelp if a particular dish has changed or fallen off the menu.

The functionality is currently only available on Yelp’s mobile apps. “We wanted to start first with the apps because that’s really where our most engaged users are,” Geddie said, though he added that desktop functionality should roll out in the future. Yelp has 30 million unique monthly app visitors and 74 million unique monthly desktop visitors across all of its listing categories. Restaurants rank second behind “shopping” as the most reviewed categories on the site, and account for 17 percent of businesses reviewed.

More from Skift Table