1 Comment

Summary:

If you’ve ever ordered a dish off a menu, chances are it’s in Food Genius’s servers. The startup has compiled a mammoth database of menus with the goal of tracking what America is eating. In January it begins selling that data to food companies.

Food Genius CEO Justin Massa. Source: Kevin Fitchard
photo: GigaOM / Kevin Fitchard

Chicago startup Food Genius knows what America is eating when it dines or orders out. It knows, for instance, whether more pizzerias are starting to put basil on their cheese and tomato pies or whether restaurants are still piling their burgers high with bacon or have switched over to avocado.

What’s more, Food Genius knows exactly how much more restaurants are charging for adding that sprinkling of basil to turn a standard pie into a pizza Margherita. Food Genius can tell you what hot combinations of ingredients, flavors and culinary buzzwords can boost the appeal – and thus the price – of a dish. In short, Food Genius has built the country’s biggest virtual test kitchen with menu data.

Food Genius has compiled a mammoth database of 100,000 unique menus from independent and chain restaurants around the country, giving it a massive data set to play with, said Justin Massa, Food Genius co-founder and CEO. The startup gets its data from some key menu aggregation partners, the biggest being fellow Chicago food outfit GrubHub. It also takes in readily available menu data from the national and regional restaurant chains, sometimes manually scraping data off of restaurants websites.

Any company can aggregate menus, Massa admits, but Food Genius is only using those raw lists of ingredients and dish descriptions as a starting point. It’s built parsing and categorization algorithms that break down those menu items into 14,000 ingredients, techniques and concepts that make up its internally developed culinary taxonomy. From there, it ferrets out the relationships between the items, like what categories they fall into and how different ingredients and techniques are commonly paired.

Food Genius screen shot

By crunching that data, Food Genius had generated more than 1 billion different food concepts, each of which represents a sort of meta-dish. Some of those concepts have been produced millions of times over in restaurants all over the country, like cheeseburgers, but many more are specific to individual restaurants. And others are just concepts inferred from Food Genius’s data.

Teaching a Kraft to cook like a Wolfgang Puck

While it might be tempting to think that Food Genius wants to use its data to automate the creative process of cooking, Massa quickly dispelled that notion. While Food Genius can suggest ingredient combinations based on patterns its database detects, pure math could never replace the chef, Massa said.

saute pan kitchen cookingInstead, Food Genius is doing the opposite. It’s tracking trends in independent restaurants and relaying that information to big food companies – restaurant chains, food distributors, consultants and most importantly food manufacturers.

“The big food companies take it on faith that innovation starts in independent kitchens,” Massa said. “The problem is the lifecycle of product development at a company like Kraft is two years. Meanwhile, in an independent restaurant, a new dish could be conceived and executed in 30 minutes.”

Food trends are becoming ever more fleeting. Bacon in desserts or gourmet sliders may be hot concepts today, but they could become passé quickly. By using Food Genius’s database, food companies can identify food trends with true mass appeal, and they can latch onto those trends early. The last thing a major food manufacturer wants to do is invest large sums of money and time bringing to market a new line of frozen southwestern chicken entrees, only to discover that the consumers are now into Thai cuisine with pork.

“We want to give them the confidence that they can catch a trend within nine months, rather than within two years,” Massa said.

Taking food analysis to the next level

The company plans to take its service live in January, and already has half a dozen food industry consultants and food product manufacturers signed up. It’s selling its data through a licensing model, with monthly fees starting at $2,000 per user.

Food Genius is one of the growing number of companies to emerge from Excelerate Labs’ accelerator program in Chicago. In September, it raised a $1.2 million funding round led by Hyde Park Venture Partners and Hyde Park Angels with participation by New World Ventures, IDEO, Amicus Capital and I2A Fund.

Noodle restaurant sharing share couple black and white pictureMassa said right now its customers are using Food Genius data to develop their own food concepts, but as the company scales it hopes to offer more custom analysis. Instead of trying to identify food trends themselves, Food Genius will tell them what trends they should be paying attention to, Massa said.

The startup is also looking for ways to refine its data by tracking what diners are actually ordering, not just what restaurants are offering. Knowing what dishes are in the menu doesn’t tell you which ones are popular. Food Genius would like to delve deeper into sentiment analysis, but Massa will that such data-mining techniques aren’t fully baked, especially when it comes to people’s mercurial tastes in food.

Food Genius could start pairing specific restaurant dishes with reviews in Yelp or dish sightings on apps like Foodspotting, but the information it would gather would be of questionable value.

“Say there’s this one guy who has written 10 Yelp reviews about 10 different brisket dishes he’s had, and he hated every single one of them,” Massa said. “Does he really hate brisket? Or does he really love it? Clearly he’s going to order brisket again at the next place he goes to.”

Saute pan photo courtesy of Shutterstock user Fedor Kondratenko; Dining photo courtesy of Shutterstock user Everett Collection

  1. Frank D. Muscarello Saturday, December 22, 2012

    Food Genius has unlocked the time it takes for large manufacturer to analyze trends and therefore identify which products to make. $2000/month should be $2m a year they make their R&D more efficient.

    Share

Comments have been disabled for this post