Bear Naked granola launches a direct-to-consumer e-commerce site with IBM’S Chef Watson suggesting custom ingredient pairings.

Bear Naked granola is exposing all of its ingredients to IBM’s Chef Watson.

Chef Watson is a branch of IBM Corp.’s machine learning Watson technology, which uses an algorithm that learns over time when it’s exposed to new data instead of being explicitly programmed by an individual. Chef Watson is a part of the Watson technology that is specific to the food industry and can predict which ingredients will taste good together based on the chemistry of the ingredients and how often those ingredients are paired together in recipes.

Bear Naked granola began selling online to shoppers in March, enabling consumers to create custom-granola, with the help of Chef Watson. The decade-old consumer packaged goods company offers about 50 ingredients for shoppers to choose from in their granola mix, including non-traditional granola ingredients such as jalapeño, green tea matcha powder and chipotle essential oil. In addition to exotic extras, Bear Naked offers three granola bases: cacao-cashew butter, chocolate or honey. And so, there are more than 500,000 possible combinations, says Chris Tutor, vice president of strategy for Bear Naked Custom.

With a vast array of potential flavor combinations, it’s not always obvious if certain pairings will taste good together, Tutor says. “We wanted to allow consumers to create their blend and wanted to let them know if their blend would taste good.”

Enter Chef Watson.


When a consumer creates a custom granola mix, she first chooses her base. The site then offers her three complementary ingredients from different categories such as a fruit, a spice and a nut. After she picks one ingredient, Watson continues to serve her ingredient choices that would pair well with what she has already chosen until she narrows it down to three ingredients.

A shopper can dismiss a choice to prompt Watson to serve her more options, or she can click “view all the ingredients” if she doesn’t want to go through Watson’s recommendations. About half of shoppers go through the Watson curation and the other half go right for the full list, Tutor says.

Watson plays a role in both scenarios. When a consumer picks her mix from the full list, Bear Naked will give the consumer’s granola a “synergy score,” which is Chef Watson’s score on how well those ingredients will taste together. The higher the synergy score, the more likely that is a good-tasting mix.

Chef Watson on also provides Bear Naked with business insights it can use when developing future products, Tutor says. For example, if many shoppers choose to include lavender in their granola mixes, the CPG company may choose to include lavender in more if its products that it sells at retail stores., which launched in March 2016, is the only place where consumers can make a custom mix of Bear Naked granola. However several retail site carry Bear Naked prepackaged granola products. also sells “Chef Blends,” which are pre-blended granola mixes, although the majority of products sold on the site are the custom mixes, Tutor says.


Bear Naked worked with digital consultancy firm Avionos to develop this feature. It took about six months to build the feature and launch the site, Tutor says. The company is getting the word out about its custom offering via social media and display ads, as well as search and email marketing. Tutor declined to say how much it spent on the Chef Watson integration.

The average time spent on is about three to three and a half minutes, he says. Tutor declined to provide sales data on the new site.