At the intersection of big data and better food choices

September 2, 2014; 00:16 am by Bill Bishop

<< Previous 

Elliott Grant’s 20+ US patents testify to his love of finding elegant solutions to hard problems. Among his creations is a technology that enables capture of supply chain insights on over $10bn worth of products a year. When we heard he was working on an app that helps consumers connect their food shopping with their nutritional needs, we wanted to know more. Here, he talks with us about how shoppers use the app, how it personalizes nutrition scoring for individual products, and what retailers can learn from the data it generates.

Tell us about the origins of ShopWell. How did it get started?  

ShopWell was dreamed up in 2008 at IDEO, the world-famous innovation and design firm in Palo Alto, CA; it spun off as an independent company in 2010.  The original vision, which remains true today, was to create a simpler link between our health and what we buy at the grocery store by personalizing and simplifying the nutrition information on a product’s packaging. The first implementation was a searchable website, but as shoppers rapidly embraced mobile phones and started using them in-store, it was obvious it had to become an app as well.

Today, ShopWell is a personalized nutrition app that’s been downloaded by more than a million Americans. Users complete a simple nutrition and allergy profile when they first register, then they scan any barcode in their kitchen or supermarket and instantly see a personalized score for how well those foods fit their unique nutritional goals (celiac, heart health, diabetic, weight loss, allergies, etc.). The app will even recommend healthier alternatives in the store or nearby.

The application’s “brain” is a sophisticated algorithm that combines the expertise of a panel of registered dietitians with the recommendations of trusted organizations, such as the American Heart Association and the Institute of Medicine. 

How do food shoppers use the app?

Today we see two major behaviors.  When shoppers first get the app and tell it their age, gender, and dietary goals, they find it really rewarding to scan their pantry or shopping cart to see if the things they’ve been choosing are actually a good fit for them. The app suggests alternatives they may not have considered, and highlights ingredients they may not have understood.  

Once a user has done this for a while, their behavior switches to discovering new items or seeing whether a product they are considering is worth trying.

We’re working on creating a third behavior – which is to be mindful of how we shop and maintain good behaviors – by adding weekly reminders, tips, nudges, and helpful, science-based suggestions. The goal is to help shoppers stick with their dietary path.  

Overall, we want to be viewed as a trusted and friendly way to get advice and ideas when planning a shopping list or while in the store.

What do shoppers tell you they like about the app? What improvements do they ask for?

We receive tons of feedback from users. The general themes are that the app makes shopping easier, quicker, and less stressful. People love discovering new brands and products that are a better fit for their personal needs. Unlike calorie counters (which can be very time consuming), ShopWell users like knowing that if they make one or two healthy ‘trade ups’ suggested by the app when they go shopping, then they’re on their way to a healthier diet. Shoppers with serious dietary restrictions – or family members with dietary restrictions – tell us the app has become an essential tool.

With regard to improvements, shoppers often ask us about three things.

• Include new food allergies (we already support 17 different allergens)
• Include support for new diets (such as low FODMAP*)
• Make more private label products accessible

We’re always working to stay on top of these. The private label products can be challenging. Our system is built on databases that use the UPC as the key to unlock the ingredient and nutrition information, but some retailers with extensive private label lines haven’t published their ingredient and nutrition data, which makes those products inaccessible from the app.

ShopWell produces a personalized score for how a product fits into the shopper’s unique needs.  How does this work? And why should shoppers and retailers have confidence in it?

Personalized scores are much more useful and meaningful than an average score or a star rating.  For example, a can of chicken soup or a frozen entrée is going to score very differently for a young woman training for a half-marathon than it is for a 70-year-old man managing his blood pressure and heart health.

Our scoring mechanism was developed by a panel of expert registered dietitians, and it’s based on the recommendations of FDA, Institute of a Medicine and USDA Dietary Guidelines and Recommended Dietary Allowances, American Heart Association guidelines, and Academy of Nutrition and Dietetics guidelines for dietary management of chronic diseases. In fact, we were awarded a US Patent (8,647,121) on our scoring method; and the USDA recently accepted ShopWell as one of its MyPlate National Strategic Partners.

It works like this: A baseline score is created based on a shopper’s age and gender.  Then, ingredients and nutrient content move the score up or down based on how they impact the user’s stated dietary goals. Exactly how much the score moves is derived from the expertise of registered dietitians using sophisticated regression analyses. Some attributes, like “added sugar,” are binary: If there are any added sugar ingredients, the algorithm will bump down the score by a specific number of points. Other attributes, like fiber and protein, are dose-dependent, so the more there is the more significant the positive impact on the score. Vitamins and minerals improve the score based on the % of daily value met with that product. The higher the concentration of the vitamin or mineral, the more positive the impact on the score.

Shoppers see a numerical score, and also a traffic-signal color overlay that helps simplify decision-making by grouping scores into “strong match,” “medium match,” and “weak match” foods. Foods with your allergens are flagged as “avoid.” The app also explains in detail why the product scored the way it did – highlighting ingredients and nutritional values. And, because it recommends a healthier alternative in the same category, the score does something that’s easily actionable – rather than a nice-to-know.

How can retailers benefit from this type of app and the information it generates?  

ShopWell is squarely positioned at the intersection of big data and healthy eating.  Our crowd of 1+ million health-seeking shoppers scan on average 10,000 products a day – 60% in store, the rest in their kitchens – and they search our website for as many again.  

For the 60% of scans that happen in grocery stores, we are able to identify exactly which store they’re in. This is extremely useful data for retailers. We anonymize all the data to protect our users’ privacy – but we are able to identify patterns and create unique sub-segments of shoppers and understand their behavior.  

Retailers and marketers can access our cloud-based tools so they can create a health-profile the shoppers in their stores, evaluate categories to find out if they are aligned with their target shoppers’ diets, and develop new insights into where their target demographic is shopping and what product categories they are considering.

How can retailers translate the insights from this data into actions that will improve the business?

Here are three examples that show how the data and insights can lead to clear action.

Example 1: For a given chain we can identify the makeup of their health-seeking shopper population. Are they over-indexing in gluten-avoidance compared to competitors’ shoppers? Which categories are women over 50 most interested in finding out about?  Having a statistically significant sample size means stores can be sure to merchandize the right items for their shopper base. Even stores that have a loyalty card program often don’t have insight into the dietary goals of their shoppers.

Example 2: In the frozen-entrée category, which products score the best for adult men seeking a heart-healthy diet? If you see that this profile is shopping in your store (or you’re trying to attract this segment) then you need to be sure to carry and market items that are a better fit for them. As you work with suppliers – or plan a private label line – this information is crucial.

Example 3: You want to target the millennial that’s interested in a gluten-free diet – where does she prefer to shop?  What categories is she considering or unsure about? Which brands is she showing a preference for? 

There’s a lot of debate about whether shoppers are really bothering to use mobile devices in store when they’re busy.  Who do you see using the app?

ShopWell users tend to be women (76%) who are looking for products that help them meet their dietary goals (esp. weight management, general health, gluten avoidance, and heart health) and avoid allergens. The fact that the app also supports more nuanced goals, such as lactose intolerance, digestive health, and diabetes has also made it really popular amongst shoppers with these kind of long-tail needs. One of our most startling insights was that 25% of our in-store scans are happening in Walmart stores.  The map of scans looks exactly like a map of the US population density.