Behind the scenes: How tech and big data enable us to personalize the meal kit experience

Helping all kinds of customers find something tasty and exciting to cook has always been crucial to our success. Each week we offer our customers around the world between 25 and 40 carefully curated recipes to choose from. By doing so, we provide them with a high amount of variety for their weeknight dinners, while taking their needs, diets and wishes into close consideration. 

Behind the scenes we’ve built a large engine; a best-in-class technology infrastructure and data science & analytics capability that helps us improve our customers’ cooking and eating experience by improving existing and creating new recipes. Based on these recipes we then create menus that target different customer segments and satisfy diverse consumer needs. 

How does it work? Interacting with our customers on a multitude of channels is key - hence, customer feedback, both direct and behavioral, is essential to personalizing the meal kit experience. We then act on the feedback that our customers provide us with every week through all channels by analyzing and evaluating the quality of our ingredients and the process of cooking one of our dinners. We use the learnings to develop new recipes, improve existing ones or to source new or better ingredients, all to continuously improve the customer experience. 

There are four components that are essential to the success of how we personalize the experience across all of our brands: 

1. We receive millions of customer feedback points 

Every month we receive, analyze and evaluate around 3 million comments and ratings from customers around the world. Language processing and text analysis technology helps us analyze them anonymously. Based on the feedback’s sentiment, it’s sorted and shared either with our recipe development team, our menu planning teams or with our procurement teams. Our recipe development teams have weekly review sessions and identify the best and worst performing recipes, empowered by the data we generate. They then incorporate the accumulated feedback into the development or improvement of the recipes. If negative feedback points to an operational issue, our procurement teams will swing into action and collaborate closely with our suppliers. The main objective is to continuously improve our product and offer our customers an even better solution to everyday cooking. 

2. The feedback is personalized 

Over the past years we have built a best-in-class system that helps us to pinpoint the feedback to the exact recipe, which suppliers the ingredients came from and which recipe developer created the recipe. Thanks to the text analysis technology we use, we are able to understand whether a recipe contained too much or too little sauce, was too spicy or too bland. We can see not only what our loyal customers bought, but also what they thought about each recipe that they shared their feedback on. That being said, we always operate in line with all GDPR rules and regulations. We see this as a significant advantage compared to the likes of traditional grocery stores, who normally aren’t able to collect their customers' feedback as precisely as we are. 

3. Feedback from multiple geographies

We significantly benefit from the fact that the feedback we receive is multinational. Looking and evaluating the feedback from a centralized lens provides us with a very unique opportunity. Not only do we gain great insights into the world’s eating habits and preferences, we’re also able to share best practices and learnings across all of our markets and can identify quickly which meals drive uptake and repeat purchases. 

4. Recipe Sorting Algorithms 

As we learn about our customers’ preferences which they either tell us directly via pre-selected preferences or which we infer through the behavioral data they unveil while surfing our menus, we use that information to sort recipes according to our understanding of an individuals’ taste. By that we make their experience highly personal and increase the relevance of recipes they see on our menu. We are in the initial stages of this but early results are very promising and allow us to serve our customers better, in particular as we expand the meal options available on the menu to avoid cognitive overload. 

We’re only at the very beginning on our journey to offering our customers the most personalized dinner time experience and we have many products and algorithms in our pipeline to come. For now, all of the feedback data helps to customize our product, preferences, add-ons, and features to meet the needs and wants of our customers. 

I will be covering more unique ways in which technology helps us to improve the customer experience over the next few weeks. It’s amazing how technology and personalization can help us to come one step closer to our mission “to change the way people eat forever”