Kum & Go presenting at Analytics Unite 2023

Data and analytics lead Kum & Go shoppers to healthier food choices

Kum & Go uses first-party loyalty data and A/B testing to modernize food offerings.
Elizabeth Christenson
Editor, Retail Leader
Elizabeth Christenson profile picture

What this means: To keep up with the ever-changing demands of consumers, retailers are investing heavily in store format changes post-pandemic. Because layout and assortment changes are beneficial across fleet, it’s critical to understand how the changes will come to life in different markets and regions, which may impact how consumers move through their shopping journey. Testing design and flow strategies before rolling them out is important to help retailers get things right in today’s chaotic retail environment.

Matt Weber, Kum & Go’s director of business insights and analytics
Matt Weber, Kum & Go’s director of business insights and analytics, at Analytics Unite 2023

Data-driven decision making is key in the retail industry, and A/B testing enables an indispensable way of tying return on investment to different initiatives. Matt Weber, Kum & Go’s director of business insights and analytics and Johnny Stoddard, MarketDial’s chief customer officer, spoke about how the convenience store chain used data and analytics to drive a strategic initiative for a healthy food launch during a Analytics Unite 2023 presentation on “How Data and Analytics Are Leading to Healthier Food Choices.”

Kum & Go, a Midwestern convenience store chain with about 400 stores in 13 states, was recently acquired by Maverik. The retailer began its healthy choices initiative in 2021 by conducting analysis to determine the consumer to which healthy food would appeal. It also leveraged first-party data from its loyalty program to discover if these consumers were already shopping its stores.

Kum & Go partnered with MarketDial to test and understand the success of this initiative. MarketDial designed an A/B test analyzing Kum & Go’s first-party data in which it selected a set of stores to implement the healthy food initiative. The analytics problem statement was: how to design a test where it could reduce bias, but also work within practical business constraints? Some of the constraints were time, speed-to-market, test size, test length and expense. Broadly speaking, the solution was to design a non-biased sample group and then pick a well-matched control group to measure against. MarketDial analyzed data to find a non-biased sample of 30 stores that would represent the different formats of the nearly 400 store fleet.

Kum & Go started with a traditional convenience store food program, and it conducted construction to change the store layout to accommodate made-to-order food along with the new equipment and technology to make and retail it. Assortment changes in center store were also necessary to compliment the new food program. Additionally, a complete app redesign was needed to accommodate made-to-order food retailing.

“The big reason you test is the unknowns like what are the unintended consequences — the things you didn’t anticipate to see,” Weber said.

Johnny Stoddard, MarketDial’s chief customer officer
Johnny Stoddard, MarketDial’s chief customer officer, at Analytics Unite 2023

Kum & Go began with an operational test in one of its Des Moines, Iowa, stores to determine construction plans and equipment. The retailer then began a full brand refresh and store remodel along with the made-to-order and grab-and-go healthy food product launches in its 30 designated market areas for the next 26 weeks.

“One of the great parts with test and control is we didn't just see an impact to the food,” Weber said. “We could measure impact to other categories. For instance, which categories were impacted positively or negatively. And it really gave us some insights into when we go into that next round of what do we need to do?”

For example, cold fountain sales were impacted because the stations were closed for a long time during the remodel. Kum & Go was able then to adjust the remodeling schedule to aid sales. Kum & Go’s loyalty program data showed a huge shift to pizza occurred, so the retailer expanded grab-and-go options in its made-to-order initiative. Signage and equipment also were adjusted based on data.

In the second phase of testing, Kum & Go conducted some ethnographic studies to examine customers’ experiences. What the retailer found was one of the greatest barriers to trial of the new food program was habits and mission-driving shopping. In turn, the retailer created a little bit of friction to motivate consumers to notice the new menu and food program. Kum & Go will be moving into new markets this year and will be optimizing its kitchen and staffing after insights from its trial.

“If every idea that we had worked, then you wouldn’t need to do A/B testing,” Weber said.

Instead, “failing fast” and learning from testing in small numbers leads to better consumer experiences across the entire fleet, he said.

What’s next: Using data and insights to inform what store changes are needed is a great use case for first-party data optimization. The keys to unlocking how consumers move through a store or how they want to shop are within their purchase or behavioral data.