When we used AI to turn compliance into a 2⁠-⁠min job

The challenge of compliance at scale

Bar managers, shop owners, and restaurant operators often have to deal with multiple things at a time. One of them is making sure their fridges meet Heineken's branding, placement, and presentation standards, but with so many things going on, this was rarely a shop owner’s top priority. To ensure these requirements were met, our operation often relied on visits from sales reps, who had to invest time and effort every day in order to cover our customer base in massive markets like Brazil and Mexico.

What did this mean for all involved?

As a result, shop owners would have products mixed up inside the wrong fridges, sales reps had to spend most of their time focusing on compliance related tasks, and Heineken products wouldn’t have the chance to communicate their brand identities.

We felt like there was something more we could do to simplify compliance, strengthen Heineken’s brand positioning and allow our users to focus on what matters most to them: running their businesses.

Could our users perform these checks?

Monitoring fridge compliance across thousands of outlets, however, is a manual, time-consuming process. Sales reps spent significant time on compliance checks rather than having strategic conversations with our business owners, creating a high reliance on them and a lack of autonomy. The goal was to give outlet owners the ability to perform these checks themselves, in real time, from their phone, saving time for both sides. In exchange we would reward their effort with loyalty points for them to redeem from a range of different prizes, products and promotional merchandise for their businesses.

To allow for this, we partnered with an AI powered image recognition solution, that would allow users to take pictures of their fridges and assess whether they aligned with our standards for compliance.

Taking compliance into their own hands

The main journey would have users take a picture of their closed fridge, and a subsequent photo inside it. These pictures would be run against Heineken's Picture of Success guidelines, showing exactly what products were missing and giving users a direct path to restock.

Business owners would then receive an amount of points proportional to their compliance score for each scan. Heineken in return, would earn real-time visibility into whether commercial and brand standards were being met across their outlet base, without relying solely on field visits.

It was important for the AI fridge scanner to be embedded as a web app inside the existing eazle platform, removing any need for a separate app install, since device storage and willingness to download could be real barriers in markets like Brazil and Mexico.

Defining our design foundations

As the design of our solution was starting to take shape, it was important to cover three main bases so we could establish a dynamic that was easy to adopt from scratch.

1. Clear communication was key

Since we were asking our users to perform tasks they had never done before, it was very important for me to provide our users with clear instructions and copy throughout the journey. Communicating clear and simple progressive steps, one action at a time would guide users through a flow that could otherwise feel pretty overwhelming.

I wanted to avoid lengthy tutorials and provide bitesize information in the right context instead. Display the right alerts to bring attention to the type of picture they needed to capture while their phone camera was on and ready to shoot, instead of overloading them with long instructions in the previous step, only for them to forget everything once they opened their camera.

3. It was crucial to test it on site

Testing the solution in a real scenario would ensure we had not overlooked the nuances of dealing with users from two markets with a different language, degree of digital literacy and potential technological constraints. So I worked with our CX Growth team to build an interview guide for on-site usability testing in Rio de Janeiro, carried out by a local agency with real shop owners, inside their own shops, on their own devices.

My intention going into testing was not only about journey completion, but rather to understand whether our users found genuine value from this dynamic: if they didn’t feel motivated to scan their fridges in exchange for a reward that got them excited, why do it in the first place? Testing taught us that we still had some fixes to tackle for the full journey to feel more intuitive, by keeping our copy sharper and removing distraction to allow our users to move on by focusing on one task per screen and taking one step at a time.

Real value for the three parts of the equation

We completed the full design journey, from concept, to testing, to live product in 2.5 months. The AI Fridge Scanner launched in Brazil and Mexico in June 2026 allowing shop owners to turn compliance into a 2 minute job. They got a full picture of what was missing in their fridges while making the most of our reward program. As for Heineken, we gained timely and accurate data on whether commercial and brand standards were met across our outlet base, at a scale field visits alone couldn't reach.