AMLD2024 - AI + Retail Insights

We organized and ran the “AI and Retail” track at AMLD 2024. Here are our insights:

D ONE has completed a plethora of projects in the Retail space with various clients, projects we’re quite fond of. They have included tasks such as AI supported shelf space optimization, self-service BI, and building supporting data architecture.

Data Driven Value Creation is a profound interest of the Retail industry in which players are creating competitive advantages with innovation, which makes it highly interesting.

That experience and joy is why we organized the AI and Retail track - and condensed it for you. Enjoy!

Recommendations @ Migros

It is difficult to recommend new products to supermarket customers, since they tend to stick to their usual purchases. General recommendations like “popular products” do not work, because they are irrelevant to a specific customer more often than not. So what does?

Max Nolte from Migros demonstrated the importance of collaborative filtering to identify and recommend personalized, relevant new products which customers like. This works by method of filtering out products from possible recommendation lists based on the purchasing behaviour of similar customers.

The takeaway Max stressed most was that to be successful, he needed a well-rounded product team. This plus great teamwork, a solid data foundation, processing and data protection standards made the production-ready recommender system a success.

Making clothes fit @ Zalando

The top barrier to online shopping is concerns about product fit. At Zalando, ⅓ of all returns are size related. How do you change that?

Sahan Ayvaz from Zalando is using state of the art computer vision technology to identify people’s silhouettes and calculate the right body measurements to help customers find the right garment size at first try. What do users have to do? Take two photos of themselves.

It was amazing to see how this innovative AI shopping experience boosts customer satisfaction and helps to reduce return rates. We also learned how important the right architecture is and the pivotal role automatic deployments play.

Shelf optimization @ Avolta

What is the optimal assortment in any supermarket? Do we need more pasta? More gin? Is the beauty section too big?

Rubén Pertusa and Pau Sempere talked about AI assortment optimization at scale. Their task was to build a system capable of performing custom recommendations for 5500 points of sale spread over 75 countries. Their solution predicts customer profiles based on the location of the shop and automatically recommends appropriate products.

They emphasized the need for a central data platform that supports scalability, real-time capabilities and supports AI, as well as how important automation and modeling was.

Using this pipeline, Avolta optimizes the assortment of duty-free stores worldwide and increases revenues by having the right product available at the right location.

Reducing food waste @ Freshflow

Up to 10% of a supermarket’s fresh produce ends up in the trash. That’s terrible for the environment and just as bad for their profit margins. Turns out, ordering the perfect amount of food each day is hard.

Avik Mukhija from Freshflow showcased how AI can help the environment as well as have a massive positive economic impact for grocery retailers. Using AI forecasts to improve the supply chain for fruit and vegetables, Freshflow has been used to reduce food waste by 22%, which increased revenues by 2% and reduced costs overall. They achieve this by looking at inventory, weather data, historical sales data, shrink, warehouse size and more.

Conclusion: AI models are much more accurate when ordering the perfect amount of produce every day compared to manual ordering, which is better for the environment and better for profits.

Becoming Data-Driven @ Valora

A clear business strategy is a key driver for data-driven value creation. Daniel Habermehl demonstrated how Valora embarks on the AI journey. The focus is on core processes and combining business understanding with data science and user centricity. This enables data-driven decision making at Valora which is well aligned with the business strategy, such as advanced pricing capabilities or forecasting and replenishment.

Innovation @ ETH AI Center

The ETH AI Center also runs several AI projects in the Retail sector - Klaus Fuchs showed how they foster exciting new startups in the Swiss AI space. One example is their study on consumer product identification, where they tracked consumer head movements to learn which products attract attention with the help of VR headsets and then nudged the consumer towards a healthier option. They also support the start-up FoodCoach, a digital Receipt-based Diet Monitoring & Interventions App, using Graph Analytics Techniques.

Find the full talk here:

For any questions, don't hesitate to reach out to us!