Germany’s retailers take smart stores to scale

German retailers are moving AI and automation beyond technology trials as self-checkout, electronic shelf labels and operational AI become established parts of the physical store. But smart carts, cashierless shopping and robotics still have to prove they can deliver value at scale.

One in 10 checkouts in German food retail is now a self-checkout, twice the proportion recorded two years ago, while electronic shelf labels (ESLs) are used by 91% of retailers, according to KPMG’s 2026 Retail Sales Monitor.

Alongside these established technologies, retailers are testing cashierless shopping, smart carts and autonomous food preparation. AI is also moving behind the scenes into assortment planning, space optimisation and inventory management.

The result is a new phase of store digitalisation. For retailers, the question is increasingly not whether AI and automation work, but where they can improve productivity, availability and customer experience enough to justify deployment across large store estates.

AI moves behind the scenes

Some of the most significant uses of AI in German retail are largely invisible to customers.

REWE Group says it uses analytical AI for space optimisation across approximately 3,400 REWE stores and for assortment optimisation across its REWE and Nahkauf formats. The technology is intended to make better use of selling space and adapt local assortments to customer demand.

REWE began using AI for personalisation and assortment optimisation in 2018 and expanded its use across analytics from 2021. It also uses anonymised sales and customer behaviour data to support decisions around product availability, inventory and waste.

While generative AI and customer-facing applications attract much of the attention, these less visible applications show how AI can support day-to-day retail decisions.

Store managers can use data to determine how much space to allocate to a category, which products to stock locally and where replenishment needs attention.

The effectiveness of such systems, however, depends on the quality of the data they use. Incomplete or inconsistent inventory, product and sales information can limit what AI can optimise, making data integration a fundamental part of the smart-store proposition.

Self-checkout reaches the mainstream

Checkout provides some of the clearest evidence that store automation is moving beyond experimentation.

KPMG reports that self-checkout now accounts for one in 10 checkouts in German food retail, after doubling in two years.

But checkout automation is taking different forms. Alongside conventional self-checkout terminals, retailers are introducing mobile scanning, automated product recognition and cashierless formats.

REWE opened its seventh Pick&Go test store in Hanover in March 2026. Customers can use a scanless option in which products are recorded automatically rather than unpacked and scanned at a conventional checkout.

The retailer has also tested the concept in convenience retail. In February 2025, REWE subsidiary Lekkerland and airport operator Fraport opened an unmanned REWE To Go pilot at Frankfurt Airport. Cameras identify selected products, with purchases automatically charged to the payment method registered when the customer enters.

The deployments illustrate the different stages of checkout automation. Conventional self-checkout is already widely deployed, while fully cashierless formats remain experimental.

Smart carts turn the trolley into a digital channel

The shopping trolley is also emerging as an interface between physical and digital retail.

Smart carts combine screens, sensors and cameras with functions such as digital shopping lists, navigation, product information and payment. Their screens can also provide retail media inventory, allowing retailers and brands to reach shoppers close to the point of purchase.

Research published in the Journal of Business Research offers an indication of the technology’s commercial potential.

Researchers analysed 12,418 shopping sessions at a major German supermarket, including 9,422 in which customers actively used smart-cart functionality.

Smart-cart users spent an average of €30.18, compared with €22.89 among non-users. They bought 12.02 items on average rather than 9.61 and spent 40.40 minutes in the store compared with 32.75 minutes.

That represents a 32% difference in average spend, 25% more items and 23% longer in store.

The findings do not show that smart carts caused those increases. The study was observational, and customers who chose to use the technology may differ from non-users. The researchers instead found an association between smart-cart use and higher spending, larger baskets and longer visits.

For retailers, the findings highlight the trolley’s potential as a digital touchpoint combining customer assistance, merchandising, navigation, loyalty and payment.

The question is whether those benefits justify the hardware, software, maintenance and integration costs of deploying smart carts across a store estate.

Electronic shelf labels become infrastructure

Other smart-store technologies are already much further along the adoption curve.

Electronic shelf labels are used by 91% of German retailers covered by KPMG’s latest research.

Their role increasingly extends beyond replacing paper price labels. ESLs allow retailers to manage price and promotional information centrally, reduce manual label changes and connect the shelf edge with other retail systems.

German specialist retailer Kölle Zoo provides one example. Around 300,000 electronic shelf labels were installed across 28 stores during an eight-week rollout in 2026. The retailer said the system reduced manual work associated with price and promotional changes, allowing employees to spend more time on customer-facing activities.

Integrated with pricing, inventory and promotional systems, ESLs can therefore become part of a connected store rather than simply digital replacements for paper labels.

Automation reaches the supermarket kitchen

Not every application is as mature.

In October 2025, REWE Region West launched its Fresh & Smart pilot with German technology company Circus Group, using an autonomous robot to prepare meals inside supermarkets.

The first deployment in Düsseldorf-Heerdt formed part of a planned three-store pilot. Customers order at a digital terminal before the robot prepares the selected dish, while AI-supported demand forecasting helps manage supply and waste.

The potential extends beyond the novelty of robotic food preparation. Retail catering requires labour, food-safety controls, forecasting and consistent production. Automation could allow retailers to provide freshly prepared food across longer trading periods while standardising some of those processes.

But the economics remain unproven.

Equipment and maintenance costs, throughput, reliability, customer demand and product range will determine whether autonomous food preparation can move beyond selected locations.

For now, the REWE deployment is a live test rather than evidence that robotic kitchens are ready for widespread supermarket adoption.

The smart store faces the scale test

The bigger opportunity – and challenge – is connecting these technologies.

Electronic shelf labels, checkout systems, smart carts, inventory technology and AI software all generate or use data. Their value can increase when that information feeds into common retail processes.

Demand detected by an inventory system can inform replenishment and assortment decisions. Digital shelf infrastructure can connect centrally managed pricing with the physical store. Smart carts can bring digital services into the shopping journey.

But connecting hundreds or thousands of stores to common data, software and operational processes, while maintaining reliability and security, is considerably harder than running a successful technology pilot.

Different technologies also require different measures of success.

Checkout automation can target transaction efficiency and queues. AI-based space planning can improve the use of selling space. ESLs can reduce manual price administration. Smart carts can connect customer services, merchandising and payment. Robotics can automate selected repetitive production processes.

Some are already reaching significant adoption. Others are generating promising evidence, while cashierless stores and autonomous food preparation remain at an earlier stage.

The challenge for retailers is therefore to define the business problem first and choose the technology second.

Automation changes the role of store staff

Labour is central to that calculation.

Automation can reduce repetitive work such as changing paper price labels, scanning purchases or monitoring routine processes. That does not necessarily mean removing employees from stores.

REWE says its approach is to use AI to support employees and improve processes. Kölle Zoo similarly says time saved through electronic shelf labels can be redirected towards customer advice and service.

For retailers facing recruitment and cost pressures, the workforce case for smart-store technology may therefore be as much about changing how employee time is used as reducing headcount.

Technology that removes repetitive administration while preserving employees for customer service, replenishment and exception handling may ultimately offer a clearer business case than automation pursued for its own sake.

From smart devices to smart stores

Germany illustrates how physical retail automation is developing at different speeds.

Self-checkout and electronic shelf labels have already achieved significant adoption. Operational AI is being used across large store estates. Smart carts are producing real-world behavioural data, while cashierless stores and autonomous food preparation remain at an earlier stage.

The next phase is unlikely to be defined by a single breakthrough device.

Instead, the smart store will depend on how effectively retailers connect AI, checkout, shelf-edge technology, inventory systems and customer-facing devices around specific operational needs.

For retailers, the question is shifting from what can be automated to what is worth automating at scale.