How Lidl forecasts perishables by weather and calendar (and the version that fits your chain)

Lidl decides each store’s perishables order against two things that change every day: the weather and the calendar. The system behind it is simpler than the scale suggests, and more expensive than it looks, for a reason that is not in the model.

Lidl is part of the Schwarz Group, the largest retailer in Europe by revenue, per Deloitte’s Global Powers of Retailing (2025). In hard discount, perishables are where the thin margin meets its biggest enemy: shrink. Every point of loss in produce eats the profit the low price already left tight.

What the system does for the operation

It shortens the distance between what the store orders and what the store will sell tomorrow. In fresh, ordering too high becomes garbage. Ordering too low becomes an empty shelf at rush hour.

The order goes out adjusted by a forecast at the store and day level, not by a regional average. A store next to the park on a sunny Sunday does not order like the downtown store.

What the public record supports

Nobody outside has seen the architecture from the inside, and we will not pretend we have. What the public record supports is the shape: a fresh forecast that uses a weather signal and the commercial calendar as first-order inputs.

This practice has been documented in European grocery for years. Temperature moves salad, barbecue meat, ice cream. Holidays and paydays move volume. The group’s merit is doing this at store granularity, every day, at the scale of Europe.

The non-obvious part

Weather does not enter as “it will be sunny”. It enters as a deviation from what is normal for that week in that location. Thirty degrees in October moves the basket. Thirty degrees in January is just a Tuesday. The model that works does not try to predict the weather. It reacts to the weather’s deviation from local expectation.

The calendar is not only national holidays either. It is payday, the competitor’s market day, the city event. Carry only national holidays and you get Christmas right and miss every Friday.

The real cost, and where it sits

The model layer is the cheap part. The expensive part is the wiring. Sales by store, day, and SKU, clean. Shelf-out flagged in the history. Integration with the automated order. Discipline in the perishables registry, where the same tomato tends to have three codes.

It is people and months, not a weekend. And there is the process temper: the manager has to trust the order to stop adjusting it by hand. While they “correct” it daily, the model never learns.

The version that fits your chain

You do not have the Schwarz Group, and you do not need it. The version at 5 to 10 percent of the scale does not try to forecast Europe.

Take one location, three fresh categories (the one that shrinks most, the one that runs out most, the one that sells most) and two inputs: local temperature deviation and a hand-built city event calendar. Measure shrink and shelf-out by day, against the baseline week. This runs in a pilot store in a few weeks.

The big model is a problem for those with a thousand stores. The criterion is the same at ten or at a thousand: the order only promises what sales within shelf life confirm. It is the same reading that misleads in stock cover, and the same platform, build, or implement decision almost every chain postpones.

Think about the fresh category that hurts most at month end. Do you know how much of it becomes expiry loss, by store, by day of week? If the answer lives in the manager’s head and not in a number, that is the project.

Tell me the operational result your operation wants, the specific number in the specific location: fresh shrink at 3%, shelf-out at 2%. In one hour I will send back a sketch of the architecture sized for your case, with a build or buy marker on each piece. It is the kind of thing we deliver. If the shape confirms, the two-week Diagnóstico becomes a signed spec: the paths, the costs, the dependencies, and the acceptance criterion for each component. That document becomes the contract for the Implementação.