Reducing returns in the bakery—using data instead of gut feelings

For many bakeries, returns are a daily occurrence. At the end of the day, there are loaves of bread, croissants, and pastries left over that can no longer be sold. What’s left is either thrown away or, at best, donated. Either way, it’s a loss—of raw materials, labor, and profit margin.

Most businesses are aware of their returns problem. But solving it is harder than it sounds. That’s because the root causes lie not in the craftsmanship, but in the planning.

The True Cost of Returns

The direct costs are obvious: ingredients, energy, and labor that went into a product that nobody buys. But the indirect costs are often greater.

If you regularly produce too much, you get used to it. Returns become “normal shrinkage,” which is factored into the cost calculations. This leads to higher retail prices or lower margins. On top of that, there’s the time involved: returns must be recorded, sorted, and disposed of. In in-store bakeries, this quickly adds up to several hours a week.

And then there’s the image factor. Customers who come in just before closing time and see full shelves might be pleased with the selection. But as a business, you pay a high price for it.

Why it's so hard to reduce returns

Demand for baked goods is unpredictable. At least, that’s how it feels to businesses. A rainy day, school holidays, a local market, or simply a Wednesday instead of a Thursday: all of these factors influence how much of each product is sold.

Most bakeries respond to this uncertainty by building in a safety margin. Better to have a few too many rolls than an empty shelf. This is understandable, but it systematically leads to overproduction.

The other classic approach: empirical evidence. "Business is slower on Mondays, busier on Fridays." That’s true on average, but it’s not much help when a sunny Friday in March plays out completely differently from a rainy Friday in October, and the store at the train station follows entirely different patterns than the one in the neighborhood.

The fundamental problem is this: It’s simply not feasible for a person to calculate the optimal quantity for hundreds of items across multiple locations every day while taking into account the weather, holidays, events, and seasonal trends all at once. It’s just too complex.

How Data-Driven Forecasts Reduce Returns

This is exactly where AI-based sales forecasts come into play. Instead of relying on experience and safety margins, the AI automatically analyzes the actual quantities needed on each day at each location.

To do this, GoNina directly to the bakery’s point-of-sale system—such as HS-Soft, ProtecData, or Lightspeed—and takes into account historical sales data, weather forecasts, holidays, school breaks, and local events. Based on this data, the AI generates a daily sales forecast for each product and location.

The result is a specific order proposal: not “roughly the same amount as last week,” but a data-driven recommendation that adjusts daily to the current situation. The company can accept the proposal as is or adjust it manually.

The key difference from gut instinct: AI takes into account dozens of factors at once that a person cannot systematically evaluate in everyday life. It recognizes that on a sunny Wednesday after spring break, Branch 3 will see higher demand for sandwiches and lower demand for farmhouse bread, and adjusts its recommendations accordingly.

What's changing in practice

Bakeries that GoNina report several benefits:

Less waste at the end of the day. The most obvious change: there’s less left over at the end of the day. Businesses are achieving a reduction in surplus of up to 52%. That means fewer returns, less food waste, and lower material costs.

Fewer emotional debates. The question "How much should we bake tomorrow?" is often debated among production staff, store managers, and executive management. With a data-driven recommendation, this discussion becomes more objective and shorter.

Higher sales despite lower production. That may sound paradoxical, but it makes sense: When forecasting is more accurate, the right products are in stock more often. Customers find what they’re looking for instead of settling for whatever’s left or leaving the store without making a purchase. As a result, businesses see sales increase by up to 6%.

A better basis for making decisions about the product assortment. The forecast data also shows which products are in constant demand and which are rarely sold. This helps with assortment decisions: Is it worth continuing to stock this specialty pastry every day, or is three days a week sufficient?

Frequently Asked Questions

How quickly will I see an impact on returns? Most businesses notice a noticeable change within the first few weeks. The AI learns from sales data and becomes more accurate over time. The full effect typically becomes apparent after two to three months.

Can I override the suggestions? Yes . GoNina recommendations, not requirements. The business can manually adjust any suggestion—for example, if a special promotion is planned or if a product needs to be highlighted more prominently.

What if I’d rather produce a little more? That’s a conscious decision that GoNina . The forecast shows the expected demand. Whether you produce slightly more or slightly less than that is up to you. But it will be data-driven rather than arbitrary.

Conclusion

Returns in the bakery are not an inevitable evil. They result from planning uncertainty, which can be significantly reduced with the right data. AI-based sales forecasts do not replace gut instinct, but they supplement it with a data foundation that no human can provide in day-to-day operations.

For those who want to reduce waste without losing revenue, data-driven forecasts offer a concrete solution.

To learn more about how AI forecasts can simplify the entire production planning process, check out our guide to AI sales forecasting for bakeries.