Restaurant food waste: measure the cause and use AI as support

Two kitchens may discard the same weight of food but need different solutions. One prepares too much lunch. In the other, guests leave part of a side dish. Start with the cause before choosing technology or setting a percentage target.

Measure waste where it occurs

Distinguish kitchen, serving and plate waste. Record weight, date and what the food was. Note causes such as excess production, incorrect storage or leftovers after serving. Separate normally inedible material, such as bones and shells, to keep the definition consistent. The restaurant handbook from IVL and the Swedish Food Agency supports this work.

Make recording practical: labelled containers, an accessible scale and an owner on each shift. Use measurement to improve work, rather than blame the person reporting.

Compare weight and guests served

In a hypothetical example, 12 kg of waste for 200 guests equals 60 grams per guest. On the next comparable day, 10 kg for 250 guests equals 40 grams per guest. Total waste fell by approximately 17%, while waste per guest fell by approximately 33%. The measures describe different things.

Also track what is discarded. A kilogram of potatoes and a kilogram of expensive fish have different ingredient costs. A percentage of food weight therefore cannot simply be treated as the same percentage of purchasing cost.

Connect measurement to an action

For recurring overproduction, reduce the first batch and decide when to start the next. For plate waste, investigate quantity, flavour, temperature and meal format before changing portion size. For unused ingredients, review purchasing quantities, delivery days and whether the menu can use the ingredient across several dishes.

Sales history shows what was sold. Add kitchen measurements, reservations and a physical stock check for better evidence. Lower waste is not an improvement if guests could no longer order the food they wanted; track unavailable dishes and service too.

What AI can contribute

AI can help summarise available records and suggest questions to investigate. Specialist waste systems may also collect their own measurements through equipment such as scales. This provides different evidence from POS sales alone.

In September 2022, Ingka Group reported a 54% reduction in production food waste across its IKEA stores against a 2017 baseline, measured to July 2022. The work combined Winnow’s AI technology, measurement and trained employees. This is a dated result from that business, rather than a general AI effect or a Vendion result.

Vendion Analytics provides sales and product-mix information, while AI Boss can answer questions about available records. This does not mean the system automatically knows ingredient weights, shelf life or plate-waste quantities.

Reduce waste while maintaining safety

Plan ingredient use around storage, dates and safe handling. Best-before and use-by dates mean different things. Businesses remain responsible for the safety of food used after its best-before date; food carrying a use-by date must not be sold, donated or used after that date. See the Swedish Food Agency’s guidance.

Leftover food is not automatically suitable for service the next day. Follow kitchen handling and safety procedures, even when AI suggests a use.

Choose one recurring cause of waste, test a change and measure again under comparable conditions. That shows what the work actually achieved.