The Last Tray Is Not a Forecast: How Too Good To Go Built a Market for Surplus Food

At 5:30 in the afternoon, a baker faces a small but expensive uncertainty. The last tray may sell before closing. It may also become tomorrow’s waste.

Too Good To Go built a marketplace around that narrow window. Partner stores offer assorted surplus food for collection; customers reserve through an app and arrive during a set pickup period. The merchant recovers some value from products that are still safe to eat but unlikely to sell through the normal counter. The buyer accepts less choice in exchange for a lower price.

That trade sounds simple. Operating it across many cities is not. Surplus is local, perishable and uncertain. A marketplace cannot move a bakery item across a continent or promise exactly what will remain at closing. It must attract enough nearby buyers, make collection convenient and help merchants list realistic quantities without training them to overproduce.

Scale came from matching two inconveniences

Too Good To Go says it now has 120 million registered users and 180,000 active business partners across 21 countries. It also reports that more than 500 million meals have been saved since its 2016 launch. These are company-reported cumulative figures, not independently audited measures of meals that would otherwise certainly have been discarded.

Still, the reach illustrates a useful platform principle. The business did not create surplus food or bargain-seeking consumers. It reduced the coordination cost between them. Stores gain a timed channel for variable inventory; customers gain discovery, payment and pickup information in one place. Density matters more than global reach at the moment of exchange: one bakery needs buyers within a practical collection radius tonight.

The model also reduces the information burden by selling an assortment rather than a precisely selected basket. A shop does not need to predict at midday whether two croissants, a sandwich or three rolls will remain. It can assemble the available items later. The customer, however, carries part of that uncertainty. Dietary restrictions, allergens, preferences and perceived value must therefore be handled clearly.

Recovered revenue is not the same as recovered profit

Too Good To Go describes its offer as helping businesses unlock revenue from surplus. That is commercially plausible, but the correct measure for a partner is contribution after every extra cost: packaging, staff time, platform charges, payment processing, customer service and the risk of disrupting full-price sales.

A bakery should compare the programme with a credible baseline, not with a retail price that was never going to be collected. If a surprise bag brings in $6 but costs $1.20 to package and prepare, the recovered contribution is not $6. Equally, if the offer causes a regular customer to wait for a discount, some revenue may have shifted rather than been created.

The operational test is straightforward. Record how many units were produced, sold normally, discounted, redistributed, donated and discarded before joining. Then track the same measures weekly. A useful surplus channel should reduce disposal and improve contribution without increasing production merely to maintain listing availability.

Prevention remains the first option

The United States Environmental Protection Agency places prevention at the top of its Wasted Food Scale. Donation and upcycling are also among the most preferred pathways, while landfill, incineration and disposal down the drain sit near the bottom. A discounted-sale platform occupies a valuable space because edible food remains in the human food supply chain. It does not make excess production desirable.

This distinction protects both the social purpose and the unit economics. Forecasting, smaller batches, better ordering, dynamic production and improved storage can avoid the cost of surplus altogether. Resale handles the residual that remains after good management. Donation may be more appropriate where community access is the priority and legal, food-safety and logistics arrangements support it.

The environmental context is substantial. UNEP estimated that 1.05 billion tonnes of food waste were generated at household, food-service and retail levels in 2022, including inedible parts. Food service accounted for 28% and retail for 12%. Those global estimates describe the size of the system problem; they do not prove that any single app prevents a corresponding quantity.

Regulation is turning food waste into an operating issue

The European Union’s 2025 amendment to the Waste Framework Directive introduced binding national targets for 2030: a 10% reduction in food processing and manufacturing, and a 30% per-capita reduction across retail, restaurants, food services and households. Progress is measured against the annual average generated between 2021 and 2023, subject to specified alternatives.

The targets apply to Member States, not as a simple identical quota handed to every café. National implementation can nevertheless increase measurement, prevention and redistribution expectations across the supply chain. For digital platforms and food businesses, credible records may become as important as a large headline count.

A merchant should retain its own evidence: quantity listed, quantity collected, cancellations, refunds, disposal after failed collection and any product excluded for safety. Without a baseline and clear definitions, “meals saved” is a communications number rather than a management measure.

Food safety and trust are part of the product

A platform can match supply and demand, but the food business remains responsible for safe handling under the rules that apply where it operates. Time and temperature controls, allergen communication, packaging integrity, hygiene, traceability and withdrawal procedures do not disappear because the product is discounted.

The mystery element should never extend to safety-critical information. A customer may accept an unknown flavour; a person with an allergy cannot accept unknown risk. Partners need a process for identifying unsuitable products, responding to dietary questions and cancelling inventory that is no longer safe or available.

What smaller marketplaces can copy

The transferable lesson is not to copy a global app feature by feature. It is to design around a recurring mismatch. Too Good To Go’s mismatch is time: good food remains when ordinary demand ends. A local marketplace might find a similar mismatch in imperfect produce, short production runs, cancelled bookings, unused delivery capacity or off-cut materials.

Start with one category, one geography and one service promise. Measure supply reliability before spending heavily on customer acquisition. Define who carries inventory risk, who sets the price, when payment is released, what happens after a failed pickup and how complaints are resolved. A purpose-led claim attracts attention; dependable operations retain both sides.

Most importantly, publish measures that cannot be improved by making the original problem worse. If success is only the number of discounted bags sold, a partner can raise the number by producing excess stock. Better indicators combine normal sell-through, avoided disposal, recovered contribution, repeat participation and customer satisfaction.

Too Good To Go demonstrates that waste can become inventory when timing, price and local demand are coordinated. Its deeper business lesson is more disciplined: a social enterprise earns trust when the commercial mechanism rewards less waste—not simply more activity on the platform.