The Surprise Bag: How Too Good To Go Made Surplus Desirable

Most retailers try to remove uncertainty from a purchase. Too Good To Go sells some of it back.

Customers pay for a bag without knowing exactly what it contains. They accept a collection window, surrender some choice, and discover the contents later. For a company selling food approaching the end of its commercial life, the proposition seems counterintuitive. Why make an already time-sensitive product less predictable?

Behind the counter, however, uncertainty is the problem that needs solving. A bakery cannot know exactly which pastries will remain. Listing every leftover item creates work while its value is disappearing. A bag lets the merchant promise an exchange without committing to a precise assortment. The customer receives a discount, perhaps a discovery, and the satisfaction of putting something useful to use.

The format accommodates uncertainty on one side of the market by finding people willing to accept it on the other. That is both an economic solution and a psychological one. But it leads to a harder question: when a company gets better at selling surplus, does it necessarily get better at preventing waste?

From a Copenhagen problem to a retail proposition

Too Good To Go dates its origins to Copenhagen in 2015, when Thomas Bjørn, Stian Olesen, Klaus Bagge Pedersen, Brian Christensen, and Adam Sigbrand sought a solution to food discarded by Danish buffet restaurants. The app launched in early 2016. Entrepreneurs helping establish the business in Norway, Britain, and France joined the wider founding story that year.¹

Mette Lykke arrived as an investor before becoming its chief executive. A co-founder of fitness app Endomondo, she learned about Too Good To Go from a fellow passenger on a bus in 2016, invested that October, and took over as CEO in spring 2017. In a later interview, she identified the passenger as entrepreneur Birgit Aaby.²

The interview also describes a practical way she won over a large retailer. Rather than begin with an abstract pitch to Coop's Danish headquarters, Lykke recruited individual franchisees. By the time head office questioned the model, she could point to 30 outlets already participating.² The demonstration was not a slide predicting interest. It was merchants choosing to try the service.

The appeal to merchants was practical: recover money from unsold food, retain discretion over the contents, and potentially introduce new customers to the store. The company's merchant FAQ describes annual fees and commissions on bags sold, although terms vary.³ The platform earns money by making recovery worthwhile for businesses and collection worthwhile for consumers.

Surprise also creates distance between a clearance purchase and an ordinary one. Someone who needs a particular lunch at noon cannot reliably substitute an unspecified bag available later. Lykke explicitly cited uncertainty over the contents when addressing retailers' fears of cannibalizing regular sales.²

That is documented commercial reasoning. It does not establish that the founders began by applying a named psychological theory. The design solved an operating problem; behavioral research helps explain why customers might accept, or enjoy, the solution.

A bargain with several kinds of value

The discount makes the exchange understandable. When Whole Foods Market and Too Good To Go announced their national rollout in July 2024, covering more than 450 stores, a prepared-food bag cost $9.99 against an advertised $30 retail value. A bakery bag cost $6.99 against $21.⁴

Those reference prices tell customers what they receive for surrendering choice. They do not establish the value of the contents to a particular household. An assortment can meet its advertised retail valuation and still contain more bread than someone can use. The relevant saving also depends on what the customer would otherwise have bought.

Surprise can supply a separate reward. In experiments published in the Journal of Consumer Research, Luxi Shen, Ayelet Fishbach, and Christopher Hsee found that uncertain rewards could motivate greater investment of effort, time, or money than certain rewards with higher expected value. The effect depended on attention to pursuing the reward rather than obtaining the outcome.⁵ These experiments did not examine Too Good To Go. They suggest why hunting for a bag might be entertaining even when planning a dependable dinner is not.

Research involving the app points to several motives rather than one. A 2019 Wageningen study surveyed 611 Dutch users. Asked for their main reason for using it, 35% selected wasting less food, 26% the surprise experience, and 20% saving money. Respondents generally reported little disposal of the contents. The study was partly funded by Too Good To Go and relied on selected users' accounts, rather than measured household waste against a control group.⁶

A German survey of 380 respondents, published in 2025, associated enjoyment with intended app use; price consciousness and sustainable consumption consciousness were not significant predictors in its model. That finding concerns stated intention and associations, not the causal drivers of purchasing.⁷ It cannot establish that price or environmental concern is unimportant.

The motives can coexist. A bargain can provide pleasure and moral satisfaction without requiring a customer to make an environmental sacrifice. The same customer may seek adventure one evening and inexpensive groceries another.

Availability and collection windows shape that choice. Limited supply can make securing a bag satisfying, while an inconvenient pickup can make it impractical. These constraints reflect perishable inventory and merchant operations as well as the app's presentation. Calling every sold-out listing a scarcity tactic would confuse the underlying economics with an inferred intention.

In a 2026 Kiplinger account, Kate Schubel described buying a $10 bakery bag with labeled prices totaling $34.96. She liked the food but did not intend to repeat the purchase: limited choice and a roughly 25-minute drive each way diminished the bargain.⁸ One experience cannot establish typical satisfaction. It shows why a successful transaction may fail to become a useful habit.

Growing the network, changing the business

Too Good To Go's expansion required enough merchants and customers to meet locally, at workable times. Millions of distant registrations cannot make a neighborhood listing useful. Repeated, satisfactory exchanges can.

The company entered the United States in 2020 and reached San Francisco in 2021.⁹ A financing announced in January 2021 raised €25.7 million, with US expansion among its purposes.¹⁰ Capital, retailer partnerships, geographic execution, and local marketplace density offer substantial explanations for growth alongside the appeal of surprise.

Its 2025 Impact Report counted more than 120 million registered users, 200,000 active partners, and approximately 157 million meals saved during the year.¹¹ Registrations are not active customers. The meal figure is an operational measure expressed in meal-equivalents, not an independently established number of meals that would otherwise have been discarded.

By October 2026, the company reported 150 million registered users and 200,000 active partners across 20 countries.¹² Those figures establish reach; they leave open how often customers return and how commercial performance differs across markets.

Its financial statements show a growing company that had not yet reached net profitability. Too Good To Go Holding ApS reported about €249 million in revenue for 2025, up roughly 29% from 2024. A measure of operating earnings before certain costs, adjusted EBITDA, was positive at €15.8 million. After other operating costs and special items, however, the group recorded a €5.4 million operating loss and a €4.8 million net loss.¹³

Operating cash flow was positive at about €26 million, helped in part by working-capital changes. That does not mean the company earned a net profit. Nor does every euro of reported revenue represent the same kind of transaction: Surprise Bags generate merchant fees, while delivered Parcels are accounted for using the full transaction value. The figures reflect more than the growth of the familiar bag marketplace.¹³

The business increasingly reaches beyond the final handover. Its 2022 acquisition of inventory-management company CodaBene helped underpin Too Good To Go Platform.¹⁴ Enterprise software now combines expiry-date management, AI-powered markdowns, marketplace redistribution, and donations. Parcels bring manufacturers' surplus directly to consumers.¹⁵

The familiar bag is evolving too. A Canadian commercial appendix updated in July 2025 describes optional algorithmic pricing that adjusts prices using factors including location, bag type, purchase history, and real-time supply and demand. Its stated objectives include bags sold and partner revenue.¹⁶ This documents analytical capability, not a measured global performance improvement.

The strategic direction is consequential. Too Good To Go can increasingly participate in decisions about how surplus is handled, rather than only connecting a leftover bag with its buyer. Better management upstream might leave less inventory to sell through the app. A business organized around waste reduction needs room for that outcome to count as progress.

What an empty store bin cannot tell us

Luyi Yang and Man Yu's Management Science study supplies a precise warning. Their mathematical model compares surprise clearance, transparent clearance, and no clearance. A predetermined bag price with variable quantities lets a merchant recover value flexibly from uncertain leftovers.¹⁷

In the model, surprise clearance maximizes merchant profit and eliminates store waste. Yet it can generate the most total waste: improved recovery encourages greater production, and additional purchases, including regular-price purchases, can go uneaten. Under other conditions, it produces the best joint outcome for profit, welfare, and waste reduction.¹⁷

These are conditional predictions from a model, not evidence that Too Good To Go actually increased waste across its network. They identify a risk that a transaction counter cannot reveal: food may leave the store without reducing the amount produced, purchased, or thrown away elsewhere.

Recent field evidence provides an important counterweight. A September 2026 working paper by Nymisha Bandi, Maxime Cohen, and Saibal Ray analyzes two years of data from 5,279 US Circle K stores. Using staggered adoption in a difference-in-differences design, the authors estimate roughly 11% lower fresh- and prepared-food waste and approximately 5% higher traffic and revenue after adoption.¹⁸

This is observational causal analysis, dependent on its comparison assumptions, rather than a randomized trial. It examines one chain and store outcomes, not household disposal or total emissions. The authors also report substitution within categories overlapping bag contents, even alongside overall commercial gains. The results strengthen the case that the partnership can work while resisting a blanket conclusion that it never displaces regular purchases.¹⁸

The model and the field study ask different questions. Evidence of lower store waste establishes something valuable. It does not settle what happens throughout production and consumption. Conversely, a theoretical possibility of rebound does not overturn measured improvement in a particular setting.

The work of keeping the promise

Too Good To Go has also had to manage the behavior of the people using its market. Alberto Bertello and colleagues' July 2026 Journal of Business Ethics study examined its Italian expansion through 41 interviews and other material collected during 2019–2022. It describes convenient transactions alongside mechanisms intended to protect the mission, respond to opportunism, and maintain emotional commitment.¹⁹

Managers reported watching unusually persistent bag volumes as possible evidence that merchants were producing for the platform. Ratings helped distinguish participants' conduct from the reputation of the whole service. The researchers also interpreted a multistep review process as encouraging customers to reconsider disappointment. Their historical case, informed by disclosed company collaborations, does not experimentally establish customer benefit.¹⁹

A pause might produce a fairer assessment. It might also discourage a valid complaint. Evaluating such a process requires more than counting how many negative reviews disappear.

Trust makes uncertainty tolerable when the uncertainty concerns which usable products arrive. It becomes fragile when customers cannot rely on promised value, condition, or remedies. The distinction matters for retention: disappointment about receiving an unfamiliar pastry is different from feeling that the bargain itself was misrepresented.

Measuring the thing that matters

The company's environmental claims have a published methodology behind them. A 2024 assessment by Mérieux NutriSciences | Blonk, reviewed by three external experts, estimated approximately 2.7 kilograms of greenhouse-gas emissions avoided per kilogram of rescued food, expressed as carbon-dioxide equivalents. In its public reporting, Too Good To Go treats an average Surprise Bag as one kilogram, or one meal-equivalent.²⁰

The calculation accounts for packaging, collection journeys, and food discarded at home. But it also assumes that the food was genuinely surplus and that 90% of the rescued quantity replaces food that would otherwise have been purchased. Much of the estimated climate benefit comes from the production that is assumed not to happen, not simply from avoiding disposal. Change those assumptions and the estimated benefit changes substantially.²¹

Consumption and displacement are different. Eating an additional pastry does not demonstrate that another purchase, and ultimately its production, was avoided. The assessment estimates an average conditional benefit; it does not directly measure the emissions prevented by every individual bag.

Nor does displaying the estimate guarantee motivation. Brendan Lawson and colleagues followed 10 Oxfordshire households through an app trial and interviews. Participants often attached little value to the carbon figure, and some questioned its consistency with varied bags and collection journeys. This small qualitative study cannot establish a population effect.²² It does challenge the assumption that a quantified environmental message necessarily becomes meaningful feedback.

There is a useful lesson here for other organizations. Too Good To Go made a beneficial action attractive through price, flexibility, and discovery. It reduced the coordination required to participate. The next discipline is to test outcomes beyond the transaction: merchant contribution after costs, purchases displaced, products used, and waste prevented. A product can count its activity accurately while only partly measuring the problem it addresses.

Beyond the bakery

On October 15, 2026, Too Good To Go announced expansion across personal care, pet care, flowers and plants, nutrition and supplements, and home products. More than 8,500 stores already supplied these areas; around one-third were existing food partners. The announcement extended activity already underway.²³

The commercial rationale is recognizable: packaging changes, discontinued ranges, seasonal cycles, and forecast errors strand useful stock. Redistribution can recover value, release space, and introduce customers to products. Examples include Interflora flowers in Denmark and Sweden and Decathlon nutrition products in Italy, with availability varying by market.²³

The transferable advantage is a network for connecting surplus with people willing to accept constraints. How much surprise those people want remains a category-specific question. An unfamiliar bouquet may be welcome; a pet's diet or a skincare ingredient may require greater certainty. These are hypotheses about customer needs, not evidence that the expansion will falter.

Tests of truthful product descriptions could examine which details improve usable value and repeat purchasing: suitability, quantity, condition, or remaining shelf life. Pricing experiments could follow merchant contribution and ordering behavior alongside sell-through. Longer-lived products require checking whether redistribution replaces purchases or simply moves inventory into consumers' cupboards. Food-based impact factors cannot answer those questions for household goods.

The company's existing software and pricing tools show that such questions belong within an established analytical business. Public sources do not reveal its complete experimentation portfolio. They do reveal why the next chapter deserves more demanding definitions of success.

Too Good To Go made a market by giving uncertainty a price and surplus a destination. Its strongest future would connect that achievement to a further alignment: merchants recovering value without acquiring an incentive to create unnecessary excess, customers receiving things they use, and environmental claims tracking changes that would otherwise not occur.

Surprise may be part of what makes the purchase appealing. Whether it creates value, for the merchant, the customer, and the environment, deserves less uncertainty.

References

Footnotes

1. Too Good To Go, “Our Founders” and “Our History”. Company chronology distinguishes 2015 origins from the 2016 app launch. ↩

2. Steve Wynne-Jones, “A Practical Solution: ESM Meets Mette Lykke, CEO, Too Good To Go”, European Supermarket Magazine, December 10, 2019; company history for investment and CEO dates. Interview recollections are attributed, not independently reconstructed. ↩ ↩² ↩³

3. Too Good To Go, merchant recruitment FAQ. Commercial terms vary; this is not a universal fee schedule. ↩

4. Too Good To Go and Whole Foods Market, national rollout announcement, July 17, 2024. Prices and advertised values are historical launch examples. ↩

5. Luxi Shen, Ayelet Fishbach, and Christopher K. Hsee, “The Motivating-Uncertainty Effect: Uncertainty Increases Resource Investment in the Process of Reward Pursuit”, Journal of Consumer Research 41(5), 2015, 1301–1315. Author-hosted paper. ↩

6. Sandra van der Haar and Gertrude G. Zeinstra, The impact of Too Good To Go on food waste reduction at the consumer household level: An explorative study, Wageningen Food & Biobased Research Report 1975, October 2019, summary, pp. 5–6; DOI. ↩

7. Louisa Boecker et al., “Investigating factors on the intention to use mobile apps against food waste”, International Journal on Food System Dynamics 16, 2025, 1–26. Accessible full text. ↩

8. Kate Schubel, “I Saved Money on a Grocery Surprise Bag: Here's Why I Won't Buy Another One”, Kiplinger, 2026. Individual consumer account, not a representative survey. ↩

9. Too Good To Go, US launch announcement, September 29, 2020; West Coast expansion announcement, May 5, 2021. ↩

10. ABG Sundal Collier, financing transaction announcement, January 8, 2021. Adviser disclosure, not total funding to date. ↩

11. Too Good To Go, Impact Report 2025, community and impact sections. Company-reported totals; 156,984,983 meal-equivalents rounded to 157 million. ↩

12. Katie Searles, “Too Good To Go expands into personal care, pet and home”, Retail Gazette, October 15, 2026, reporting the company's scale; the original announcement supplies the country count. This is corroboration of the disclosure, not an independent user audit. ↩

13. Too Good To Go Holding ApS, Annual Report 2025, consolidated income statement, printed p. 41; group cash flows, p. 47; note 3, pp. 57–58; note 8, p. 66. Uses consolidated group accounts, not individual subsidiaries. ↩ ↩²

14. Too Good To Go Holding ApS, Annual Report 2024, strategy discussion of CodaBene acquisition in 2022. ↩

15. Too Good To Go, enterprise Platform documentation and impact methodology, Parcels description. Product disclosures establish capabilities, not causal performance uplifts. ↩

16. Too Good To Go, Canadian “Dynamic Pricing Appendix”, updated July 28, 2025. Optional commercial tool; no worldwide availability inferred. ↩

17. Luyi Yang and Man Yu, “Too Good To Go: Combating Food Waste with Surprise Clearance”, Management Science 72(3), March 2026, 2307–2325; online July 22, 2025. Full text, especially sections 3–5. ↩ ↩²

18. Nymisha Bandi, Maxime C. Cohen, and Saibal Ray, “Reducing Food Waste Through Social Impact Partnerships”, working paper posted September 15, 2026, abstract; McGill research account, October 2026. Full manuscript could not be retrieved for this review; estimates and design are attributed to the abstract and university account. ↩ ↩²

19. Alberto Bertello, Canio Forliano, Paola De Bernardi, and Domenico Dentoni, “Too Good to Fail: Unpacking Legitimation Work in Social Platform–Based Enterprises”, Journal of Business Ethics, July 6, 2026. Methods and findings; historical Italian case and disclosed researcher collaborations. ↩ ↩²

20. Too Good To Go, “Our Impact Calculation”, methodology, average-bag convention, and external review description. ↩

21. Mérieux NutriSciences | Blonk, Environmental footprint of the food saved by Too Good To Go, v1.1, 2024, sections 1.2.5, 2.3–2.4, and 4. Displacement is a modeled assumption; an additional 10% household-waste factor is applied. ↩

22. Brendan T. Lawson, Marianna Coulentianos, Emilie Vrain, and Val Mitchell, “CO2e (best) avoided? How people experience CO2e avoided on the Too Good To Go app”, Big Data & Society, December 2, 2025. Longitudinal qualitative study involving 12 people in 10 households. ↩

23. Too Good To Go, “Too Good To Go expands its Marketplace beyond food”, October 15, 2026. Original strategy announcement; store count and partner examples are company-reported. ↩ ↩²