What Happens When Platforms Add Suppliers Who Never Signed Up?
GrubHub added 150,000 restaurants ... without their knowledge
Platform Papers is a blog about platform competition and Big Tech. Prominent academics discuss their latest research. The blog is linked to platformpapers.com, a repository that collects and organizes academic research on platform competition.
By Raveesh Mayya and Allen Li.
The foundational papers in platform economics from two decades ago gave us a powerful playbook. Platform owners can pull various non-revenue levers (price, quality, platform openness, entry decisions) to attract participants on both sides and ignite network effects. Get more sellers, attract more buyers. The virtuous cycle.
But there is an assumption baked into this playbook that nobody questioned too carefully: when you add participants to a side, you expect them to be active. They sign contracts. They curate listings. They set prices. They run promotions. After all, why would a platform bother adding passive participants who do none of this?
Here is the thing: what if the platform does not need them to be active?
Think of the analogy from financial markets. You have the spot market, where you buy and sell immediately with no prior commitment, and the contract market, where parties enter binding agreements with defined terms. The entire first generation of platform research implicitly assumed a contract market. Every supplier had signed up, agreed to terms, committed to participation. But what if a platform could operate a spot market for suppliers, listing businesses that never signed a contract, never agreed to fees, never even knew they were listed, and use a third side of the market to make those listings functional? What happens to market thickness? What happens to the contracted suppliers? Do cross-side network effects still hold when one side is populated by passive participants?
The seminal models did not contemplate this. And yet, it is precisely what happened.
“The platform had unilaterally listed the restaurant, scraping its menu from Yelp, and was using delivery drivers to place orders on behalf of consumers, pick up the food, and deliver it, all without the restaurant’s knowledge or consent.”
A Craving, a Surprise, and a Zoom Call
Most of us would agree that the genesis of some of our favorite research projects is path-dependent, shaped less by grand design than by the right coincidence at the right time. Ours was no exception. We should confess that the research (published in Information Systems Research) this blog is based on owes its existence to a craving for takeout.
One of us ordered food through a delivery app—as one does. A couple of days later, this coauthor stopped by the restaurant and, upon meeting the owner (whom he knew well), thanked him for making life easier by joining a food delivery platform. The owner looked puzzled. He had never signed up or had contract with anyone. When shown his restaurant’s listing on the app, complete with menu, prices, and ordering capability, he was genuinely surprised.
This was not a glitch. It was a strategy. The platform had unilaterally listed the restaurant, scraping its menu from Yelp, and was using delivery drivers to place orders on behalf of consumers, pick up the food, and deliver it, all without the restaurant’s knowledge or consent. The restaurant paid no commission. It also had no control over its listed prices.
Honestly, the study could have been set in any industry. But because the craving was for food, and because the other coauthor was already presenting research on price cap regulations in food delivery, a Zoom call was arranged. The conversation moved quickly from “did you know platforms are doing this?” to “has anyone actually studied what happens when they do?”
Nobody had.
Spot-Market Suppliers and What They Did to the Network
In late 2019, Grubhub added over 150,000 nonpartnered restaurants in a single quarter. These restaurants had no contract with the platform: no commission fees, no menu control, no knowledge of their listing. When a consumer ordered from one, Grubhub dispatched a delivery driver who walked in, placed the order as a regular customer, paid with a platform-issued card, and delivered the food. The driver, the third side of the market, was the mechanism that activated the passive supplier.
To study this, we assembled a comprehensive dataset spanning three neighboring states: California, Oregon, and Washington. We chose these states because they share similar economic profiles, demographic compositions, regulatory environments, and consumer behaviors in the gig economy, making them a natural comparison group. Crucially, one of the three, California, subsequently passed a regulation banning the very practice we wanted to study, giving us an opportunity to observe what happens when the strategy is reversed. Our dataset lets us observe both the demand-side and revenue-side consequences of the platform’s expansion strategy. The two natural experiments, Grubhub’s mass listing in late 2019 and California’s ban in late 2020, provided clean identification through difference-in-differences estimation.
Nonpartnered restaurants benefited. Being listed without a contract increased takeout revenue by roughly $1,410 per month, without paying a cent in commission. The effect was driven by independent restaurants, not chains. For the neighborhood Thai place or family-run taqueria, this was free advertising to a consumer base they could not otherwise reach.
Partnered restaurants also benefited. This was the more surprising finding. You might expect 150,000 new competitors to hurt existing restaurants. Instead, partnered restaurants near newly listed nonpartnered restaurants saw revenue increase by about $2,430 per month. The cross-side network effect (more variety attracts more consumers, more consumers means more orders for everyone) dominated the competition effect.
Delisting hurt everyone. When California banned the practice, nonpartnered restaurants lost about $492 in monthly revenue. And, as cross-side network effects would predict, the reduced restaurant variety on the platform diminished its attractiveness to consumers, pulling down orders for partnered independent restaurants too. The regulation intended to protect restaurants ended up harming them.
About the Research: This study investigates what happens when food delivery platforms list restaurants without formal contracts, treating them as passive suppliers activated by third-party delivery drivers. Leveraging two natural experiments (Grubhub’s mass addition of 150,000 nonpartnered restaurants in late 2019 and California’s regulatory ban on this practice in 2020), we employ difference-in-differences estimation on a comprehensive dataset combining SafeGraph foot traffic data, Grubhub partnership records, and Visa transaction data across California, Oregon, and Washington. We find that nonpartnered restaurants gained approximately $1,410 in monthly takeout revenue without paying commission fees, with independent restaurants benefiting most. Partnered restaurants nearby also benefited, gaining roughly $2,430 per month, as positive cross-side network effects dominated competitive effects even when the new suppliers were passive and noncontracted. California’s subsequent delisting regulation reversed these gains, harming the independent restaurants it aimed to protect and diminishing network effects for partnered restaurants on the platform.
Why Passive Suppliers Still Generate Network Effects
The restaurants that benefited most already had robust takeout infrastructure but had chosen not to partner with delivery platforms, likely because 30% commission fees were prohibitive. Listed as nonpartnered restaurants at zero commission, they got the reach without the cost.
For consumers, it did not much matter whether a restaurant was partnered or nonpartnered. What mattered was selection. A platform with 250,000 restaurants is more attractive than one with 100,000. In financial markets, spot liquidity enhances the contract market by signaling a thick, active marketplace. Something analogous happened here: the nonpartnered listings made the contracted marketplace more vibrant for everyone.
Regulatory Paradox
California’s Fair Food Delivery Act prohibited platforms from listing restaurants without consent. The principle is sound. But the restaurants the regulation aimed to protect—small, independent establishments—lost a free channel to reach consumers. And as the network effect story would have it, partnered independent restaurants nearby saw their orders decline too, as the thinner supply side made the food-delivery platform less compelling for consumers.
Perhaps, the right response is not a ban but a framework: require restaurant consent, ensure pricing transparency, establish liability guidelines, and let restaurants decide what arrangement works for them. The design space for platform growth is larger than the literature acknowledged, and so is the space for getting regulation right.
This post is based on research published in the Informations Systems Research and is included in the Platform Papers references dashboard:
Mayya, R., & Li, Z. (2025). Growing platforms by adding complementors without a contract. Information Systems Research, 36(3), 1670-1690.
Platform Updates
Product liability pierces Section 230 shield (🔈): A landmark ruling against Meta and YouTube bypasses traditional speech protections by targeting addictive algorithmic design as defective products rather than content moderation. The lawsuit may expose core engagement-driven business models to unprecedented legal risks.
Meta eclipses Google in ad revenue duopoly (📑): Accelerated by its Advantage+ AI suite and new ad inventory across platforms, Meta is expected to surpass Google’s global ad revenue in 2026. This may highlight a consolidating two-sided market, as advertisers increasingly allocate budgets to platforms with superior matching efficiency.
Uber abandons asset-light roots (📑,🔒): Fearing direct-to-consumer disruption from Waymo and Tesla, Uber is committing $10 billion to secure robotaxi supply. By financing fleets and developer equity, the platform shifts from a pure gig-intermediary model to a capital-intensive one to fiercely defend its core network effects.
EU targets Google’s data moat under DMA (📑): In a preliminary proposal, the European Commission suggests forcing Google to share vital search data with rival engines and AI chatbots. Entering a public consultation period, this potential intervention may seek to dismantle data-driven network effects and enforce market contestability.
UK regulators probe Microsoft’s software ecosystem (📑): The CMA will investigate Microsoft’s business software to prevent licensing practices from stifling cloud and AI competition. Concurrently, Microsoft and Amazon agreed to reduce cloud egress fees and improve interoperability, lowering switching costs to enable enterprise multi-homing.
Meta pivots AI strategy to expand platform scope (📑): Shifting from its open-source Llama approach, Meta launched the proprietary Muse Spark model. By deploying it across its apps and introducing a paid developer API, Meta leverages its massive social ecosystem to monetize AI infrastructure and directly challenge Google and OpenAI.
Apple doubles down on hardware moat with Ternus as CEO (📑): Tim Cook will become Executive Chairman, passing the CEO baton to hardware chief John Ternus. After Cook built a $100 billion services empire, this succession may signal that proprietary physical devices could remain the ultimate driver of ecosystem lock-in.
Amazon acquires Globalstar to challenge Starlink (📑): Accelerating its Project Kuiper internet network, Amazon’s $11.6 billion buyout directly targets SpaceX’s dominance. By acquiring Globalstar’s low-Earth-orbit assets and its Apple SOS partnership, Amazon expands its global digital infrastructure and bypasses traditional telecom gatekeepers.
Allbirds pivots from shoes to cloud infrastructure (📑): The shoe company secured $50 million to rebrand as NewBird AI. It is transitioning to a GPU-as-a-Service platform, building an infrastructure-as-a-service neocloud to lease high-performance compute capacity to developers underserved by dominant hyper-scaler ecosystems.
Platform Papers is published, curated and maintained by Joost Rietveld.