The Hierarchy of Marketplaces
Focus, tip, dominate — the only sequence that produces a marketplace worth owning
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 95%
A three-level hierarchy for marketplace builders, reasoned backwards from the goal. Marketplace equity value comes from dominance — profitability rises sharply with how much bigger you are than the number two. Dominance requires tipping a market, and tipping requires saturating a market, which is only achievable if you have brutally constrained the market you attack first. So: Level 1 focus (the thimble), Level 2 tip the market, Level 3 dominate.
Origin
Sarah Tavel's marketplace counterpart to her Hierarchy of Engagement, developed at Benchmark from the same allergy to vanity metrics — this time GMV rather than MAUs. She kept meeting founders fixated on the $1M annualised GMV series-A milestone while diffusing that GMV across cities. The dominance-versus-profitability relationship comes from a graph she saw as a board observer at the online classifieds company OLX (led by Fabrice Grinda). Bill Gurley's 'great marketplaces create the new incumbents' informs Level 2.
Core principles
- 01GMV is the marketplace's MAU — necessary but not the race you should think you are running, and it can actively pull you the wrong way.
- 02Not all GMV is created equal: $1M concentrated in one city is worth vastly more than $1M spread over five.
- 03Being barely number one gives you none of the benefits of the model — you fight for every incremental point of share.
- 04The only scalable route to dominance is tipping a market, and you cannot tip a market you have not saturated.
- 05Assume scarcity of three things: capital, the founder's attention, and the customer's attention.
- 06Focus your ambition like a laser on a thimble, not like the sun trying to warm an ocean.
How to run it
- 1
Work backwards from dominance
Start from the endgame — Airbnb, Amazon, eBay, Google — where dominant share produces a highly profitable cash-generating business. Accept that the profit curve is steeply tied to how far ahead of number two you are, and let that dictate everything upstream.
Pro tip If you cannot articulate a market you could plausibly dominate, you do not yet have a marketplace plan.
Watch out Middle-of-the-pack, or number one by a nose, is the worst place to be: all the operational pain of a marketplace with none of the winner-take-most economics.
- 2
Level 1 — pick the thimble
Constrain to a market small enough to heat to boiling: a single city, a single category, or both. DoorDash chose the suburbs — a market others thought was uneconomic and therefore uncontested. Etsy chose craft-fair goods.
Pro tip Uncontested and 'too small' is a feature: you get customers desperate for attention and no competitor bidding against you.
Watch out Racing to $1M GMV pushes you toward skimming the cream off a huge market — easier, faster, and the opposite of what builds enduring value.
- 3
Reach minimum viable happiness
Do the unscalable work — product experience, friction removal — until a defined percentage of people retain after a transaction. That threshold, not GMV, is the gate to Level 2.
Pro tip Use Sean Ellis's question rather than NPS: if 40%+ say they would be very disappointed if the product disappeared, you have a white-hot centre.
Watch out You will not make everyone happy — look for a core persona you make really happy, not a broad average.
- 4
Level 2 — tip the market
Swap the unscalable tactics for scalable tipping loops (growth loops plus happiness loops) and drive toward saturation until suppliers and buyers start coming to you. Early signals: cohorts improving, organic growth appearing, individual suppliers or buyers tipping to you.
Pro tip Stay close enough to both sides that when they lean in you are there to receive them and can build momentum behind it — Rekki did not predict its own tipping mechanism.
Watch out Some markets simply cannot be tipped. Run the tippability test (supply concentration, supply homogeneity, entrenched competition) before pouring capital in.
- 5
Level 3 — dominate and expand on three vectors
Only now go wide. Vector one: keep penetrating the market you already tip. Vector two: expand use cases within it (Uber: black cars, then UberX, then Pool). Vector three: land-grab adjacent markets as fast as you can run the now-proven playbook, funding each new market with contribution profit from the tipped ones.
Pro tip Put more wood behind fewer arrows — each market where the flywheel spins produces contribution profit, which buys capital, which buys the next market.
Watch out The playbook does not always transfer: Rekki found London's restaurant supply highly fragmented and Berlin's not. Expect it to rhyme, not repeat.
In the wild
Postmates started with big ambition across big cities and across restaurants, retail and more, so it was always being compared to whatever substitutes existed and spread thin across many vectors of customer and seller preference. DoorDash famously and controversially went after the suburbs, lost money early on long drives, but faced almost no competition and had customers desperate for the service.
→ DoorDash made both sides happy enough to retain in a market it could actually saturate, then went from strength to adjacent market. It is the larger company today.
Founders treated $1M annualised GMV as the series-A trigger and, to hit it, told themselves they had to prove the value proposition in multiple cities — $500k here, $200k there. Tavel's read: they were making their own job dramatically harder, because $1M of GMV skimmed off a huge market is easy and worthless, while $1M inside a constrained market is hard and is the actual path.
→ The hierarchy was written to break the reflex: diffused GMV never saturates anything and so never tips anything.
Spread $1M of go-to-market spend across ten cities and a competitor puts $1M into one of them. They win that city outright, which earns them contribution profit and more capital to keep going — while you own nothing decisively anywhere.
→ Tavel's argument for why focus is a defensive necessity, not just an efficiency preference. And you never face only one competitor.
Common mistakes
Blitzscaling at Level 1
Ambitious founders start with the land grab — go big immediately, win before anyone catches up. Blitzscale is Level 3; done first it plants a thousand flags and decisively wins nothing.
Diffusing focus to 'prove the value prop travels'
Expanding cities early to validate the model is the most common rationalisation. It guarantees you saturate nowhere, and saturation is the precondition of tipping.
Chasing GMV once you have a brand
Etsy chased GMV growth ahead of its IPO and let mass-produced goods flood a handmade-goods marketplace. Handmade sellers and buyers reacted, trust eroded badly, and the period preceded a CEO change.
Resting on your laurels after tipping
The history of marketplaces is a history of disruption — HomeAway/VRBO by Airbnb, GrubHub by DoorDash. Consumer expectations only ever rise; a tipped market can be untipped.
Is it for you?
Best for
Marketplace founders and product leaders (B2B or B2C) deciding how wide to go and what to measure at seed through series B
Not ideal for
Markets structurally incapable of tipping (concentrated or homogeneous supply), or products with genuinely low repeat frequency where the buy side cannot be cornered
From the transcript
“basically layer one is focus Layer Two tip the market layer three dominate the market”
“the best ambitious Founders do is they focus that ambition like a laser beam on a small Market the thimble”
“so there's there's kind of three vectors then that any Marketplace can grow at this point”
“you always want to put more wood behind fewer arrows”
From the episode
The hierarchy of engagement
Sarah Tavel (Benchmark, Greylock, Pinterest)