Three Ways to Find Marketplace Whitespace
Unbundle a low-NPS segment, own an ignored niche, or change the atomic unit of supply
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 4
- Confidence
- 87%
Tavel's three repeatable patterns for finding a new marketplace opportunity. One: take a horizontal marketplace, find the segment with the worst experience, and build a vertical marketplace for it (eBay to sneakers gives you GOAT). Two: find a niche nobody is paying attention to, bring it online, take the friction out and reveal it is bigger than anyone thought (Etsy, Hipcamp). Three: change the atomic unit of supply so that supply nobody could previously access becomes available (GrubHub required restaurants with their own delivery fleets; DoorDash/Postmates/Uber Eats supplied the fleet, so every restaurant became supply).
Origin
Sarah Tavel's whitespace post, summarised here. She extends pattern three to LLMs: automated outreach and onboarding may make the long tail of supply reachable for the first time, opening supply types nobody can currently anticipate.
Core principles
- 01Low-NPS segments of horizontal marketplaces are pre-validated demand with a broken experience.
- 02Ignored niches look small from the outside — that is the source of both their winnability and their upside.
- 03Changing the atomic unit of supply expands the addressable supply pool without inventing new demand.
- 04LLMs most plausibly attack pattern three: automating the reach and onboarding of long-tail supply.
- 05Whichever pattern you use, you still owe the market a Level 1 thimble.
How to run it
- 1
Pattern one — unbundle the low-NPS segment
Take an established horizontal marketplace, identify the category where buyer or seller satisfaction is worst, and build a dedicated vertical marketplace for it. Sneakers on eBay were a low-NPS segment; GOAT is the vertical that resulted.
Pro tip The horizontal marketplace has already proved the demand exists — you are only fixing the experience.
- 2
Pattern two — bring an ignored niche online
Find a niche nobody is paying attention to, bring it online, remove friction, and demonstrate that the market is bigger than people assumed. Etsy did it for handmade goods; Hipcamp for campable private land.
Pro tip The lack of attention is the asset — no competition means you can saturate and tip.
Watch out Confirm the adjacency: the niche must have room to grow into something larger.
- 3
Pattern three — change the atomic unit of supply
Redefine what counts as a unit of supply so that previously inaccessible supply becomes available. GrubHub's atomic unit was a restaurant that owned a delivery fleet. DoorDash, Postmates and Uber Eats redefined it as any restaurant, because the marketplace supplied the fleet.
Pro tip Ask what constraint currently disqualifies most potential supply, and whether you can absorb that constraint yourself.
Watch out This is the pattern that disrupts incumbents. GrubHub was too slow to change the atomic unit of its own supply and got disrupted by exactly this move.
- 4
Apply the LLM lens to pattern three
Ask whether LLMs let you reach, onboard and support a long tail of supply that was previously too expensive to serve — which would expand the supply pool in ways nobody can currently anticipate.
Pro tip Also ask whether LLMs can create a materially better experience inside a low-NPS segment or an offline niche, unlocking patterns one and two.
Watch out Tavel says she has not yet seen an example of this working — treat it as a hunting ground, not a validated pattern.
In the wild
GrubHub's supply was restaurants that already ran their own delivery. DoorDash, Postmates and Uber Eats said the marketplace would provide the delivery fleet, so any restaurant became eligible supply.
→ The available supply pool expanded enormously, and GrubHub, too slow to change its own atomic unit and constrained on capital as a public company, was disrupted.
eBay was a horizontal marketplace where the sneaker segment had a poor experience — authenticity, trust and merchandising were all weak for that category.
→ GOAT built the vertical marketplace for exactly that segment, on demand eBay had already proved existed.
Common mistakes
Looking for new demand instead of new supply
Two of the three patterns are supply-side moves. The unlock is usually making supply that already exists reachable, transactable or better-experienced — not conjuring demand that did not exist.
Assuming LLMs unlock the demand side
Tavel's read is that the clearest LLM opportunity is pattern three — automating the effort of reaching and onboarding long-tail supply. She has not yet seen a working example, so the honest position is that it is a hunting ground.
Is it for you?
Best for
Aspiring marketplace founders and investors hunting for a category to attack, before any product exists
Not ideal for
Operators already inside a chosen market — this is an idea-generation lens, not an execution one
From the transcript
“you always find a horizontal Marketplace and you find the kind of low NPS part of that segment and you create a marketplace around that”
“the second is finding a niche that no one else is paying attention to”
“the third is changing the atomic unit of the supply”
From the episode
The hierarchy of engagement
Sarah Tavel (Benchmark, Greylock, Pinterest)