How I launched a marketplace with no buyers and no outside sellers

Why does it matter? Because fake inventory, a giveaway that flopped, and a pivot nobody planned for — that's the actual $20B playbook

Grant LaForce built Whatnot by listing Funko Pops he didn't own and scrambling to source them after each sale. The live selling feature that actually made the business work wasn't in the original product. VCs passed at 100% month-over-month growth. None of it looked inevitable from inside — and that's the point.

• Faking supply is a real cold-start strategy — Whatnot ran it systematically, with a pricing algorithm to find sourceable inventory
• The product that saved the company launched seven months in; the original marketplace would have died at $25K total GMV
• Live commerce in the US sits at single-digit percent penetration; China is at 40%
• A/B tests are structurally unreliable on two-sided platforms — most marketplace teams don't catch this

Whatnot launched with fake inventory: Grant listed Funko Pops he didn't own, then scrambled to source them only after each sale

Every listing on Whatnot's platform in the early days was fake — in the sense that the inventory didn't exist yet when it was listed. When a buyer checked out, a team in Brazil hunted the specific Pop across a hundred online stores, then shipped it authenticated.

That process required real infrastructure. With 100,000+ Funko Pop variations in the wild, they built a pricing algorithm that scraped sell-through data across every major site and set a liquidity floor. "We'd look at the amount of liquidity in each version of the Funko Pop" — if something hadn't moved 10 units in 30 days, you couldn't reliably source it. The early algorithm still had gaps: buyers spotted mispricings and flipped $500 Pops bought at $150.

Whatnot didn't open to a second external seller until February 2020 — three months in — when buyer demand was dense enough to keep real sellers around. The playbook: simulate one side of the market yourself until demand is real enough to attract the other.

The feature that built Whatnot didn't exist at launch — without it, the company dies somewhere around $25K total GMV

December 2019: 30 sales. March 2020 after YC demo day: $25,000 in total GMV. Those are genuinely grim numbers.

July 2020: live auctions go live. "That's when the business really started." By September, Whatnot was doing $250–300K in a single month at 100%+ month-over-month growth.

The static collectibles marketplace was failing slowly. Live selling was a different product — different retention mechanics, different seller economics, different community dynamics. Founders often credit the original idea when the actual breakthrough was a late pivot they almost didn't make. Whatnot's version is unusually clean: the original product was going to die. If one feature bends your growth curve 90 degrees, that feature IS the business. Everything before it was research.

Track which feature or channel is driving actual retention and GMV. The business-defining insight often arrives 6–18 months in, not at launch.

From $2.3M to $163M to $1B in three straight years — this is what a category inflection looks like when it actually happens

End of 2020: $2.3M GMV, 17–18 employees. End of 2021: $163M, ~100 employees. End of 2022: $1B.

70x in two years. When Shaan heard "163 million," there was a beat before anyone moved on.

Grant credits two drivers: the live format genuinely transformed seller economics during COVID — brick-and-mortar comic shops migrated entirely to Whatnot and started outperforming their physical storefronts — and the team cracked category expansion fast. Launching new verticals triggered buyer-seller cross-pollination that compounded everything working in the original Funko Pop market.

When a marketplace goes 70x in two years, you're either building on it now or you're watching someone else do it.

Live commerce in the US is at single-digit percent — China is at 40% — and Grant calls 20–30x growth from here the base case, not the bull case

Sam asks what share of e-commerce live shopping represents today. Grant: "Single-digit percentages." His 10-year target for the West: 30%+. His growth multiple to get there: 20–30x.

The three biggest US and European live commerce platforms combined sit at roughly $30 billion. China is at 40% of all e-commerce. That gap is the opportunity.

What grows alongside it: wholesale supply marketplaces already pulling significant raises on the back of Whatnot demand alone; multi-channel network companies managing sellers' live operations; inventory and analytics tooling. Whatnot's biggest seller is already at multiple nine figures. In China, the largest live sellers individually do over $1B per year. The Streamlabs equivalent for live commerce hasn't been built in the West.

The giveaway that defined Whatnot's growth had 104 entries and zero purchases on its first run — the 2.5x improvement per iteration was the only signal that mattered

104 entries. Zero purchases. Grant was funding this from personal savings. "The bank account's dwindling."

The mechanic: weekly raffle for a rare $500–$1,000 Funko "Grail." Entry requires a referral share; more shares equals more tickets. First run: flat.

What made him keep going — the second giveaway hit roughly 250 entries. "Maybe it wasn't exactly 3x, but it was 2 and a halfx." That compounding rate, even off a base of 100, is the data point worth betting on. Not that it worked, but that the failure rate was improving at a predictable rate. Eventually they took over every Funko Pop subreddit and Facebook group in the hobby.

Don't kill a growth mechanic after one failed run. Kill it when the improvement rate stops.

A/B tests are structurally broken on two-sided platforms — the right diagnostic is individual seller elasticity, not aggregate metrics

Your A/B test results on a marketplace are probably wrong. Grant explains the mechanism: change the feed, buyers shift into different streams, sellers adjust their behavior in response, and you've contaminated both sides of your experiment simultaneously. "AB tests are notoriously bad on network systems."

The real diagnostic: plot individual seller elasticity curves — what happens to a specific seller's business when their impressions increase versus decrease. Grant found sellers on Whatnot who shrink faster when impressions drop than they grow when impressions rise. That asymmetry is invisible in aggregate A/B data and only surfaces when you trace seller trajectories over longer time windows.

His heuristic for all of it: "If you start to use a bunch of fancy words, I'm going to make you explain it to me like I'm in middle school. If you can't do that, you probably don't know what you're talking about." Complexity is usually a cover story.

Investors passed on Whatnot at 100% month-over-month growth because the market was 'too small' — that was the feature, not a flaw

Growing fast, still getting rejected. "Investors were very nervous to invest in a Funko Pop company." The objection was always the same: small market.

Grant's reframe: "If you're a four-person company like we were in the early days, you can't compete in a mass market." Every dominant consumer company started in an impossibly small corner. eBay's actual first sale was a broken laser pointer to a broken laser pointer collector — the Pez dispenser story was invented for PR. PayPal was a niche tool for eBay power sellers before it was anything else.

Pick your first market by how completely you can own it with today's team — not by how large it could theoretically become.

Whatnot rejected the visionary founder identity from day one — and treated every founder conviction as a hypothesis to disprove

"We're not a sort of vision-driven company. We're a user-driven company." Grant says it flatly and means it.

Shaan pushes back with the obvious counterexample: Steve Jobs. Grant doesn't budge. Direction matters, but the customer always wins the stack-ranking. "If you have this amazing idea for something and customers do not want that, you're dead to rights."

More pointedly: "Stories are sort of BS. A lot of stories in terms of like business are really a mechanism to convince people to follow you versus something of like real substance." The Whatnot origin narrative sounds clean in retrospect. In real time, it was a series of user-driven pivots — the live feature that made the company work wasn't in the original pitch. Vision sets the direction. Users tell you if you're on it.

The infrastructure layer for live commerce in the West is almost entirely unbuilt

The trajectory Whatnot traces — 30 sales to $1B GMV in under 24 months — is less a success story than a live demonstration of where Western e-commerce is heading. Live commerce in China represents 40% of all online spending. The supply chain, analytics, and multi-channel management tooling that live sellers need at scale is largely unbuilt here.

Grant's parting line on speed applies to all of it: "If you think you're moving fast, you're probably not."

---

Topics: live commerce, marketplace cold start, whatnot, two-sided platforms, growth hacking, founder story, e-commerce, collectibles, startup strategy, YC, niche markets