Live commerce is 30-40% of e-commerce in China and still single digits in the US
Whatnot started with Pokémon cards and grew into a platform doing $8 billion in annual sales by letting users sell to each other on video. Live commerce is already 30-40% of all e-commerce in China and still single digits in the US — that gap is going to close.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Not knowing the market was actually an advantage
Grant says they started out knowing almost nothing about the market — they had never heard of live shopping in Asia, and their impression of QVC was ‘grannies shopping on QVC’. David George says flatly that this may have been an asset rather than a liability: if you understand the market deeply, you iterate on what already exists instead of building something entirely new. Grant calls it a ‘1,000% advantage’, because users don't care about the market at all — they care about whether the thing is fun and whether it has value.
— Grant LaFontaineLive shopping isn't e-commerce, it's a mall
Grant likens Whatnot to walking through a mall: you know you want something, but not specifically what, so you go in, wander around, play for a while, and maybe end up buying a pair of shoes. Traditional e-commerce assumes you already know what you want to buy, so it just moves offline selling online and makes it more efficient — it's not good at discovery or entertainment. Live does the opposite: it expands demand, because you find things you didn't know you'd be interested in. David adds that this is closer to the demand expansion that came with the birth of the department store or the shopping mall than to an extension of e-commerce.
— Grant LaFontaineOver 80% of users buy nothing that day
Whatnot users spend about 95 minutes a day on the platform, a number already on the scale of social and entertainment platforms. But Grant stresses that on any given day the vast majority of people buy nothing — over 80% of users don't transact that day, they just come to watch, chat and play. So it's fundamentally an entertainment format, with shopping as one part of it. David says he himself spends a lot of time on it, both for entertainment and to transact.
— Grant LaFontaineTop sellers make over $100M a year at 20-40% margins
The biggest merchants on Whatnot can do over $100 million a year in revenue, with EBITDA margins of 20%, 30% or even 40%-plus. The platform did over $8 billion in sales last year, and has already done over $8 billion this year to date — growing fast — but the platform only takes a small cut. Grant says they don't plan to raise rates: let merchants keep the lion's share and the platform grows along with them. David calls it one of the least-reported stories in American small-business enablement.
— Grant LaFontaineFrom collectibles to more than 100 categories
Today Whatnot sells across more than 100 categories. Grant names a few: fresh fish shipped straight off the dock in San Diego, barbecue joints, golf — golf is growing several hundred percent year over year. Women's clothing is already the largest category, closer in shape to TJ Maxx, selling off-season inventory at good prices. His test is: anything worth opening a physical store for is a fit for Whatnot, because you can do that store visit from your couch on your phone.
— Grant LaFontaineGoing global came with no localization problem
Whatnot now covers about 10 countries: North America (the US and Canada), five European countries, Japan and Australia. Grant says the most counterintuitive part is that it works exactly the same way in all of them as it does in the US — the same small businesses get created, the same collectibles, fashion and electronics, with highly similar dynamics. That's what convinced him live is a very durable shopping medium.
— Grant LaFontaineTrust and safety is 40% of the staff
Grant says trust and safety is roughly 40% of Whatnot's employees, a huge investment, because nobody wants to put their credit card somewhere they don't trust, let alone buy fresh fish from somewhere they don't trust. Mechanically they have a rules engine that processes billions of data points per second and can identify and act on bad behavior at sub-second speed: harassment in a livestream gets detected by an LLM, fed into the rules engine, combined with seller history and other signals, and programmatically warned, suspended or banned; late shipping and excessive refund rates run through the same system. David's analogy: this is Whatnot's police and judicial system, equivalent to running two New York Cities.
— Grant LaFontaineAI is for efficiency, not for replacing people
Grant says the platform is self-reinforcing: if someone livestreams with an AI avatar, users won't come back, and that's not a good business. The core of Whatnot is walking into a store, getting to know the owner, and getting to know the regulars. So good sellers don't want to remove the human part, they want to amplify it. AI lands on helping sellers run the business more efficiently: listing products, automatically inferring from the livestream the metadata the discovery system needs, and telling sellers analytically where they're doing well and where they can improve. They don't offer avatar tools, because that's not why users come to Whatnot.
— Grant LaFontaineIn their own words · checked verbatim
users don't give a shit about the market. Yeah. They don't care about the market at all. They care about whether they're going to enjoy the thing. It provides value.
Grant LaFontaine10:21
if you look at a lot of the e-commerce companies that have been built in the past, prior to us, you sort of know exactly what you want. It's not really great for discovery.
Grant LaFontaine13:25
most people on whatnot don't purchase anything. The vast majority of people don't purchase anything on whatnot. They're just going and they're watching and they're talking with people.
Grant LaFontaine14:25
if you look at e-commerce today, it's so, you have to know exactly what you're looking for. So all it's really done is take a bunch of offline sales and bring them online in a more efficient way.
Grant LaFontaine19:28
one of the ways to make it so efficient is to obscure who you're buying from and obscure the ability for people to like build brands and businesses around it and whatnot, sort of the polar opposite.
Grant LaFontaine21:38
I think trust and safety in particular on whatnot is about 40 percent of employees. Wow. Yeah. It's a huge, huge investment from us because it's a challenging one.
Grant LaFontaine33:09
how do you help someone keep it human but run their business more efficiently?
Grant LaFontaine38:12
Figures
| Average daily time spent per Whatnot user | about 95 minutes | 19:28 |
| Live commerce as a share of all e-commerce in China | 30-40% | 14:25 |
| Live commerce as a share of all e-commerce in the US | single-digit percentage | 14:25 |
| Share of Whatnot users who don't buy on a given day | over 80% | 20:31 |
| Annual revenue of top Whatnot sellers | over $100 million | 24:44 |
| EBITDA margin of top Whatnot sellers | 20-40%+ | 24:44 |
| Whatnot sales last year | over $8 billion | 24:44 |
| Whatnot sales this year to date | over $8 billion | 24:44 |
| Number of countries Whatnot covers | about 10 | 30:03 |
| Trust and safety as a share of Whatnot employees | about 40% | 33:09 |
Glossary
- live commerce
- A form of e-commerce in which goods are shown and sold via real-time video livestream.
- rules engine
- Whatnot's backend system that processes billions of data points per second and automatically judges and acts on bad behavior on the platform.
- EBITDA margin
- A common measure of a business's operating profitability.
How to listen
Founders and investors building in e-commerce, live streaming and marketplaces; product leads who care about small-business enablement, building trust and safety systems, and where the line on AI actually falls.
The opening Pokémon card and domain-name stories can be fast-forwarded; start at 10:21 with the market read.