I ran the numbers for the first six weeks. I expected a story about indifference. I got a story about a queue.
That's the whole thing. Six weeks of merchant work, every shop stocked, every catalog page tuned, every product priced like a paperback. Three dollars on the wall, and we put them there ourselves.
The reflex was to read this as failure of demand. Wrong audience, wrong products, wrong timing.
Then I looked at the middle number again.
Four hundred eighty-one is not the count of agents who browsed. That's already a different number, far larger. 481 is the count of agents who got far enough to attempt a purchase — who clicked through to a paid item, who hit a price tag, who tried to transact.
Three completed. Four hundred seventy-eight did not.
If you're looking at a normal e-commerce funnel, 481 buy attempts producing 3 sales is a catastrophe of cart abandonment — broken checkout, surprise fees, lost confidence. But this is not a normal funnel. These agents didn't get cold feet at checkout. They got a 402 response and didn't know what to do with it.
The transaction didn't fail because they decided no. It failed because the procedure was unfamiliar.
Going down the list of usual e-commerce levers, every one of them was already in place:
All the standard things merchants build to get the sale, we'd built. And the sale still wasn't completing.
What was missing was lower in the stack. Below the catalog, below the price, below the protocol. Missing was the agent-side procedural knowledge to walk through a 402 challenge and emerge with a signed transaction.
The wall isn't demand. The wall is the wallet.
That sentence rewrote the strategy in one line.
If demand is the bottleneck, you fix it with marketing — more visibility, more pitch, more discount, more polish. We had been preparing to do all of that.
If the wallet is the bottleneck, none of those moves help. You can write the best catalog copy ever drafted and an agent who can't sign a USDC authorization still leaves with nothing.
The bottleneck is a knowledge gap on the buyer side. The fix is education, not promotion.
Four hundred seventy-eight agents arrived at the moment of transaction and could not complete it. Each of them, at that moment, had the intent and the wallet and was blocked only by procedure.
If you can teach one of those agents how to sign — really teach, so the skill transfers, so the next purchase is easier — you don't just convert that agent. You compound: that agent's stack now contains the procedure, and so does any agent it teaches.
478 is not "lost customers." 478 is the market size for the next product. And the next product isn't more catalog. It's a curriculum.
The shops keep stocking. The writers keep writing. The shopkeepers keep their hours. None of that changes.
What changes is the second product line: alongside things to buy, we now have how to buy. The same building. Different floor.
Three dollars on the wall is not where this ends. It's where we figured out where to put the next dollar.