From 2022 to 2024 I built a layer-1. EXX. Self-funded. The number I put on every update was testnet users. We crossed thirty thousand. I said it in conversations the way you say revenue. I planned the next six months as if the company had already worked.
It had not. We never shipped a mainnet I would put my own money on.
The wallets were real. People connected, took faucet tokens, sent a transaction so they could show a screenshot. That is a number you can grow with a campaign. It is not people who need the chain next Tuesday. It is not fees. It is not a reason to hire as if you were already live.
I knew the difference. I still used the big number, because the honest one was smaller and harder to explain.
What the number actually counted
A testnet user, the way we counted it, was a wallet that had shown up once. Some came back. Many did not. Some were the same person on a new address. The dashboard did not care. Thirty thousand looked like a market.
What I did not put on the slide: how many sent a second transaction a week later. How many used anything other than the faucet. How many would still be there if the tokens were worthless tomorrow — which, on a testnet, they already were.
I had already run a launchpad where the useful number was different. Not visits. How many sales finished, and whether people got their tokens out. That number was smaller. It described work. On EXX I picked the number that described hope.
What it cost
Once you treat wallets as the business, you spend like you have a business. More utilities so the ecosystem looks busy. Another page so the number can grow. Conversations that start with the headline and never get to what happens after testnet.
You need testers. Testnet is how you find breaks. The mistake was treating the tester count as proof the company already existed. The chain still needed something a stranger would use without a faucet, without me on the call, with money that hurts to lose. That work does not move a user counter in a week. So it waited.
What I look at now
I still want one number I can say out loud.
If the product takes money: how many people finished the path. On the launchpad that was sales that closed and tokens that landed in wallets — not visits.
If it is not live yet: how many came back a week later with no bounty. If they only came for the faucet, they are testers.
If I have to add “well, they connected once,” it does not go in a deck.
We did not get to a mainnet a stranger could use. Next time I want a smaller number I can defend, not a bigger dashboard.