The second customer is the real test

This week my product failed in the most useful way possible.

I built it around my first customer. Every setting was tuned to them. Every step had been walked through by me personally, by hand, behind the scenes. And for that customer it worked perfectly, every single day.

Then a second customer arrived. They could not get past step one.

Not step five. Not some rare edge case deep inside. Step one.

Why did a working product fail at step one?

Because the product had never actually onboarded anyone. I had been the onboarding.

When you set your first customer up by hand, you personally carry them over every gap in the product. You do not notice the gaps, because your own hands are the bridge. The product looks finished. It is not. It is finished for exactly one person.

The second customer arrives without you attached, and every gap you have been quietly bridging is suddenly a wall.

What did the audit find?

I ran one audit with a simple question: can a brand-new customer get from signing up to a finished result without me touching anything?

It found six blockers. The pattern behind them is worth more than the list, but here is the flavor:

1. A required permission step was never attached for new customers. So one whole group of them could never get any output at all. Not bad output. Nothing.
2. The built-in quality check had been written around customer one. It measured everyone against the first customer, so everyone who was not them failed.
3. There was no way for a new customer to set their own schedule, no way to start their own new project, and no way to lock in their own core details. All of that had been set up once, by hand, for customer one, and the buttons to do it simply did not exist.
4. Even the last step, getting the finished piece with its words attached, had no path a customer could reach.

Every one of these was invisible while I was the one driving. Every one of them was a wall for anyone else.

The two problems I did not expect

The audit surfaced two more things that had nothing to do with onboarding, and they scared me more.

One: a supplier my product depends on had a single bad moment, the kind that fixes itself in a minute. My product treated that one bad moment as a permanent no, and quietly cancelled two weeks of planned work. Nobody was told.

Two: settings from the first customer were bleeding into the second customer’s results. Customer two was getting little pieces of customer one.

Neither shows up when you only have one customer. Both are unacceptable the moment you have two.

All six fixed the same day

Here is the part I am actually proud of. The audit ran, the list existed by the afternoon, and all six blockers were fixed the same day. The second customer now gets from step one to a finished result on their own.

That speed is not because I am fast. It is because finding the problems was the hard part. Once the audit asked the right question, the fixes were obvious.

What this means if you have one happy customer

Think of it like a tailored suit. It fits one body perfectly. That tells you nothing about how it fits anyone else.

One happy customer proves the idea. The second customer tests the product. So if you have something working today, ask the question I waited too long to ask: what happens when someone you have never met tries it, without you in the room?

That answer is your actual product. Everything else is you, doing favors.

I build AI agents that run all parts of my online business. Check them out at pelleforsman.com.