This morning a founder wrote something most founders only admit after the postmortem: an early advisor told him you will always be wrong about how close you are to product-market fit, and he then spent months mistaking love for his team's labor for love for his product. His customers were happy. His revenue was real. And a good part of both belonged to a category he had not priced: people renting his engineers, one Slack message at a time.
Why demand for you looks exactly like product-market fit
From the inside, the two are indistinguishable. Usage is up because your engineer is in the customer's workspace daily. Retention is perfect because leaving would mean losing that engineer. Expansion happens because you personally found the next use case during a call. Every dashboard says fit. What the dashboard cannot show is the counterfactual: what all those numbers do the week your people leave the building.
This is the standard trap of services-first and forward-deployed models, and it is getting more common, not less, because AI made hands-on delivery cheap enough that every product company can afford to smother its customers with attention. The attention works. That is the problem. It works whether or not the product does.
What people say will not help you here
Asking the customer does not resolve it. They will say they love the product, and they are not lying, they just have no reason to separate the software from the person who makes it work. Stated satisfaction bundles everything. Only behaviour unbundles it, and only behaviour in your absence.
Three readings that separate the product from the people
Usage in your absence. Pull your team back from an account for 30 days, planned and communicated, and watch the curve. If usage holds, the workflow depends on the product. If it decays, you were the product.
Renewals without rescue. A renewal that arrives without a heroic save, a discount or a founder dinner is a vote for the software. A renewal you had to go get in person is a vote for you.
Complaints that keep coming. Bug reports and feature requests are the least glamorous fit signal and among the most honest. Nobody documents the flaws of a tool they have quietly abandoned. Silence is what churn sounds like before it happens.
The cost of reading it wrong
Founders who mistake rented labor for product demand scale the wrong thing. They hire sales against revenue that cannot be resold, they raise against retention that is actually a staffing arrangement, and the truth surfaces at the worst possible moment, mid-scale, when the service layer stops stretching. The diagnosis was available two quarters earlier for the price of one uncomfortable experiment.
Key takeaways
Demand for your labor and demand for your product produce identical dashboards while your team is involved. Customers will not separate the two for you, their stated satisfaction bundles both. The clean readings are behavioural: usage in your absence, renewals without rescue, complaints that keep arriving. Services-first is a fine way to learn and a terrible thing to scale unexamined. Run the absence test before you hire against the revenue.
If the product only works when you are in the room, the market has already told you what it is paying for. It is just being polite about it.