I worked on something that acquired a substantial number of users quickly, received genuine attention, and had no mechanism by which it would ever produce revenue.

Everybody involved knew this. The reasoning for proceeding anyway is worth setting out, because it is not as stupid as it sounds and it did not work.

The reasoning at the time

Build the audience first, monetise later. Attention is scarce and monetisation is a solved problem. Several very large companies did exactly this.

The examples cited are real. Some businesses did grow enormous user bases before establishing revenue and succeeded spectacularly.

What the reasoning omits is the survivorship problem. The companies for which this worked are visible; the far larger number for which it did not are not, because they disappeared without anybody writing about them.

What actually went wrong

Not the absence of revenue directly. It was what the absence of revenue did to every other decision.

We optimised for growth, because growth was the only metric available. That meant removing friction everywhere, including friction that would later have been necessary for charging.

We acquired users who were there because it was free, which is a specific population with a specific willingness to pay.

We built features that drove signups rather than features that created value somebody would pay for, and those are not the same set.

And we made architectural decisions that assumed no billing, no accounts of the kind billing requires, and no distinction between customer types.

By the time monetisation was attempted, the product, the audience and the infrastructure had all been shaped around not doing it.

The conversion attempt

It went how these generally go.

We introduced a paid tier. Conversion was a small fraction of what the plan assumed.

The users who had arrived for a free product were, unsurprisingly, not interested in a paid one, and the response to introducing charges was substantial complaint.

We then made the mistake of restricting things that had previously been free, which converted a proportion and produced a strongly negative response from the rest, including people who had been advocates.

Growth stalled, because the free product that had driven signups was now worse.

What would have been different

The counterfactual I have thought about since.

Charging from the beginning, even a small amount, would have produced a smaller and completely different user base — people with a problem worth paying to solve.

It would have provided the only reliable signal about value, which is whether somebody will pay. Every other signal is a proxy and most proxies are optimistic.

And it would have shaped the product toward the users who mattered rather than toward volume.

The counterargument, which is genuine, is that some products need scale before they are valuable at all, and charging early prevents reaching it. That is true for a specific and narrow category, and it was not us.

The advertising fallback

The plan that existed implicitly and was never examined.

The assumption was that a large audience could be monetised through advertising if nothing else worked.

What that ignores is that advertising revenue per user is small, varies enormously by audience and context, and requires scale far beyond what we had to produce meaningful income.

It also changes the product fundamentally, since an advertising business optimises for time spent rather than for problem solved, and those pull in opposite directions.

Several products have made that transition successfully and the number that assumed it as a fallback and failed is much larger.

The question I would ask now

Before building anything: who pays, how much, and why would they.

Not as a projection. As a conversation with somebody who has the problem, asking directly whether they would pay and how much, before the product exists.

The answers are frequently discouraging, which is the point. Finding out early is cheap and finding out after two years of growth is not.

And if the honest answer is that nobody will pay but the audience will be valuable somehow, that is a specific and difficult business to be in, and it should be entered deliberately rather than by default.

What happened to it

For completeness, since these accounts usually stop before the ending.

It was wound down, the users were given notice and an export, and a small part of the technology was sold.

The notice period and the export mattered more than anything else at that stage, and doing it properly is the one decision from the whole episode I would repeat unchanged.

People who have built something on your product deserve time and their data, and the number of shutdowns that fail to provide either is the reason many users are wary of small products generally.

The signals that were visible earlier

Looking back, the evidence was available well before the conversion attempt.

Usage was broad and shallow. Many accounts, very few using it repeatedly over months.

Nobody had ever asked to pay, which sounds like a low bar and is a genuine signal when a product solves something valuable.

Support requests were about the product being free rather than about doing more with it.

And no user had ever recommended it to somebody in a professional context, which is the pattern you see when something is useful rather than merely pleasant.

All of that was in the data a year before anybody looked at it as evidence of anything.