I accidentally built a feature for twelve people

Render My Email was supposed to become a business.

Instead it became market research.

Expensive market research.

Well.

Approximately sixteen dollars a month.

The premise seemed solid.

Developers build HTML emails.

HTML emails break.

Therefore developers require software that tells them when HTML emails break.

This is known in the startup industry as:

looking at a problem for roughly twelve seconds.

The Product

The software worked.

This is unfortunately relevant.

You could render emails across clients.

Preview dark mode.

Compare outputs.

Catch layout regressions.

The infrastructure behaved.

The queues behaved.

The billing behaved.

The codebase behaved.

The customers behaved exactly like customers who had no intention of becoming customers.

This distinction would later prove important.

The Investigation

Naturally, I assumed the problem was the product.

Because I am a software engineer.

When software engineers fail to find users, we never suspect the market.

We suspect CSS.

Perhaps the onboarding was wrong.

Perhaps the button needed more contrast.

Perhaps I should rewrite the rendering pipeline.

Maybe Redis.

Maybe Kafka.

Maybe both.

At no point did I consider the revolutionary hypothesis that nobody particularly cared.

This is why engineers need adult supervision.

The Competition

There are already companies solving email rendering.

They have been doing this for years.

They are quite good at it.

This did not discourage me.

Because I had committed the classic engineering error.

I believed technical superiority produced commercial superiority.

History contains surprisingly little evidence for this.

Internet Explorer once had ninety-five percent market share.

PHP still exists.

Reality has never cared about architecture.

The Actual Customer

Eventually I started talking to people.

An underrated debugging technique.

The conversations went approximately like this.

"Do your emails ever break?"

Yes.

"How often?"

Occasionally.

"What do you do?"

Open Gmail.

Then Outlook.

Then ship it.

This is horrifying.

It is also apparently sufficient.

The companies that genuinely care already have enterprise contracts.

The companies that do not care were not waiting for my landing page.

I had accidentally positioned myself between two groups that were perfectly happy.

The Market Size

Every founder claims their market is enormous.

Mine technically was.

Millions of developers write emails.

This is true.

Millions of developers also occasionally use regular expressions.

Nobody has built Regex-as-a-Service.

Frequency matters.

Pain matters.

Budget matters.

An inconvenience repeated twice a year is called life.

Not SaaS.

The Emotional Support Dashboard

Naturally, I added analytics.

Every unsuccessful product eventually develops analytics.

Not because users need them.

Because founders do.

The dashboard informed me that nobody was using the dashboard.

I admired the recursion.

At one point I had more graphs describing user behaviour than users exhibiting behaviour.

This is difficult to scale.

The Numbers

The infrastructure cost around sixteen dollars a month.

This sounds responsible.

Until you realise the revenue was approximately the infrastructure cost.

With the sign removed.

Technically the unit economics worked.

If I acquired customers.

This turns out to be an important dependency for recurring revenue.

The Real Bug

The software did exactly what I designed it to do.

I had simply solved a problem that most people encounter with the emotional intensity of forgetting where they parked.

Annoying.

Recoverable.

Not subscription-worthy.

I kept trying to optimise the implementation.

The implementation was innocent.

The product had perfect product-market fit.

For a market that barely existed.

Archive

Eventually I archived the repository.

This felt dramatic.

It was not.

The code still works.

The product still works.

The business model politely declined to participate.

GitHub has a button called Archive repository.

There should be another one called:

Congratulations. You only wasted three weeks instead of three years.

I would press that much more often.

Render My Email was not a failure.

It was a successful experiment that disproved my own assumptions before my bank account had to.

That is probably the highest ROI feature I have ever built.

Still here?

Bold choice.

The measured version — with fewer jokes and more principles — lives on engineering.