Idea Autopsy: Minutely Built a Great Niche Tool, Then Walked Away From It

• Name: Minutely (fictional)
• What it did: Turned recorded meetings into minutes and action lists, starting with construction site meetings
• Lifespan: 2 years, 4 months
• Peak: About 1,100 paying teams
• Cause of death: Left a defensible niche to chase a crowded general market

A very good start

Minutely began with a narrow, muddy, wonderful problem. On construction projects, weekly site meetings decide who fixes what and by when. Somebody, usually the most junior person in the room, writes the minutes, and those minutes become something close to a legal record when things go wrong. Get them wrong and a dispute over a leaking roof can turn into months of finger-pointing.

The founders, one of them a former site engineer, built a tool that recorded the meeting on a phone, transcribed it and produced minutes in the format contractors already used: numbered items, responsible party, due date, all linked to the project's list of open issues. It understood construction vocabulary like snag lists, variations and requests for information, and it flagged any commitment made without a date.

Contractors loved it. Within a year Minutely had a few hundred paying teams and almost no churn. Site managers recommended it to each other at the pub, which is the kind of growth you can't buy.

The fork in the road

Then the investors' question arrived, as it always does: how big can this get?

Construction, the team was told, was a ceiling. Every company has meetings. If Minutely could summarise a site meeting, surely it could summarise a sales call, a board meeting or a parent-teacher evening. So the company rebranded as "minutes for every meeting," stripped the construction language from its marketing and launched a generic version with a free tier.

Sign-ups jumped and revenue grew, slowly. The team celebrated a new peak of about 1,100 paying teams.

Walking into the bundle

While Minutely was broadening, meeting summaries became a standard feature. The video-call and office software most companies already pay for began offering transcripts and summaries built in, often at no extra cost. For a generic meeting, "good enough and already installed" beats "slightly better and another subscription" almost every time.

Minutely's generic customers churned at five times the rate of its construction customers. Worse, the construction customers felt neglected. Their requests for links to drawings and subcontractor sign-off sat for months while the team chased integrations with general sales tools. Then a small rival appeared, another fictional upstart called Siteword, with a construction-only product, and within a year it had poached many of Minutely's original champions.

By the time the founders tried to swing back, they'd lost the niche and never won the mainstream. The company wound down and sold its code to a project-management vendor for a modest sum.

The niche wasn't Minutely's ceiling. It was Minutely's moat, and the company walked out of the castle to fight in an open field.

The lessons

1. You don't have to leave your wedge. Starting narrow is common advice. It's less often said that the narrow thing can be the whole business. Construction is a huge industry. A tool that became essential infrastructure for site meetings could have grown into drawings, inspections and handover documents without ever touching a sales call.

2. Ask what the platforms will bundle. Any AI feature that works on generic data, such as summaries, transcripts or rewriting, is likely to become a default feature in software people already use. Specific workflows, vocabulary and formats are much harder to bundle. Build where the platforms won't bother.

3. Watch churn by segment, not in total. Minutely's blended numbers looked fine for a long time because loyal construction teams masked the leaky generic ones. Churn broken down by segment would have told the story a year earlier.

4. Your best customers are your roadmap. The site managers had been telling the team exactly what to build, and the roadmap went elsewhere.

5. Grow deeper before you grow broader. If investors need a bigger market story, look for neighbouring work in the same industry before you look at neighbouring industries. When the broad market is someone else's default feature, going deeper wins.

If you're building something like it

Before you broaden, run a simple test. Write down the three things your niche customers would lose if you disappeared, and ask whether any general-purpose platform could give them those things within a year. If the answer is no, you have a moat. Don't drain it.