Notebook · 29 May 2026 · 5–6 min read · Alpesh Patel

Why pay-once works at our scale.

Every time a subscription-business person sees the Keepwright price, they ask the same question. It’s a polite version of aren’t you about to go broke?. Here’s the unromantic answer.

The question, said straight

The polite version is “won’t you need recurring revenue to stay alive?”. The honest version is the one that gets asked when the polite version has been answered: if you only sell something once, what stops the line going to zero?

It’s a serious question. We’d be lying if we said it never crosses our minds. So here is the actual answer, in three parts, plain enough to argue with.

Part one: the overhead is small on purpose

Keepwright doesn’t have an office. It doesn’t have a sales team. It doesn’t have a customer-success organisation, a board, an investor relations function, a marketing department, or a quarterly all-hands. It has one Director who writes the code, ships the apps, runs the support inbox, and reads every App Store review.

That sounds romantic. It’s actually arithmetic. Most software companies you’ve heard of charge what they charge because they have to. A team of 30, an office in central London, a Series A burn rate, and a board that wants to see ARR growth — those things compose into a per-customer cost that needs to be paid, somehow, every month, forever. The subscription model isn’t arbitrary. It’s a consequence of the cost structure.

Take away the team, the office, the burn, and the board, and the per-customer cost collapses. What’s left is the marginal cost of one more user installing the app and (sometimes) opening it for support. That number is genuinely close to zero. Which means a one-time price of fifteen pounds isn’t a discount — it’s a real margin.

Part two: the apps are durable on purpose

The second answer to “what stops the line going to zero?” is that the apps are built to keep working without us behind them.

This is the part that surprises people, so let’s be specific. Keepwright apps work offline. They have no account system. They have no cloud back-end we have to keep paying to operate. There is no server in our infrastructure whose uptime determines whether your trips get captured today. If Keepwright went quiet for a year — if the Director was hit by a bus — your Mileage Tracker would still capture every business trip you made tomorrow, still calculate the right figure for your country, still export to PDF.

That’s not a happy accident. It’s the most important architectural decision we’ve made. It means we owe you years of operation but it costs us nothing per user per month to give them to you. The cloud-bill cost of one user is zero. The cloud-bill cost of a hundred thousand users is also zero, because we built it that way.

Compare that to a tracker that runs on someone else’s SaaS infrastructure. Every business day, every active user, there’s a database query, a sync call, a notification ping. Each one rounds to a fraction of a penny but the fractions compound. The subscription pays the fractions. Without the subscription, the company would lose money on every user every month.

We don’t have that problem. So we don’t need the subscription.

Part three: distribution, not retention

This is the part the venture-backed software industry doesn’t want pay-once founders to think about, because it’s the part where the maths actually flips.

A subscription business at £85/year needs a customer it can keep. The whole machine is tuned around retention. The product, the emails, the help articles, the discount-just-for-staying offers — all of it is engineered to stop people leaving. Because if they leave in year two, the lifetime value is £85 instead of £425, and the cost of acquiring them in the first place no longer makes sense.

A pay-once business doesn’t need a customer it can keep. We need a customer it can find. Once. The unit economics flip from retention to distribution. The question stops being how do we get this person to stay another twelve months and starts being how do we let one more person know we exist.

Now, distribution isn’t free either. Search ads cost money. App store ranking is a brutal lottery. Word-of-mouth takes time. We’re not pretending it’s easier than retention. We’re saying it’s a different game, and it’s a game where the playing field is more level for a small studio than the retention game is. The big subscription companies don’t out-distribute us. They out-retain us — and we’ve already stepped out of that contest.

Where the model breaks

Pay-once isn’t a magic spell. There are three obvious places it falls over, and we should say them plainly.

• Categories that genuinely need a server. Anything with real-time collaboration, anything that syncs across multiple users, anything that needs a database we own. Mileage tracking doesn’t. Plenty of other categories do, and they’re not our market.
• Categories that demand constant change. If a regulator alters the rate or rule annually and we’re forever shipping new builds for free to existing buyers, the marginal cost of one extra year of ownership starts to add up. We’re committed to absorbing reasonable updates inside the original purchase. The day “reasonable” becomes “ruinous”, we’ll have to say so. We don’t expect that day.
• Companies that need venture-style returns. We don’t. Keepwright is a UK Ltd with one Director, no outside capital, and no exit number we’re building towards. The plan is to ship apps people keep and be paid for it.

So why are we confident?

Because the maths is honest about itself. We’ve picked a category — small mobile utility apps — where the marginal cost of a user is genuinely close to zero, where the apps can run for years without us behind them, and where the buyers happen to be the kind of people who would prefer to pay once if you let them.

The model isn’t for everyone. It probably isn’t for most things. It’s for this thing, and the next two or three things like it. Beyond that, we’d be reaching.

We’d rather build small and right than big and rented.

Previous: The real maths → · Back to Notebook