All writing

Should you build your own software?

Guide3 min read

Should you build your own software?

Guide · November 20, 2026 · 3 min read

Most businesses should not build their own software. A few should, and for them it becomes the most valuable thing they own. The difference is rarely size. It is whether the way you work is part of why customers choose you.

This guide is for an owner or manager weighing a bespoke system against the products already on the market. It is drawn from systems we have built for businesses of ten to a few hundred people.

Start with what you have

Nobody begins by commissioning software. You choose the product everyone in your trade uses, it covers most of what you need, and it costs less than a developer. That is the right way to start.

What changes is everything around it. You hire, and the per-seat bill grows. The way you actually work drifts from the way the product assumes you work, and people fill the gap by hand: the spreadsheet beside the system, the step everyone knows to do that the software has never heard of.

Those workarounds are worth noticing. They are the outline of the system you actually need.

When building pays off

Bespoke software earns its cost when the gap between how you work and how the product works is large, permanent and expensive.

  1. Your process is part of what makes you different, and the product forces you to work like everyone else
  2. Important work lives in spreadsheets beside the system, maintained by one or two people
  3. You pay for several products that do not talk to each other, and someone re-types between them
  4. Per-seat fees grow with every hire, for a product that still does not fit
  5. You have outgrown the product, and the vendor's roadmap is not yours

When it does not

Building is the wrong answer more often than it is the right one. Keep what you have if any of these is true.

  1. Your process is standard. Software built for thousands of businesses is cheaper than software built for one.
  2. You are still changing shape. Build once the way you work has settled enough to write down.
  3. The fee is small. A modest subscription is rarely where the money goes.
  4. Nobody inside will own it. A system of your own needs someone to answer for it, even when someone else maintains it.

What it really costs

The build is the visible cost. The honest comparison includes everything on both sides, over several years.

Renting a productOwning a system
To startLow: a subscription and some setupHigher: a specification and the build
Each yearSubscription fees, rising with headcount and price increasesHosting and upkeep; changes priced when you want them
The workaroundsPaid in staff time, every weekBuilt into the system
When you growMore seats, more feesUsually nothing extra
If you leaveYour data, exported in the vendor's formatEverything: the code, the infrastructure, the credentials

At a handful of seats, renting usually wins. As headcount and workarounds grow, the lines tend to cross. Put your own numbers in before deciding.

How to start without betting the business

The risk in bespoke software is not that it costs money. It is spending that money before anyone has checked that the thing being built is the right one.

  1. Pick the one process that costs the most hours, not the whole operation.
  2. Have it written down, in your words, before anything is built, and correct it until it is right.
  3. Build that piece, use it for real, and keep the old system running beside it until the new one has earned trust.
  4. Make sure everything is yours from the start: the code, the hosting, every login.
  5. Only then decide what comes next.

Done this way, the worst outcome is a small system you own that solves one real problem. The best is the start of the software your business runs on.

Start with one problem

If you have a process like that in mind, bring it to us as you would describe it to a colleague. We will tell you whether it is worth building, and say so plainly if it is not.

Start with one problem, or bring us the whole operation.