Build vs buy calculator
Enter what you pay now and what it costs your team to work around it. See the three-year comparison against building something custom, and the month it breaks even.
Free, runs in your browser, nothing stored. Updated September 2026.
Licences you pay for today
Your current per-seat price
Re-keying, exports, checking what the tool gets wrong
Salary plus employer costs
A focused custom tool is typically £15,000–£50,000
Industry norm is 15–25%
Building looks worth it
Building breaks even at month 22 and saves £17,840 over three years. At that margin it is worth getting a real quote.
Buy · 3 years
£57,840
£1,000 licences + £607 workaround, monthly
Build · 3 years
£40,000
£25,000 once + £417/mo upkeep
Break-even
Month 22
When building becomes the cheaper option
Assumes subscription and workaround costs continue unchanged, and that a build removes the workaround entirely. Both are simplifications — a build usually removes most of it, not all, and licence prices tend to rise. Treat the break-even month as an order of magnitude, not a forecast.
Why the workaround hours matter more than the licence fee
Most build-versus-buy comparisons put the subscription on one side and the build cost on the other. That understates buying, because it ignores the largest recurring cost of a product that nearly fits: the manual work it forces on your team.
Someone re-keys data between two systems. Someone exports to a spreadsheet every Friday. Someone double-checks the thing the software reliably gets wrong. Those hours are permanent and they rise with wages. Four hours a week at a loaded £35 an hour is roughly £7,300 a year — frequently more than the licences themselves.
Why seat count usually decides it
Per-user pricing scales with hiring. A custom build does not. That is why a company that was right to buy at ten people is often right to build at forty, without anything about the software changing. If you are growing, run this calculation again at the headcount you expect in eighteen months, not today's.
What this calculator deliberately does not do
It does not account for the risk of building, and that risk is real. A build can overrun, arrive late, or end up dependent on one person who then leaves. Those are the genuine reasons to keep buying even when the arithmetic favours building.
If the break-even lands beyond about two years, buying is usually still the better call — the margin is too thin to justify the risk. Read the full build vs buy guide for the decision framework, and the AI-specific version if the tool you are considering involves models.
Common questions
How do you calculate build vs buy for software?
Take the subscription cost (seats × price × 36 months) and add the cost of manual workarounds the tool forces on your team, at loaded hourly rates. Compare that against the build cost plus three years of maintenance at 15–25% of build per year. The workaround cost is the figure most comparisons omit, and it is frequently larger than the licence fee.
At what point does custom software become cheaper than SaaS?
It depends mostly on seat count, because per-user pricing scales with headcount while a custom build does not. As a rough guide, a £25,000 build against a £40 per-seat product breaks even inside three years at roughly 20–25 users once workaround time is included, and sooner above that.
What maintenance percentage should I use?
The long-standing software industry norm is 15–25% of the original build cost per year, covering hosting, security patching, dependency updates and small changes. Use the upper end for anything built on third-party AI model APIs, because those dependencies change faster than conventional software.
Does this calculator work for AI tools too?
Yes, with one adjustment: add your expected model API usage to the maintenance figure. For most business workloads that is modest, but it is not zero, and it scales with usage rather than with headcount.