Know the Figures

Build vs Buy Software

The licence looks expensive until you price what building it really costs.

Replaces: Vendor business cases and internal engineering estimates

How to use it

Price both paths across the same period and discount them to today. Building front-loads engineering cost and never fully stops; buying spreads a licence that rises each year. The comparison is only honest once the build estimate carries an overrun allowance and a permanent maintenance line.

Where the number comes from

  • The build estimate is multiplied by the overrun allowance to give a realistic duration, then costed at engineers times loaded cost per engineer.
  • After launch, the maintenance FTE cost continues every year for the rest of the period.
  • The licence rises each year at the inflation rate you set, and implementation is charged once in year one.
  • Buying still carries internal effort — someone administers the tool — which is charged at the same loaded engineer cost.
  • Every year is discounted at your cost of capital, so early build spend is penalised relative to later licence payments.
  • The break-even licence is the first-year price at which the discounted cost of buying equals the discounted cost of building.

What goes wrong

The part most calculators leave out.

  • The overrun allowance is doing most of the work here, and 40% is optimistic by the standards of most software estimates. Try it at 100% and see whether the decision survives.
  • Opportunity cost is not priced. Engineers building this are not building something only your business can build, and that foregone work is frequently worth more than the licence.
  • Time to value is missing entirely. Buying delivers in weeks and building delivers in quarters; if the capability drives revenue, that gap is a real cost this comparison ignores.
  • Build cases routinely underestimate what "finished" means — security review, access control, audit logging, documentation, on-call and the second version users ask for immediately.
  • Vendor risk cuts the other way: price rises above the modelled rate, acquisition, sunsetting, or a roadmap that diverges from your needs. A licence is a dependency as well as a saving.
  • Building can be right for genuine differentiation. If the capability is the thing customers buy from you, this arithmetic is the wrong question.

The seven-month project that is really ten

Three engineers at 145,000 loaded, quoted at seven months, looks like about 254,000 to build against a 90,000 licence — payback in under three years, and the build case writes itself. Apply a 40% overrun and it is 9.8 months and roughly 355,000. Add 0.75 of an engineer forever to maintain it — 109,000 every year — and the ongoing cost alone exceeds the licence. Meanwhile buying carries implementation and a quarter of an FTE to administer. Over five years, discounted, buying wins comfortably, and it would still win at a licence well above the quoted 90,000. The build only looked cheaper while maintenance was invisible.

Questions

What overrun allowance should I use?
Whatever your own last three projects actually ran over by. In the absence of that history, 50–100% is closer to observed software estimation error than the 20% most business cases assume.
Why charge maintenance forever?
Because software does not stand still. Dependencies age, security patches arrive, integrations break, and users ask for changes. Anything you build becomes something you own indefinitely.
When is building genuinely right?
When the capability is a real differentiator, when no adequate product exists, or when the data cannot leave your control. Building to save licence fees on a solved problem is the case that most often disappoints.
Does anything I type get sent anywhere?
No. The whole calculation runs in your browser. Nothing is transmitted, stored, or logged, and there is no account to create.

Last updated .

Put this calculator on your site

Free to embed, on any site, commercial or not. No sign-up and no tracking script — the calculator runs in your reader’s browser exactly as it does here. All we ask is that you keep the credit line.

HTML

The small script resizes the frame as the reader changes inputs. Drop it if your CMS strips scripts — the calculator still works, it will just stay at a fixed height.

Related calculators