Two totals, one comparison
The build side adds the one-off construction cost to the annual maintenance burden multiplied by the years you will run it. The buy side adds the implementation cost to the annual licence multiplied by the same years. Putting both on one screen matters because build-versus-buy debates are rarely lost on arithmetic. They are lost on estimates, and the estimates arrive pre-loaded: the build figure is produced by the engineers who want to build, and the buy figure by the vendor who wants to sell.
A five-year example
An internal pricing engine is estimated at €500,000 to build, with €80,000 a year to maintain. The commercial alternative costs €150,000 to implement and €120,000 a year to licence. Over five years: build totals €500,000 + (€80,000 × 5) = €900,000, while buy totals €150,000 + (€120,000 × 5) = €750,000. Buying is €150,000 cheaper on these inputs. Shorten the horizon to three years, though, and the gap narrows to €740,000 against €510,000, and every one of the four estimates deserves a fight before either total is believed.
What neither column contains
The totals compare cash. The decision is mostly about things that are not cash.
- The opportunity cost of engineering time. Every quarter your best engineers spend rebuilding a solved problem is a quarter not spent on whatever actually differentiates you, a cost that never appears in the build column.
- The buy option's exit. Data held in proprietary formats, per-seat pricing that ratchets with success, and the migration project you will fund the day the relationship sours.
- The build option's key-person risk. Systems have authors, and when the author leaves, the maintenance line stops being a number and becomes a search.
- Estimate decay on both sides. Built systems accrete scope and their maintenance grows; vendor pricing at year six rarely resembles the pricing that won the deal.
Where the comparison ends
Five inputs cannot carry strategy. The calculator will not tell you whether the capability is differentiating enough to own, whether your organisation can genuinely sustain software over a decade, or whether the vendor will exist at the end of the term. When the two totals land close together, that is the signal to stop arguing about the arithmetic and start arguing about exactly these questions.
