Free tool
Business Case Calculator
Take an initiative you need to justify right now. Enter what the rollout costs once, what it costs every year afterwards, and the benefit you expect per year. This page puts the three numbers side by side and shows how far your assumptions can be wrong before the case stops working.
What belongs in the running cost?
What belongs here: hosting and infrastructure, licences, and for AI initiatives also API, token and inference costs, support and maintenance, model and version updates, governance effort (sampling, audits, evidence), and a share of business-team time if someone looks after the system on an ongoing basis.
What does not belong here: one-off costs up to go-live. That goes in the field above. Training at rollout is project cost; training new joiners from year two onwards is running cost.
If you have no figure, make a rough estimate rather than leaving it out. A zero in this field is almost always wrong, and it shifts the result more than any imprecision would.
What counts as benefit?
Three kinds, in descending order of how well they can be backed up: cost avoided (a mistake that no longer happens demonstrably cost X), time saved and additional revenue. Be conservative on the last one, because reviewers attack it first.
The uncomfortable question about time saved: two person-days a week become a benefit only once someone decides what happens to that time: a vacancy goes unfilled, a backlog gets cleared, an extra order is accepted. Without that decision the time quietly disperses into everyday work and shows up in no figure at all.
A year, not the first year. Enter the benefit in its steady state. The ramp-up during year one is deliberately left out here. The limits are shown below the result.
The arithmetic is the easy part
What comes after is harder: describing the benefit so that it becomes measurable, having a baseline before you start, and proving after go-live how much of it actually arrived. That is exactly the course “Making the case”, which launches on 16 November 2026: making benefit measurable, taking a baseline, proving it after go-live.
Browse all courses →