The AI ROI Calculator Guide
Most "AI ROI" projections are made up. This is how to build one that survives a CFO and a board meeting — for less than a day of work.
If you've ever sat through a vendor pitch with a slide claiming "10× ROI in year one," you've felt this pain. The numbers are derived from a magical formula nobody can explain, and yet you're being asked to defend them to your CFO. This playbook is the antidote: a model you build yourself, with assumptions you own, that you can stress-test.
We hand the underlying spreadsheet to every consulting client at engagement kickoff. It's the same model we use to decide whether our own AI investments are worth making.
Chapter outline
- Why most AI ROI models are fiction — the three common math errors that inflate AI savings claims by 5–20×.
- The three-tier ROI framework — direct savings (clearest), capacity gain (real but harder to value), and option value (hardest, often biggest).
- The spreadsheet structure — inputs, assumptions, sensitivities. Color-coded so your finance team will actually use it.
- Estimating realistic time-savings — methods to measure baseline (timesheets, screen recording sampling, interviews) and post-deployment.
- Pricing risk and downside scenarios — what does it cost when the agent is wrong? What does it cost when the project ships late?
- The sensitivity table — the one chart that convinces CFOs more than any other.
- Worked examples — three real projects with the actual ROI math (anonymized).
- How to present this to leadership — the slide template we use to brief boards.
TODO: expand — the spreadsheet template and worked examples will be included in the next iteration. Drop your email above to receive it.
Book a call for help building your model.