Finance · 7 min read
Building an ROI case your finance lead will accept
Finance teams are right to be sceptical of AI claims. This is how to build a return-on-investment case that survives their questions.
Why do AI business cases get rejected?
Usually because they borrow figures from vendor marketing, ignore running costs, or count saved hours as saved money when nobody will actually be let go or redeployed. Finance leads spot this quickly. A smaller, honest number is far more persuasive than a large one that falls apart under questioning.
What should the case include?
Five parts, each on one page or less:
- —The current cost of the process, measured from your own data.
- —The full cost of the change: build, licences, usage fees and staff time to adopt it.
- —Hard savings: costs that will genuinely stop, such as agency fees or overtime.
- —Capacity and revenue gains: hours freed and what they will be used for, plus revenue no longer lost.
- —Payback period and how results will be measured after launch.
How should freed-up time be valued?
Be specific about what happens to the time. If the hours go to selling, serving more customers or work you currently turn away, you can value them by the revenue they bring. If they simply disappear into the day, count them as quality of work, not money.
Worked example: freeing 25 hours a week of a sales coordinator's time, redirected to following up warm leads, is worth more than the coordinator's wage for those hours if it lifts conversion even slightly.
How do you prove it afterwards?
Agree two or three measures before you start, such as response time, cost per enquiry or conversion rate, and record them for a month beforehand. Review them at 30, 60 and 90 days. This is what turns a projection into a result your finance lead can report.
Key takeaways
- ✓Use your own figures, never vendor averages.
- ✓Count running costs and adoption time, not just the build.
- ✓Value freed time only by what it will actually be used for.
- ✓Agree the measures before launch and review at 30, 60 and 90 days.
Frequently asked questions
What is a typical payback period for AI automation?
For well-chosen operational processes, often within three to twelve months. If a case needs more than eighteen months to pay back, it is worth questioning the scope.
Should AI running costs be included?
Yes. Usage fees, hosting, licences and maintenance are ongoing and must be in the case. They are usually modest, but leaving them out damages trust.
Can I estimate ROI before speaking to anyone?
Yes. Our ROI calculator on the homepage gives a first estimate from your own volumes and costs in under a minute.
Next step
See what this could be worth in your business.
Read about our Workflow automation work, map a process with the Blueprint tool, or talk to us directly.
