AI ROI
AI ROI is the return on an AI investment measured against its full cost, including licences, implementation, maintenance and the hours your own people spend. It is only meaningful when calculated on your own figures, because the same application returns very different amounts in two organisations.
The cost side is where it goes wrong
Most AI business cases understate cost rather than overstate benefit. The licence is the visible number and usually the smallest one. What gets left out:
- The hours your own people spend during implementation, which are real even though nobody invoices for them
- Maintenance and adjustment in year two, when the process has changed and the model has not
- The cost of the work that has to happen before AI can help at all, typically making data mean the same thing across departments
- The cost of a tool that is bought and then not used, which is a total loss rather than a partial one
The benefit side needs more than hours saved
Time saved is the easiest benefit to calculate and the least convincing to a board, because saved hours only become money if the hours are removed or redeployed. Benefits that hold up better:
- Fewer errors and less rework, which you can often price from existing correction costs
- Faster response to customers, priced from won or lost orders rather than from a feeling
- Work that becomes possible at all, such as quoting on requests you previously declined
Why a payback period beats a percentage
A percentage invites debate about assumptions. A payback period asks a simpler question: after how many months has this paid for itself on our own numbers. It is easier to defend and easier to disprove, which is what makes it useful.
The calculation belongs in the AI Readiness Scan, on the opportunities the prioritisation matrix ranks highest.