What ROI really measures
ROI compares the value an automation creates against everything it costs — to build and to keep running. A common mistake is comparing gross time saved against only the model's token cost, which ignores build effort, ongoing operations, exceptions, and the fact that saved hours do not automatically become cash. A credible estimate accounts for all of it.
Framing: net benefit ≈ (value of reliably automated work) − (build cost + recurring operating cost + oversight cost). Every term deserves a range, not a point estimate.
The inputs that matter
- Baseline effort: how long the task takes today and the fully loaded labour cost of the people doing it.
- Automatable share: the realistic percentage the system handles end-to-end without a human.
- Exception rate: the cases that still need people — these erode savings quickly if underestimated.
- Recurring operating cost: inference, tools, hosting, monitoring, and maintenance (see the agent costs guide).
- Build and change-management cost: engineering, integration, training, and process change.
Safeguards on your assumptions
Optimistic assumptions are where automation business cases go wrong. Apply these safeguards deliberately:
- Adoption: assume ramp-up, not day-one full usage; benefits arrive gradually.
- Exception handling: the last 10–20% of cases often costs more per unit than the easy majority.
- Quality assurance: budget for ongoing checking and correction, especially early on.
- Change management: workflows, training, and buy-in take real time and money.
- Recurring cost drift: prompt growth and volume increases raise operating cost over time.
Worked scenario (illustrative)
Illustrative assumptions — not a quote: a document-intake process handles 10,000 items/month at ~12 minutes of loaded labour each. An automation reliably handles 70% end-to-end; the remaining 30% still needs a person, at reduced time.
The honest projection nets out the 30% exceptions and the recurring operating cost — not the headline "70% automated × full labour rate," which would overstate savings. It also treats the saved time as capacity that must be redeployed to become real value. Put your own figures and cost assumptions into the AI Automation ROI Calculator to see savings and payback as a range.
Report ranges, not guarantees
Savings from AI automation are estimates, not guarantees. Present a conservative, expected, and optimistic case, state your assumptions plainly, and revisit the numbers with real data after launch. A projection you can defend beats a single impressive figure you cannot.
Pricing and capabilities change. Provider rates, model line-ups, and features are updated by the providers over time. Any figures on this page are clearly labelled illustrative planning assumptions, not quotes. Official sources are checked on a recurring schedule. Unusual or structural changes are held for human review. See the Pricing Sources & Data Status page for verification dates, and always confirm current rates on the official provider pages below.
Official sources
Confirm current pricing and capabilities directly with each provider: