← All capabilities
ROI & measurement
Phase: ImproveProve with numbers that AI saves or earns more than it costs.
ROI measurement connects AI to the business: how much it costs to operate and how much it returns in time, cost or revenue. It’s the capability that decides what to scale, what to switch off and what budget to defend to leadership.
What it includes
- Baseline before automating: cost and time of the manual process.
- Total cost of operating the AI (models, integration, supervision, maintenance).
- Return per process: time saved, cost avoided or revenue added.
- Portfolio decisions: which agents to scale and which to switch off for underperforming.
When you need it
- Leadership asks what the AI investment returns and there’s no clear answer.
- You have several agents and don’t know which are worth it.
- AI cost grows and isn’t tied to any measurable result.
How it’s measured
- ROI per process (return / total cost to operate).
- Cost per resolved case vs. cost of the manual process.
- Payback: time to recover each agent’s investment.
Common mistakes
- Counting gross savings and ignoring the cost of supervision and maintenance.
- Having no baseline, so no improvement can be proven.
- Keeping agents alive out of inertia even when they don’t perform.
FAQ
- How do you measure ROI if the manual process was never measured?
- You estimate a baseline before automating, even a rough one, and measure the AI process rigorously. With no starting point, any return figure is an opinion.
- Is AI’s cost just the models?
- No. The real cost includes models, integration, human supervision and maintenance. Ignoring those line items makes a project look profitable when it isn’t.
Related capabilities
Missing any of these?
Start with an assessment: which capabilities you have, which you’re missing, and where it pays off most to start.
Talk to us →