Skip to main content

AI glossary Strategy & Business

How do you measure the ROI of AI?

ROI (Return on Investment)

← All glossary terms
In short

Return on investment: the net gain from an investment relative to its cost; for AI projects, the key measure for justifying and sustaining the spend.

Explained

Return on investment (ROI) compares what an investment returns with what it costs, usually as net gain divided by cost. For AI projects it is the measure that decides whether a pilot gets funded, scaled or stopped. Showing it means linking the system to measurable value, such as cost savings, revenue growth, faster processes or fewer errors, and counting the full cost, not just the model. Because many AI benefits are indirect or arrive over time, a clear baseline and agreed metrics before launch matter more than the formula itself.

How to calculate it

The basic formula is ROI = (benefits - costs) / costs, expressed as a percentage over a defined period.

  • Benefits: hours saved multiplied by their cost, extra revenue, lower error or rework rates, faster cycle times, avoided risks.
  • Costs: model and infrastructure usage, licences, integration and data work, the team that builds and runs it, change management and training.
  • Period: AI systems have running costs, so compare benefits and costs over the same time frame, not a one-off build against years of savings.

Why AI ROI is hard to prove

  • There is no baseline: the process was never measured before the AI arrived.
  • Benefits are spread out: a few minutes saved per person per day is real but hard to see in the accounts.
  • Usage costs grow with adoption, so a pilot's economics can change at scale.
  • Some value is strategic or defensive, such as quality, speed or risk reduction, and does not fit neatly in a formula.

Good practice

  • Agree the business metric and the baseline before building.
  • Track adoption: a system nobody uses returns nothing.
  • Measure running cost per transaction or per user from the first day.
  • Review ROI at fixed points and be ready to stop use cases that do not pay off.

Frequently asked questions

What is ROI in simple terms?

ROI is how much you get back for what you put in. If a project costs 100 and brings benefits worth 130 over the period, the net gain is 30 and the ROI is 30 per cent.

How do you calculate the ROI of an AI project?

Estimate the benefits the system brings over a period (time saved, extra revenue, fewer errors), subtract the full cost of building and running it in that same period, and divide by the cost. The hard part is not the arithmetic but measuring a reliable baseline and real usage.

Why do many AI projects fail to show ROI?

Common reasons are a missing baseline, low adoption, benefits that are spread too thinly to measure, and running costs that were not planned. Projects that start from a measured business problem show ROI more easily than those that start from the technology.

What costs should be included in AI ROI?

Include model and infrastructure usage, licences, data preparation and integration, the people who build, run and supervise the system, security and compliance work, and training for users. Leaving out running costs is the most common mistake.

What is a good ROI for AI?

There is no universal figure. A good return depends on the risk of the project, the time horizon and what else the money could be used for. Many organisations set a target per use case and a deadline to reach it.

Decision guides that use this term

How to put it to work

Where it shows up at Thinkia

Products

Further reading

Sources

Updated: 2026-10-21

Want to see how this applies to your organisation?

Talk to an AI Expert