Return on investment (ROI) compares the financial benefit of a project with its cost. An estimate is useful for a decision, but it is not a guaranteed result.
Start with a baseline
Record what happens today: time spent, costs, errors and delays. Estimate what might change, how long it will take to see a benefit and how you will measure it.
Count benefits carefully
Reduced work does not automatically mean lower payroll costs: people may use the time elsewhere. Count a cash saving only if spending actually falls. Faster invoicing may improve cash flow without increasing revenue. Avoid counting the same benefit twice as both time saved and extra capacity.
Include the full cost
Allow for setup, software and usage charges, connections, testing, training, staff time, support and ongoing maintenance. Compare costs and benefits over the same period.
A worked example
Suppose a project costs £10,000 to set up and £2,000 to run in its first year. If measured, non-overlapping financial benefits in that year are £18,000, ROI is (£18,000 − £12,000) ÷ £12,000 = 50%. This is an illustration, not a promise. If benefits are only estimated, label them as estimates and check them after launch.
Make a decision
Consider benefits that are hard to price, such as staff experience, as well as risks. Ask a finance professional to check assumptions where a decision is significant.
Compare the estimate with the trial
Write down the assumptions behind the estimate: how often the task occurs, what checking is needed, when costs arise and which financial benefit is expected. Use a consistent period for costs and benefits, and keep non-financial benefits visible without assigning them an unsupported cash value.
After a trial, replace estimates with measured results where possible. Compare a normal case and a less favourable case, such as more corrections or higher usage charges. A finance professional can help assess the assumptions for a significant decision.
Use the AI business case guide to capture the wider decision.
