How to calculate the return on an automation project

The basic calculation

Start with the benefit: the annual cost of the current manual process, using hours x loaded hourly cost x frequency, plus errors and delays. Then estimate how much of that the automation will realistically remove. It will rarely be everything, because exceptions still need people.

Net annual benefit = benefit achieved - annual running costs. Payback period = one-off build cost / net annual benefit. Simple return = (net annual benefit x years - build cost) / build cost, over a period you choose, such as three years.

Illustrative worked example

All numbers here are round and made up to demonstrate the method. Suppose a process costs £12,000 a year in staff time and rework. The automation removes about 75% of that, so the benefit is £9,000 a year. Running costs (licences, hosting, support) are £2,000 a year, so net benefit is £7,000.

If the build cost is £14,000, payback is £14,000 / £7,000 = 2 years. Over three years, net benefit is £21,000, minus the £14,000 build, leaving £7,000, a simple return of 50%. Your own figures, and your own costs, will differ.

Do not forget the running costs

• Software licences and usage-based charges for any AI or cloud services.
• Hosting and monitoring.
• Support and maintenance when connected systems change.
• Staff time to review exceptions and AI outputs.
• Periodic reviews, retraining or prompt and rule updates.

Allow for risk

Projects can deliver less than planned. Test your case by lowering the benefit and raising the cost, for instance assuming only half the time saving, and see whether it still makes sense. If a project only works under optimistic assumptions, treat it with caution or reduce the scope.

Also consider the risk of not acting: growing volumes, key-person dependence and compliance weaknesses. These are hard to price but belong in the discussion.

What to measure

1. Record a baseline before you start: time per item, error rate and turnaround.
2. Define the same measures after launch and check them at set intervals.
3. Track exceptions: how many items still need manual handling and why.
4. Capture staff and customer feedback, not just numbers.
5. Compare actual running costs with your estimate.
6. Review the result with the sponsor and decide what to improve next.

Benefits that are not just savings

Savings are not always cash. Many projects free up capacity, so the same team handles growth without extra hiring, or they speed up responses to customers, or improve consistency and audit trails. Put a value on these if you can, but be honest about which are estimates. A clear, cautious business case is more persuasive to a board than an ambitious one.

Frequently asked questions