Is automation worth it? Start with the work and the numbers
A recurring task is a useful starting point for automation. Measure the work, include what remains, and compare a few plausible outcomes before committing to a solution.
Measure the work that happens today
Choose one task with a clear beginning and end. Count how often it happens in a representative month and measure the hands-on minutes per task. Include checking, correcting and handing work over. Waiting for a system is only working time if someone must actively attend to it.
The GOV.UK discovery guidance recommends understanding the problem and current cost before committing to a service. Apply that principle here: a clearer form, a shared template or an existing system feature might remove the problem without a new integration.
Illustrative baseline: 200 tasks per month at 6 minutes each take 20 hours. This is an example calculation, not a typical saving or a client result.
Count review and exceptions before claiming a saving
Automating part of a task rarely removes all the work around it. Estimate the time still needed to check results, resolve exceptions, correct mistakes and handle cases outside the automation. Your saving percentage should already include those recurring steps.
Keep additional monthly monitoring and maintenance separate. Enter that as upkeep hours in the calculator. Do not subtract the same review time in both the saving percentage and upkeep. If upkeep exceeds the recovered hours, the net result is negative—and should stay negative.
Compare three assumptions, not three promises
Use a conservative, expected and optimistic percentage for the same workload. The calculator starts at 25%, 50% and 75% only to make the comparison easy to explore. These are illustrative values, not industry benchmarks. Replace them with observations from your task or pilot.
The calculator holds volume, cost rates and performance constant. It does not model a gradual rollout, seasonal changes, financing, tax effects or discounting. A result is conditional on those assumptions; it is not an instruction to automate.
Recovered capacity is not automatically cash saved
The UK Government Efficiency Framework distinguishes benefits that directly reduce spending from benefits that improve output without reducing spending. That distinction matters for a business too. A team gaining ten hours may use them for other useful work while its payroll remains unchanged.
In this calculator, net hours multiplied by your hourly labour cost give an estimated capacity value. Subtract monthly running costs to obtain the net monthly value. Dividing setup cost by a positive net monthly value gives capacity-based payback. It assumes that the recovered time can be used productively; it is not a forecast of money returning to your bank account.
With the 20-hour baseline, 50% time savings recover 10 hours. Subtract 2 hours of upkeep: 8 hours remain. At €50 per hour, their capacity value is €400. After €100 in monthly running costs, the net value is €300. A €1,200 setup cost corresponds to 4.0 months of capacity-based payback.
Keep unknown costs visible
Include implementation, preparation, testing and training in setup cost. Include subscriptions, usage charges and other recurring expenses in monthly running cost. Upkeep labour is already deducted through hours; do not charge it a second time. Use the same currency and cost basis throughout.
Blank costs remain unknown. You can still compare hours without supplying money figures. Enter zero only when zero is the intended assumption. If the monthly net value is not positive, there is no break-even under the stated assumptions. The calculator does not hide that result.
Test the assumptions with a small pilot
GOV.UK’s alpha guidance recommends testing the riskiest assumptions with prototypes. For this task, compare the same kinds of cases before and after the change. Record time, review effort, errors and upkeep. Replace estimates with measurements, then recalculate all three scenarios.
A checklist to reuse
- Choose representative cases, including exceptions.
- Agree on acceptable quality before measuring speed.
- Name an owner for monitoring and maintenance.
- Decide how the recovered hours would actually be used.
- Keep the assumptions with the exported result.
Sources and upkeep
These sources inform the guidance above. The examples and checklists are modeleven’s editorial recommendations, not promises of results. Sources reviewed 8 September 2026.