Build the estimate

Use one period and name each assumption.

This is a simple screening calculation, not a forecast of guaranteed savings. For a larger or longer investment, include the timing of costs and benefits, uncertainty and any discounting your finance team requires.

  1. 01

    Measure the current task

    Count completed instances over a representative period and time the handling, corrections and exceptions. Note quiet and busy periods. Use your actual operating weeks when annualising, rather than assuming every week has the same volume.

  2. 02

    Separate money from capacity

    Record spending that would actually stop or be avoided, such as a contractor charge. Record employee time released separately. An hourly valuation can help compare options, but a salary does not fall just because a task takes less time. Do not count both the wage valuation and the same spending reduction.

  3. 03

    Include setup and running costs

    Estimate implementation, data cleanup, testing and training, then subscriptions, usage fees, monitoring and exception handling. Record internal work as well as invoices where relevant. Check what is already included in a supplier quote so costs are not counted twice.

  4. 04

    Calculate ROI, then test payback

    Simple first-year ROI = (first-year benefit minus initial cost minus first-year recurring cost) / (initial cost plus first-year recurring cost) x 100. Simple payback = initial cost / net recurring benefit. Use a consistent basis for benefits and costs, and test lower volume, a slower start and higher support effort.

What remains manual

Price the work that remains.

Automation can remove a task while leaving the awkward cases with the team. Include that remaining work in the estimate.

The Government Efficiency Framework distinguishes reductions in spending from capacity released for other work. That distinction is useful here: four hours returned to a salaried colleague can improve service without changing the payroll.

Check the estimate

Costs and assumptions to write down

Exception handling: how many cases will still need review, and how long will they take?

Operating costs: subscriptions, usage charges and the time needed to monitor or repair the workflow.

Starting period: training, parallel running and a delay before benefits begin.

Change: likely process updates, connector changes and the cost of stopping or replacing the automation.

Decision

A short payback cannot make an unsafe workflow worth building.

There is no universal six-month approval rule. Decide what payback is acceptable for the investment, how reliable the assumptions are and what failure would cost. A required control may be justified for reasons other than financial return.

Fictional worked example

GBP 2,400 setup, GBP 400 net benefit a month

Assume a paid external service costing GBP 500 a month can genuinely be cancelled when the automation starts. The replacement costs GBP 100 a month, including support. Setup is GBP 2,400. All figures use the same cost basis, with no other costs assumed.

Benefits begin immediately and remain constant for 12 months. A delayed start or an uncancelled service changes the result. These are example assumptions, not a typical Mannvit outcome.

CalculationAmount
First-year avoided spending: 12 x GBP 500GBP 6,000
First-year cost: GBP 2,400 + (12 x GBP 100)GBP 3,600
First-year net benefit: GBP 6,000 - GBP 3,600GBP 2,400
First-year ROI: GBP 2,400 / GBP 3,600 x 10066.7%
Monthly net benefit: GBP 500 - GBP 100GBP 400
Simple payback: GBP 2,400 / GBP 4006 months

Common questions

Questions about automation ROI

For a simple first-year estimate, subtract initial and first-year recurring costs from the first-year benefit, divide by those costs and multiply by 100. Use the same period and currency throughout. Keep cash savings separate from a valuation of time released.

Divide the initial cost by the recurring benefit after recurring costs, using matching time units. If monthly net benefit is GBP 400 and setup costs GBP 2,400, simple payback is six months. This assumes benefits start immediately and stay constant.

Include setup, data preparation, testing, subscriptions, usage fees, ongoing support and exception handling. Include internal staff time where relevant, and avoid counting the same cost or benefit twice.

A useful starting point

Check the estimate before commissioning the build.

Bring the task volume, current costs and expected exception handling. Use those to identify which assumptions need testing before making an investment decision.

Bring the page, report, or workflow as it is now.

We reply with the clearest next step, or an honest no.