How to Measure Whether an Automation Is Paying for Itself

An automation pays for itself when the time or money it saves each month, minus what it costs to run and maintain, is bigger than what you spent building it, and you can show that with numbers you measured rather than guessed. The simple version: work out the monthly net saving, divide the build cost by it, and that's your payback period in months. The rest of this post is how to get honest inputs.

What do I need to measure before I build anything?

Take a baseline first. Without one, you can't prove anything later. For the task you want to automate, record for two to four weeks:

1. How often it happens (per week).
2. How long it takes a person each time, including the interruptions and rework.
3. How often it goes wrong, and what a mistake costs (a missed lead, a late invoice, a chaser email).
4. Who does it, so you can put a realistic hourly cost on their time.

Use real timings, not memory. People usually underestimate repetitive admin because it's scattered across the day.

What's the formula?

Most ROI guides use the same shape, for example this walkthrough:

• Monthly saving = hours saved per month × hourly cost of the person's time
• Monthly net saving = monthly saving − monthly running cost (software, usage fees, maintenance time)
• Payback (months) = one-off build cost ÷ monthly net saving
• First-year ROI % = (12 × monthly net saving − build cost) ÷ build cost × 100

Example with made-up round numbers, purely to show the arithmetic: if a task takes 10 hours a month at 25 an hour, the saving is 250 a month. If running it costs 40 a month, net saving is 210. A build cost of 1,260 pays back in six months.

For hourly cost, use what the time costs you in salary plus overheads, not just the wage. Some guides suggest a multiple of base pay; treat any specific multiple as a rough assumption and use your own figures.

Which costs do people forget?

• Maintenance. Apps change, connections break, rules shift. Sources vary widely on how much to budget, so choose your own allowance and track the actual time you spend fixing things.
• Checking the output. If a person reviews every result, count that time. It may be small, but it isn't zero.
• Your own setup time, if you built it yourself.
• Usage fees that rise as volume grows.

Are "hours saved" always real savings?

No, and this is where most ROI claims fall apart. Ten minutes saved here and there only counts if it turns into something: more calls made, a faster reply to enquiries, a person who no longer needs overtime. If the time just dissolves into the day, you've saved comfort, not money.

So decide up front what the freed time is for. For a recruiter, it might be more outreach to prospective clients. For a shop owner, it might be answering customers faster. Then measure that outcome too.

Some automations pay in other ways: fewer errors, faster response times, enquiries that no longer get lost. Put a cost on these if you can (what is one missed lead worth?), but keep them in a separate line so you don't blur firm numbers with estimates.

When is it paying for itself, and when should I stop?

There's no universal threshold. A common rule of thumb in ROI guides is that payback inside six to twelve months is an easy yes and beyond eighteen months needs a strong reason, but those are opinions from vendors and consultants, so treat them as a starting point and apply your own tolerance.

A practical review routine:

1. After 30 days, compare real numbers with your baseline.
2. After 90 days, recalculate payback with actual running and maintenance cost.
3. Every six months after that, ask whether the task has changed, the volume has dropped, or the tool has become more trouble than it's worth.

If it isn't working, fix it, shrink it, or switch it off. An automation is a tool, not a commitment.

What are the limits of this method?

It works best for repetitive, countable tasks. It works poorly for creative work, relationship-heavy work, or anything where the value is quality rather than speed. For those, ask whether the outcome improved (replies, conversions, customer satisfaction), not just whether time was saved.

Small volumes also make the maths unforgiving. If a task happens five times a month, almost nothing will pay back quickly. In that case, pick what to automate first using a scoring method rather than chasing the numbers.

If you'd like help choosing where to start, our automation service and consulting pages explain how we work, and the DCA Recruit pricing page shows an example of a flat monthly cost you can plug straight into this formula. Contact us if you want to talk through your own numbers.