How to Calculate the ROI of Operations Automation

Every automation project has a business case. Most of them are bad. They either count time savings that never materialize, or they ignore the maintenance cost of the automation itself, or they bundle "morale" as a number to make the spreadsheet work.

Here is the framework we use with startups to evaluate whether an automation project is worth doing, with the actual math that holds up after a year.

The four-input model

For any candidate automation, you need four numbers.

Frequency. How often does the task run today? Per day, per week, per month. Be specific.

Time per run. How long does it take a human, end to end, including context-switching cost? Most teams underestimate this. A "five-minute task" that requires opening three tools and answering a question in Slack is closer to 15 minutes of real-world cost.

Build cost. Engineering hours to build the automation, plus integration costs, plus the cost of any new tooling.

Maintenance cost. Engineering hours per quarter to maintain it. Use 15% of build cost as a default. Higher if the automation depends on third-party APIs that change often.

The math

Annual time saved = Frequency per year × Time per run × Adoption rate

Adoption rate matters. If you automate a process but half the team keeps doing it manually because the automation is confusing, you only get half the savings. Realistic adoption rates start at 70%.

Annual cost = Build cost (amortized over 2 years) + Annual maintenance + Tool subscription

If annual time saved (in hours) × loaded hourly cost > annual cost, the automation pays for itself. If the payback period is under 12 months, build it now. If it is 12 to 24 months, build it if you also expect to scale into more usage. If it is over 24 months, skip it.

A concrete example

You want to automate onboarding new employees. Today it takes 4 hours per hire across IT, HR, and the hiring manager. You hire 30 people per year.

• Annual time today: 30 × 4 = 120 hours
• Loaded cost: 120 × $100 = $12,000
• Build cost: 80 engineering hours × $150 = $12,000
• Maintenance: $1,800 per year
• Tool cost (workflow platform): $2,400 per year
• Adoption: 80%, so realistic savings = $9,600 per year

Year-one math: $9,600 saved minus ($6,000 amortized build + $1,800 maintenance + $2,400 tools) = negative $600. Marginal. But year-two and beyond is solidly positive.

This is a borderline build. If you expect to hire 60 people next year instead of 30, it is clearly worth it. If hiring stays flat, you might be better off batching onboarding instead.

The numbers that get missed

Two soft savings are real and worth counting carefully.

Error reduction matters when manual processes have a meaningful failure rate. Onboarding mistakes that delay an employee's start by a week have a real cost. Quote calculations done by hand that go out wrong cost deals.

Cycle time matters when the manual process is on a critical path. If a customer is waiting 48 hours for something that could be automated to under an hour, that latency has a revenue cost.

Both are real. Both are also where most ROI calculations cheat. Only count them if you can defend the number in a board meeting.