AI Automation ROI: How to Calculate Whether Automation Is Worth It

To calculate AI automation ROI, estimate the hours the automation saves each month, multiply by the loaded cost of that time, add any measurable revenue effect, subtract monthly running costs, and divide the build cost by the net monthly benefit to get a payback period. If payback is well under a year and the assumptions are conservative, the automation is usually worth it.

The formula

Monthly benefit = (hours saved × loaded hourly cost) + measurable revenue effect − monthly running costs

Payback period (months) = build cost ÷ monthly benefit

First-year ROI = (monthly benefit × 12 − build cost) ÷ build cost

Step 1: Measure time honestly

Time the task as it's done today for a week or two. Record how often it happens and how long each instance takes, including switching between tools. Estimates from memory usually overstate or understate — measure.

Then estimate what share of that time the automation removes. Rarely 100%: someone still reviews exceptions. For a first estimate, assume a portion remains manual.

Step 2: Value the time

Use the loaded hourly cost — wages plus benefits, payroll taxes and overhead — not just the base wage. If the owner does the task, use what their time is worth on revenue-generating work.

Step 3: Add revenue effects carefully

Some automations affect revenue: faster lead response, fewer no-shows, fewer abandoned quotes. Only include these if you can measure the before and after. If you can't, leave them out and treat any gain as upside.

Step 4: Count all the costs

A worked example (illustrative numbers)

A company spends about 25 hours a month processing inbound inquiries. An automation handles classification, CRM entry and first replies; about 20 of those hours are removed. Loaded cost is $40/hour.

• Time value: 20 × $40 = $800/month
• Running costs: say $60/month
• Net monthly benefit: $740
• Build cost: $1,600
• Payback: $1,600 ÷ $740 ≈ 2.2 months
• First-year ROI: ($740 × 12 − $1,600) ÷ $1,600 ≈ 4.5×

These numbers are an example of the method, not a prediction. Use your own measurements.

When the numbers say no

• The task is infrequent.
• Exceptions are so common that most instances still need a person.
• Running costs rival the time saved (very long documents, high volumes on expensive models).
• The process is about to change anyway.

Beyond the spreadsheet

Some benefits are real but hard to price: faster response to customers, fewer errors, less burnout, the ability to grow without hiring immediately. Mention them in the decision, but don't let them carry a case the numbers don't support.

Track it after launch

Keep measuring time spent, exception rate and running costs for the first few months. Compare with your estimate and adjust.

What drives build cost is covered in how much AI automation costs; the estimator gives a range for your project. If you're choosing which process to automate first, see 10 business processes you can automate with AI.