Cost & Hire · 5 Nov 2026 · 7 min read
How to Calculate Automation ROI (With a Free Calculator)
Calculate the ROI of automating a business process: formula, worked invoice example, payback period, and the mistakes that inflate savings. Includes calculator inputs.

Short answer: automation ROI is annual savings minus annual cost, divided by annual cost, times 100. Savings are hours saved multiplied by loaded hourly cost, plus the cost of errors avoided, plus any revenue unlocked. Cost includes the build, the platform, third party usage and ongoing maintenance. Always work out year one and year two separately. Year one carries the build, and year two shows the steady state you actually live in.
Most guides skip the uncomfortable parts: run costs, exceptions that still need a human, and the awkward fact that “hours saved” only counts if somebody puts those hours to use. This one includes them.
The formulas
Time saved per year is tasks per period times minutes per task times people times periods per year, divided by 60. For daily work, use tasks per day times minutes per task times people times about 260 working days, divided by 60.
Hours per year = tasks x minutes per task x people x periods per year / 60
Savings = (Hours saved x loaded hourly cost)
+ (Errors avoided x cost per error)
+ (Revenue enabled by faster response or freed capacity)
Cost = Build (one time) + Platform fees + Infrastructure
+ Third party APIs and AI usage + Maintenance and monitoring
ROI % = (Savings minus Cost) / Cost x 100
Payback (months) = Build cost / (Monthly savings minus Monthly run cost)
The method, in the order we do it
Start by mapping the process as it really runs: every step, who does it, how long it takes and how often. Use a stopwatch, not a guess. Then set a baseline of volume per month, handling time, error rate and rework, because that is the “before” you will measure against later.
Price the time with loaded cost, salary plus overhead, not base salary. If you only know the salary, add a reasonable overhead percentage and say what you assumed. Then define the automated state honestly: what share of cases will the workflow handle on its own, and what stays with humans for exceptions and approvals? Don’t assume 100%.
Next, price the automation itself: build, platform, hosting, APIs and maintenance. For platform costs see n8n pricing in 2026, and for build costs see what an n8n agency charges. Work out year one, year two and payback. Finally, pick the metric you’ll track and make the workflow report it by itself; how to tell if your automation is working shows how.
Image: Pixabay (free to use)
Worked example: invoice processing
Every number here is illustrative. Swap in your own.
Imagine 400 supplier invoices a month. Manual entry takes six minutes each, so 2,400 minutes, or 40 hours. About 3% need rework, 12 invoices at 20 minutes each, another 4 hours. That is 44 hours a month of manual effort. At a loaded cost of $25 an hour, the manual process costs $1,100 a month, or $13,200 a year.
Now automate it. The workflow extracts data from emailed PDFs, validates it and posts to the accounting system, but 10% of invoices still go to a person for review. That is 40 invoices at six minutes, 4 hours, or $100 a month ($1,200 a year). Annual savings: $13,200 minus $1,200, so $12,000.
| Cost item | Amount |
|---|---|
| Build (one time) | $6,000 |
| n8n Cloud (Starter, billed annually) | $240 a year |
| OCR and AI usage | $60 a month, $720 a year |
| Maintenance and monitoring | $100 a month, $1,200 a year |
| Year one cost | $8,160 |
| Run cost per year (year two on) | $2,160 |
| Result | Value |
|---|---|
| Year one net benefit | $12,000 minus $8,160 = $3,840 |
| Year one ROI | 47% |
| Monthly net benefit (savings minus run cost) | $1,000 minus $180 = $820 |
| Payback on the build | $6,000 / $820, about 7.3 months |
| Year two net benefit | $12,000 minus $2,160 = $9,840 |
| Year two ROI | 455% |
The lesson is in the gap between 47% and 455%. Year one looks modest because it carries the build. If you only quote year one, you may reject a very good project.
Calculator inputs for a spreadsheet
[EMBED: AnnamNode ROI calculator here once built]
Until the tool is live, make these cells: tasks per month (B1, example 400), minutes per task (B2, 6), people involved (B3, 1), rework rate in percent (B4, 3), minutes per rework (B5, 20), loaded hourly cost in dollars (B6, 25), share still handled by humans after automation in percent (B7, 10), build cost (B8, 6000), platform and hosting per month (B9, 20), APIs and AI per month (B10, 60) and maintenance per month (B11, 100).
Then these formulas:
Manual hours per month = (B1*B2*B3 + B1*B4/100*B5) / 60
Residual hours per month = B1*B7/100*B2*B3 / 60
Monthly savings ($) = (Manual hours minus Residual hours) * B6
Monthly run cost ($) = B9 + B10 + B11
Monthly net benefit ($) = Monthly savings minus Monthly run cost
Payback (months) = B8 / Monthly net benefit
Year 1 ROI (%) = (12*Monthly net benefit minus B8) / (B8 + 12*Monthly run cost) * 100
With the example inputs you get about 44 manual hours, 4 residual hours, $1,000 of monthly savings, $180 of run cost and a payback of about 7.3 months. The year one ROI formula divides net benefit by total year one cost, which reproduces the 47% above.
Mistakes that inflate ROI
Counting 100% of hours as saved is the most common, because exceptions and reviews remain. Ignoring run costs is a close second; hosting, APIs and maintenance add up. An unrealistic hourly rate misleads in both directions, since base salary understates the true cost of an hour while an inflated rate overstates savings, so state the rate and the overhead you assumed. Counting “freed time” that is never redeployed makes savings theoretical. Forgetting adoption matters too: a workflow nobody trusts delivers nothing. Having no baseline means you can’t prove improvement later, and single point estimates hide risk, so show a pessimistic, an expected and an optimistic case.
If you can’t measure a metric, say so before you build. That is our rule on every project: baseline and target metric first.
Benefits beyond hours
Time is the easiest thing to put a number on, but it isn’t the only value. Fewer errors, faster lead response, a shorter cash cycle, more leads followed up, growth without extra headcount, and less dependence on one person all count. Put a number on one only when you can defend it, and list the rest as qualitative upside.
Image: Pixabay (free to use)
Which process to calculate first
Rank candidates by hours per month times error cost, divided by the effort to build. The best first automations tend to be repetitive handoffs between tools: lead capture and routing, invoice data entry, reporting, support triage. For marketing teams see what to automate first, and for invoices in depth see our invoice automation guide.
FAQ
What is a good ROI for automation?
It depends on the process and the risk. Some published guides suggest well chosen automations commonly deliver 200% to 500% annual ROI, but those are generalisations. Calculate your own and compare it with your cost of capital and alternatives.
How do I calculate the payback period?
Divide the one time build cost by monthly net benefit, which is monthly savings minus monthly run cost.
Should I include maintenance?
Yes. A workflow that isn’t maintained eventually fails. Include monitoring, fixes and monthly improvements.
How do I value time savings if headcount doesn’t change?
Value it as redeployed capacity (what the team does instead) or avoided hiring. If neither applies, the saving is soft, so say so.
Does the platform choice change ROI?
Yes, especially at volume. Task based and execution based billing diverge as workflows grow; see n8n vs Zapier: cost per 1,000 jobs.
Related reading
- How much does an n8n automation agency cost in 2026?
- n8n pricing 2026: cloud plans vs self hosting
- n8n vs Zapier: cost per 1,000 finished jobs
- How to tell if your automation is working