How to calculate the ROI of business automation
Dravik teamUpdated 6 min read
Automation ROI is the annual value an automation creates minus its annual cost, divided by that cost. Value comes from hours saved, errors avoided and revenue gained through speed. Cost includes software, build time, maintenance and the review work that remains. Measure the manual baseline first, use conservative estimates, and check the numbers after launch.
Why most ROI estimates are wrong
Automation business cases tend to fail in one of two ways. Some are wildly optimistic: they assume every minute saved turns into profit, ignore maintenance, and count benefits that never appear. Others are so vague that leadership cannot judge them, with lines like "improves efficiency" and no numbers at all.
A useful ROI estimate sits in between. It uses real measurements of today's process, counts costs honestly, and separates hard savings from softer benefits. It also gets checked after launch, which makes the next business case more credible.
The basic formula
ROI equals annual net benefit divided by annual cost.
Annual net benefit is the total annual value minus total annual cost. You can also express the result as a payback period: the number of months until cumulative value covers the upfront cost.
Both views matter. ROI tells you how good the investment is. Payback tells you how long you are exposed before it pays for itself.
Step 1. Measure the manual baseline
You cannot calculate savings without knowing what the task costs today. For the process you plan to automate, collect:
- Volume: how many times the task happens per week or month.
- Time per instance: measured, not guessed. Watch a few people do it, or have them log time for a week.
- Who does it: role and fully loaded hourly cost, including salary, benefits and overhead. Your finance team can give you a standard loaded rate.
- Error rate: how often the task goes wrong.
- Cost per error: rework time, refunds, late fees, lost deals, or customer churn linked to the error.
- Delay cost: what a slow turnaround costs, such as leads going cold or discounts missed.
Step 2. Estimate the value categories
| Value category | How to estimate | Confidence |
|---|---|---|
| Time saved | Volume times minutes saved per instance times loaded hourly rate | Usually high if the baseline is measured |
| Errors avoided | Current errors per year times share prevented times cost per error | Medium |
| Faster turnaround | Depends on the process, such as more leads converted or discounts captured | Lower, estimate conservatively |
| Capacity | Hiring avoided as volume grows | Medium, only if growth is planned |
| Compliance and audit | Reduced risk, fewer audit findings | Usually qualitative |
Be careful with time saved. Hours saved only turn into money if the time goes somewhere useful: avoided hiring, overtime cut, or people moving to revenue-generating work. If the freed time just disappears, the benefit is real for morale but not for the P&L. Say which one you expect.
Step 3. Count every cost
| Cost | Typical items |
|---|---|
| Software | Automation platform, AI model usage, any new tools |
| Build | Internal staff time or outside help to design, build and test |
| Change management | Training, documentation, updated procedures |
| Maintenance | Fixing breaks when connected apps change, updating rules, monitoring |
| Residual manual work | Review queues and exceptions that still need people |
Maintenance is the line most often left out. Automations break when an app changes its API, a field gets renamed, or a login expires. Budget ongoing time for it.
Step 4. A worked example
The numbers below are hypothetical. They are made up to show the method and do not describe any real company or client. Replace every figure with your own measurements.
Example scenario: a company processes vendor invoices by hand and wants to automate capture, extraction and approval routing.
Example baseline:
- 1,000 invoices per month.
- 6 minutes of AP staff time per invoice for entry and routing.
- Loaded AP cost of 40 dollars per hour.
- 2 percent of invoices have an error that needs rework, at 30 minutes each.
- 1 duplicate payment per quarter, averaging 1,500 dollars, with half recovered.
Example time saved:
- Current entry time is 1,000 invoices times 6 minutes, which is 6,000 minutes, or 100 hours per month.
- After automation, assume 1.5 minutes per invoice remains for review and exceptions. That is 25 hours per month.
- Hours saved: 75 per month, or 900 per year.
- Value: 900 hours times 40 dollars, which is 36,000 dollars per year.
Example errors avoided:
- Rework: 20 errors per month times 0.5 hours is 10 hours, or 120 hours per year. If automation prevents half, that saves 60 hours, worth 2,400 dollars.
- Duplicate payments: 4 per year, losing 750 dollars each after recovery, is 3,000 dollars. If validation prevents all of them, that saves 3,000 dollars.
Example total annual value: 36,000 plus 2,400 plus 3,000, which is 41,400 dollars.
Example costs:
- Upfront build: 12,000 dollars.
- Software and model usage: 6,000 dollars per year.
- Maintenance: 4 hours per month at 75 dollars per hour, which is 3,600 dollars per year.
For the first year, total cost is 12,000 plus 6,000 plus 3,600, which is 21,600 dollars.
Example first-year ROI: net benefit is 41,400 minus 21,600, which is 19,800 dollars. ROI is 19,800 divided by 21,600, which is about 92 percent.
Example payback: monthly value is 3,450 dollars and monthly running cost is 800 dollars, so net monthly benefit is 2,650 dollars. Payback on the 12,000 dollar build is about 4.5 months.
Example second-year ROI: no build cost, so cost is 9,600 dollars against the same 41,400 dollars of value, and ROI is far higher.
Again, every number in this example is invented for illustration.
Step 5. Stress test the estimate
Run three versions.
- Conservative: smaller time savings, higher maintenance, slower adoption.
- Expected: your best estimates.
- Optimistic: everything goes well.
If the conservative case still pays back within a year, the project is a strong candidate. If only the optimistic case works, scope it smaller or pick a different process.
Step 6. Check after launch
Most teams skip this. Sixty to ninety days after launch, measure the same baseline numbers again: time per task, error rate, turnaround. Compare them with the estimate. Report the result, good or bad. A track record of accurate estimates makes it much easier to approve the next project.
Softer benefits worth naming
Some benefits are real but hard to put a number on. List them separately so they do not inflate the ROI.
- Staff satisfaction and retention when tedious work is removed.
- Faster, more consistent customer experience.
- Better data quality for reporting and decisions.
- Less dependence on one person who knows the process.
A template for your business case
- The process and why it matters.
- Measured baseline: volume, time, errors, cost.
- Proposed automation and what stays manual.
- Annual value, by category, with assumptions stated.
- Upfront and ongoing costs.
- ROI and payback in conservative, expected and optimistic cases.
- How and when you will measure results.
Keep it to one or two pages. Our guides on what to automate first and invoice processing automation help with choosing and scoping the process.
Getting a business case you can defend
The hardest part of an ROI calculation is getting an honest baseline and a realistic view of costs. If you want both worked out for your highest-value processes, book a business audit and we will deliver a written plan with estimated savings and the assumptions behind them.
Frequently asked questions
- What is a good ROI for an automation project?
There is no universal threshold, but many teams look for payback within a year in the conservative case. High-volume, rule-based processes often pay back faster. If a project only works under optimistic assumptions, scope it smaller.
- Should we count time saved as cost savings?
Only if the time goes somewhere measurable, such as avoided hires, reduced overtime, or work that produces revenue. Otherwise, report it as capacity gained and keep it separate from hard savings.
- How do we estimate maintenance costs?
Assume some ongoing time every month for monitoring and fixing breaks, and more for complex workflows touching many apps. Track actual maintenance time after launch and use it to improve future estimates.
- What if we do not have baseline data?
Collect it before building. Have the team log time on the task for one or two weeks and count errors over a recent period. Even a short measurement is far better than a guess.
- Do AI-based automations change the ROI calculation?
The method stays the same, but add model usage costs and the human review that AI outputs need. Accuracy also affects value, so measure how many AI outputs are accepted without changes.
Book a business audit
Find the leak. Then we hunt it down.
We look at your busywork, your Google rankings and how AI assistants describe your category, then hand you a prioritized plan.
- $199 fixed price
- 30 minute call
- A written plan you keep
- Full refund if not worth it