AI ROI calculator
See whether an AI project pays for itself. Enter the task, the time saved, and the costs, and get payback, first-year return, and break-even volume.
Runs in your browser. Nothing you enter is sent anywhere.
Include the time a person spends reviewing and fixing the output.
Salary plus benefits and overhead, divided by working hours.
Model usage and any subscriptions. The cost calculator can estimate this.
Build time, integration, training, and rollout.
This pays back in under 1 month and returns $194,120 in the first year.
- Hours saved per month
- 196.7
- Net savings per month
- $16,677
- After $40 in AI costs
- Payback period
- Under 1 month
- Setup cost recovered
- First-year return
- $194,120
- 2,996% on total cost
- Break-even volume
- Under 1 task a month
- Below this volume, the monthly AI cost is more than the time saved is worth.
A worked example: support ticket triage
Say a support team handles 4,000 tickets a month. Today, sorting and routing each one takes 8 minutes. With an AI assistant doing the first pass, a person still spends 3 minutes checking and adjusting. Time is worth $40 an hour, the AI costs $600 a month to run, and building and rolling it out costs $8,000 once. These numbers are examples, not real client data.
| Result | 4,000 tickets a month | 100 tickets a month |
|---|---|---|
| Hours saved per month | 333.3 | 8.3 |
| Net savings per month | $12,733 | -$267 |
| Payback period | Under 1 month | Never |
| First-year return | $144,800 | -$11,200 |
| Break-even volume | 180 tasks a month | 180 tasks a month |
Each ticket saves the same 5 minutes in both columns, but the outcomes are opposites. At 4,000 tickets a month the project saves 333 hours, nets $12,733 a month after AI costs, and pays back in under 1 month. At 100 tickets it saves only about $333 of time, which is less than the $600 it costs to run, so it never pays back.
What the example teaches
- Volume decides. The break-even volume, 180 tasks a month, is identical in both columns. It depends only on what each task saves and what the AI costs to run, so it’s the first number to check for any small-savings, high-volume idea.
- Count the review. The 3 minutes with AI include a person checking the result. Use the time someone spends from start to finish, not the time the model takes.
- Ask where the hours go. 333 saved hours only count as savings if the team spends them on something worth paying for, such as faster replies or a bigger workload.
How the estimate works
- How do you calculate ROI for an AI project?
- Multiply the hours saved each month by the hourly cost of the people doing the work, then subtract the monthly AI running cost. Compare that net savings figure to the one-time setup cost to find the payback period, and to the total first-year cost to find ROI.
- What should I count as time with AI?
- Count everything a person still does: prompting, reviewing, correcting, and handing off the result. A tool that drafts in seconds but needs 20 minutes of checking has saved less than it appears to.
- What hourly cost should I use?
- Use the fully loaded cost: salary plus benefits and overhead, divided by productive hours. It is usually well above the hourly wage.
- Does saved time count as real savings?
- Only if the freed time goes somewhere useful. If people spend it on higher-value work or the team handles more volume, it counts. If it just becomes idle time, the saving is on paper only.
- What does this calculator leave out?
- Quality gains, error reduction, faster turnaround, revenue effects, ramp-up time while people learn the tool, and ongoing maintenance. Those can matter more than the time savings, so treat the result as one input to a decision.
I’m Courtney Stephenson, an AI strategy and implementation leader, and I built this tool. If you’re deciding where AI belongs in your organization, I’d like to hear about it.
Want a second opinion on your results?
Book a free 15-minute call and we’ll talk through what this means for your team.