AI ROI Calculator

An AI return on investment calculator based on published experiments: choose a study, enter team size, hours and costs, and see the time saved, value and net return.

Updated November 16, 20267 peer-reviewed or working-paper studies of AI productivityEvery row sourcedMethodologyCite or embed

Studies in the calculator
7
2023 to 2025
Largest time saving
55.8%
GitHub Copilot RCT (2023)
Smallest result
-19.0%
Experienced developers (METR 2025): AI made tasks slower
Randomised trials
5
of 7 studies

Calculate your AI return

Pick a study, then enter your own numbers. Everything is calculated in your browser.

Time saved on that work
Hours saved per year
Value of time saved
Tool cost per year
Net per year
Return per $1 of tool cost

What the studies found

Share of task time saved, %

  1. Copilot RCT55.8
  2. Writing tasks40
  3. BCG consultants25.1
  4. Dev field trials20.7
  5. Support agents12.3
  6. US workers (all)5.4
  7. METR developers-19
Download chart (SVG)

The studies behind the numbers

Effect: what the study reports. Time saved: the figure the calculator uses.

StudyAuthorsYearDesignWhoSampleEffect reportedTime saved usedSource
Generative AI at WorkBrynjolfsson, Li, Raymond2023Staggered roll-out at one firmCustomer support agents5,179 agents+14% issues resolved per hour (+34% for novice and low-skilled agents)12.3%nber.org (opens in a new tab)
The Impact of AI on Developer Productivity: Evidence from GitHub CopilotPeng, Kalliamvakou, Cihon, Demirer2023Randomised controlled trialFreelance software developers95 developersTask completed 55.8% faster55.8%arxiv.org (opens in a new tab)
Navigating the Jagged Technological FrontierDell'Acqua et al.2023Randomised experimentBCG consultants758 consultants12.2% more tasks, 25.1% faster, over 40% higher quality25.1%hbs.edu (opens in a new tab)
Experimental evidence on the productivity effects of generative artificial intelligenceNoy, Zhang2023Randomised experimentCollege-educated professionals453 participantsTime taken fell by 40%, output quality rose by 18%40.0%science.org (opens in a new tab)
Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer ProductivityBecker, Rush, Barnes, Rein (METR)2025Randomised controlled trialExperienced open-source developers on their own repositories16 developers, 246 tasksTasks took 19% longer with AI tools-19.0%arxiv.org (opens in a new tab)
The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software DevelopersCui, Demirer, Jaffe, Musolff, Peng, Salz2024Three randomised field experimentsDevelopers at Microsoft, Accenture and a Fortune 100 firm4,867 developers+26.08% completed tasks20.7%papers.ssrn.com (opens in a new tab)
The Impact of Generative AI on Work ProductivityBick, Blandin, Deming (St. Louis Fed)2025Nationally representative surveyUS workers who use generative AIReal-Time Population SurveyUsers saved 5.4% of work hours5.4%stlouisfed.org (opens in a new tab)

Methodology

Each study's headline effect is copied from the paper or the authors' summary. We convert it into a share of task time saved: reported time savings are used as stated; productivity gains of p percent become 1 minus 1/(1+p). The calculator multiplies that share by the hours your team spends on comparable work, then by your hourly cost, and subtracts the tool cost you enter.

Limits. Experiments measure specific tasks under study conditions. Real gains depend on the task, the tool, training and whether saved time is used productively. One rigorous study found experienced developers were slower with AI, so the range includes losses. This is an estimate, not a forecast.

Data note. The calculator applies the time saving measured in one published study to your inputs. Where a study reports more output per hour rather than time saved, we convert it (time saved = 1 minus 1 / (1 + gain)); the conversion is shown for every study. Studies measured specific tasks with specific tools, so treat the result as a scenario, not a forecast. Two studies (St. Louis Fed, METR) give a share of all hours or a slowdown; we use them as stated.

All 7 sources
  1. https://www.nber.org/papers/w31161 (opens in a new tab)
  2. https://arxiv.org/abs/2302.06590 (opens in a new tab)
  3. https://www.hbs.edu/faculty/Pages/item.aspx?num=64700 (opens in a new tab)
  4. https://www.science.org/doi/10.1126/science.adh2586 (opens in a new tab)
  5. https://arxiv.org/abs/2507.09089 (opens in a new tab)
  6. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4945566 (opens in a new tab)
  7. https://www.stlouisfed.org/on-the-economy/2025/feb/impact-generative-ai-work-productivity (opens in a new tab)

Spotted an error or a missing row? Write to the newsroom via Contact. We correct and date every change.

How to cite or embed

Free to quote, chart and embed. Please credit i10X News and link to this page.

Download CSVJSON

More from the data desk

All datasets