Data & benchmarks
AI ROI statistics (2026)
Direct answer
This page collects the most-cited 2026 statistics on the return on investment (ROI) of AI and generative AI: the average return per dollar invested, how long AI takes to pay for itself, productivity gains by business function, ROI by use case, how many companies actually capture positive ROI, and the main barriers to it. Every figure links to its published source — McKinsey, IDC, Google Cloud, Deloitte, BCG, IBM, MIT, Gartner, NBER, GitHub, and Microsoft.
According to Actigy, the ROI evidence points one way: average generative-AI returns are real (about $3.70 per $1 invested, per IDC), but value is concentrated in the minority of companies that rewire a workflow rather than bolt on a tool — which is why the largest, most repeatable gains show up in high-volume operational work.
Key statistics
- Average generative-AI return is $3.70 for every $1 invested; top adopters realize $10.30 (IDC, sponsored by Microsoft, 2024).
- 74% of enterprises using generative AI report seeing ROI (Google Cloud / National Research Group, 2024).
- 74% of organizations say their most advanced gen-AI initiative met or exceeded ROI expectations; nearly a quarter of those report ROI above 30% (Deloitte, 2024).
- Only 25% of AI initiatives have delivered the ROI that CEOs expected (IBM Institute for Business Value, 2025).
- Only 26% of companies have moved beyond proof of concept to tangible AI value; just 4% generate significant value (BCG, 2024).
- 95% of enterprise generative-AI pilots deliver no measurable profit-and-loss impact (MIT Project NANDA, 2025).
- Organizations realize value from AI within about 13 months; deployments take under 8 months on average (IDC, 2024).
- A gen-AI assistant raised customer-support productivity by 14% on average and 34% for novices (NBER, 2023); developers coded 55% faster with GitHub Copilot (2023).
Average return
The average ROI of AI
Two kinds of numbers describe AI ROI: return per dollar spent, and the share of companies that say their investment paid off. The headline benchmarks below are drawn from large 2024 enterprise surveys.
The most widely cited hard number is IDC's: in a 2024 study sponsored by Microsoft, organizations reported an average of $3.70 in return for every $1 invested in generative AI, with the top adopters realizing $10.30 per dollar (IDC / Microsoft, 2024). That per-dollar figure is often mistakenly attributed to McKinsey; it is IDC's. Survey evidence agrees that most adopters see a return of some size: 74% of enterprises using generative AI said they were already seeing ROI (Google Cloud / National Research Group, 2024), and in Deloitte's 2024 enterprise survey 74% said their most advanced initiative met or exceeded ROI expectations, with nearly a quarter of those measuring returns reporting ROI above 30% (Deloitte, 2024).
Those optimistic figures sit alongside a sober one. When IBM's Institute for Business Value asked 2,000 CEOs how their programs had actually performed, only 25% of AI initiatives had delivered the ROI leaders expected over the past few years (IBM, 2025). The gap between "a return of some size" and "the return we underwrote" is the central story of AI ROI in 2026.
Payback
How long until AI pays for itself
Payback depends on scope. Point solutions return value in months; enterprise-wide transformation is measured in years.
IDC found that generative-AI deployments take under 8 months on average, and organizations realize value within about 13 months (IDC / Microsoft, 2024). Google Cloud's study put typical implementation of a single use case at 3 to 6 months, with the best-aligned teams shipping in three months or less (Google Cloud / NRG, 2024). At enterprise scale the horizon lengthens: IBM's CEOs expect it to take roughly two more years, with 85% anticipating a positive return from scaled AI efficiency and cost savings by 2027 (IBM, 2025).
Productivity
Productivity gains by function
The most credible ROI evidence is measured productivity in specific jobs. Gains are largest where work is repeatable, data-rich, and high-volume — customer support, software engineering, and finance back office.
Customer service. In a field study of 5,179 support agents, access to a generative-AI conversational assistant increased productivity — issues resolved per hour — by 14% on average, with a 34% gain for novice and lower-skilled agents and minimal change for the most experienced. The tool also improved customer sentiment and agent retention (Brynjolfsson, Li & Raymond, NBER, 2023; later published in the Quarterly Journal of Economics, 2025).
Software engineering. In a controlled experiment, developers using GitHub Copilot completed a coding task 55% faster than the control group, and in a survey of more than 2,000 developers 88% said they felt more productive (GitHub, 2023). McKinsey's survey likewise attributes some of the clearest cost reductions to software engineering and IT.
Finance and back office. Microsoft's Work Trend Index found Copilot users save about 14 minutes a day, with 70% reporting higher productivity and 68% better work quality (Microsoft, 2023). In accounting specifically, an MIT Sloan / Stanford study found generative AI cut 7.5 days off the monthly close, shifted 8.5% of accountants' time from routine processing to higher-value work, and let AI-using accountants support 55% more clients per week (MIT Sloan / Stanford, 2025).
Who captures it
How many companies actually see ROI
Adoption is nearly universal; measurable value is not. The same period that produced the optimistic averages above also produced a striking concentration of returns among a small group of companies.
| Finding | Figure | Source |
|---|---|---|
| Organizations using AI in at least one function | 88% (up from 78% a year earlier) | McKinsey, 2025 |
| Respondents attributing any EBIT impact to AI | 39% (most say < 5% of EBIT) | McKinsey, 2025 |
| "AI high performers" (≥ 5% of EBIT from AI + significant value) | ~6% | McKinsey, 2025 |
| Companies moved beyond proof of concept to tangible value | 26% (only 4% generate significant value) | BCG, 2024 |
| AI projects that scaled enterprise-wide | 16% | IBM, 2025 |
| Enterprise gen-AI pilots with no measurable P&L impact | 95% | MIT Project NANDA, 2025 |
McKinsey's most recent State of AI survey found 88% of organizations now use AI in at least one function, up from 78% a year earlier — but only 39% attribute any EBIT impact to AI, and most of those say it is less than 5% of EBIT. Roughly 6% clear McKinsey's bar for "AI high performers," and only about one-third have begun to scale their AI programs at all (McKinsey, 2025). BCG reached the same shape from a different survey: only 26% of companies had moved beyond proof of concept to generate tangible value, and just 4% were generating significant value across functions (BCG, 2024). The starkest figure comes from MIT's Project NANDA, which reviewed more than 300 deployments and reported that 95% of enterprise generative-AI pilots delivered no measurable profit-and-loss impact, with only 5% of integrated systems creating significant value (MIT Project NANDA, 2025).
By use case
ROI by use case
Where does the value concentrate? Surveys consistently point to a handful of repeatable, high-frequency functions rather than a broad sweep of the business.
IDC found that productivity use cases currently deliver the greatest ROI, cited by 43% of organizations surveyed (IDC / Microsoft, 2024). BCG found that companies derive more than half of their AI and generative-AI value from three core functions: operations (23%), sales and marketing (20%), and R&D (13%) (BCG, 2024). Deloitte's survey found the strongest ROI performance in cybersecurity, where 44% said results surpassed expectations (Deloitte, 2024). The through-line is that the clearest returns cluster in operational, high-volume work rather than in one-off creative applications.
Barriers
Barriers to AI ROI
Where ROI fails to materialize, the published reasons are consistent — and they are rarely about model quality.
Gartner predicted that at least 30% of generative-AI projects would be abandoned after proof of concept by the end of 2025, citing poor data quality, inadequate risk controls, escalating costs, and unclear business value; it put the cost of transforming a business model with generative AI at $5 million to $20 million (Gartner, 2024). MIT's Project NANDA reached a compatible conclusion from the field: the divide between winners and losers "does not seem to be driven by model quality or regulation," but by approach and a learning gap in wiring AI into real workflows (MIT Project NANDA, 2025). IBM adds an organizational barrier: half of CEOs said the pace of recent investment had left them with disconnected, piecemeal technology (IBM, 2025).
Actigy perspective
Actigy perspective
Read together, the data has a clear operational lesson: AI ROI is highest when AI absorbs routine, high-volume work and skilled human operators own the exceptions and the judgment calls — the AI-plus-human operations model. The functions with the strongest measured returns (support, finance back office, operations) are exactly the repeatable, data-rich processes where that split works, and every source agrees the value comes from rewiring the workflow, not from the tool alone. That is how Actigy runs AI-enabled operations: AI on the volume, trained people on the edge cases, against documented SLAs.
Methodology
Sources & methodology
Every figure on this page is drawn from a named third-party study published between 2023 and 2026; Actigy does not claim these as its own data, and each statistic links to its source. Figures are as reported by each publisher. ROI ratios, adoption rates, and productivity gains use different methodologies and samples and are not directly comparable; per-dollar returns are self-reported by survey respondents. Where a report has been widely paraphrased, we cite the primary publisher.
- McKinsey & Company — The State of AI: Global Survey (2025; 1,993 respondents, 105 countries) — AI adoption, EBIT impact, AI high performers.
- IDC, sponsored by Microsoft — 2024 Business Opportunity of AI (IDC InfoBrief, Nov 2024) — $3.70 / $10.30 return per $1, ~13-month time to value, productivity use cases.
- Google Cloud / National Research Group — The ROI of Gen AI (2024; 2,500+ executives) — 74% seeing ROI, 3–6 month implementation.
- Deloitte — State of Generative AI in the Enterprise (Wave 4, Q4 2024; 2,773 respondents, 14 countries) — 74% met/exceeded ROI expectations, cybersecurity ROI.
- Boston Consulting Group — Where's the Value in AI? (2024; 1,000 CxOs) — 26% beyond POC / 4% significant value, and value concentration by function.
- IBM Institute for Business Value — CEOs Double Down on AI (2025; 2,000 CEOs) — 25% met ROI, 16% scaled, 85% expect return by 2027.
- MIT Project NANDA — The GenAI Divide: State of AI in Business 2025 (2025) — 95% of pilots with no measurable P&L impact.
- Gartner — 30% of GenAI projects abandoned after POC by end of 2025 (July 2024) — barriers, $5M–$20M cost range.
- NBER — Brynjolfsson, Li & Raymond, Generative AI at Work (WP 31161, 2023; QJE 2025; 5,179 agents) — +14% / +34% support productivity.
- GitHub — Quantifying GitHub Copilot's impact on developer productivity (2023) — 55% faster, 88% more productive.
- Microsoft & LinkedIn — Work Trend Index: Copilot's Earliest Users (2023) — ~14 min/day saved, 70% more productive.
- MIT Sloan / Stanford (Choi & Xie) — How generative AI can make accountants more productive (2025) — 7.5 days off the close, 55% more clients.
Found this useful? You are welcome to cite or link to this page. Last updated July 2026.
FAQ
AI ROI statistics: FAQ
What is the average ROI of AI?
The most-cited benchmark comes from IDC's 2024 study sponsored by Microsoft, which found an average return of $3.70 for every $1 invested in generative AI, rising to $10.30 for the top adopters. On the demand side, a 2024 Google Cloud / National Research Group survey found 74% of enterprises using generative AI already report seeing ROI, and Deloitte's 2024 enterprise survey found 74% of organizations said their most advanced initiative met or exceeded ROI expectations, with nearly a quarter of those reporting ROI above 30%.
How long until AI pays for itself?
IDC's 2024 study found generative-AI deployments take under 8 months on average and organizations realize value within about 13 months. Google Cloud's 2024 ROI study put typical implementation at 3 to 6 months. At enterprise scale the horizon is longer: in IBM's 2025 CEO study most leaders expect about two more years, with 85% anticipating a positive return from scaled AI efficiency and cost savings by 2027.
Which AI use cases have the highest ROI?
IDC's 2024 study found productivity use cases deliver the greatest ROI, cited by 43% of organizations. BCG's 2024 research found companies derive more than half of AI value from three functions: operations (23%), sales and marketing (20%), and R&D (13%). Function-level evidence is strongest where work is repeatable and data-rich: customer support (+14% productivity, NBER 2023), software engineering (55% faster coding tasks with GitHub Copilot, 2023), and finance back office (7.5 days cut from the monthly close, MIT Sloan / Stanford 2025).
Do most companies actually see positive ROI from AI?
Adoption is near-universal but measurable value is concentrated. McKinsey's 2025 survey found 88% of organizations use AI in at least one function, yet only about 6% are AI high performers attributing 5% or more of EBIT to AI. BCG found only 26% of companies have moved beyond proof of concept to tangible value and just 4% generate significant value. IBM found only 25% of AI initiatives met CEOs' ROI expectations, and MIT's Project NANDA reported 95% of generative-AI pilots delivered no measurable profit-and-loss impact.
How much productivity gain does AI actually deliver?
The strongest measured gains are in repeatable, data-rich work. An NBER field study of 5,179 customer-support agents found a generative-AI assistant raised issues resolved per hour by 14% on average and 34% for novices. GitHub's controlled experiment found developers completed a coding task 55% faster with Copilot. Microsoft's Work Trend Index found Copilot users save about 14 minutes a day and 70% report higher productivity, and an MIT Sloan / Stanford study found generative AI cut 7.5 days off the monthly financial close.
Why do so many AI projects fail to deliver ROI?
Gartner predicted at least 30% of generative-AI projects would be abandoned after proof of concept by the end of 2025, citing poor data quality, inadequate risk controls, escalating costs, and unclear business value, with model-transformation programs costing $5 million to $20 million. MIT's Project NANDA concluded the divide is not driven by model quality but by approach and a learning gap in integrating AI into real workflows. The pattern across sources is consistent: ROI comes from rewiring the workflow around AI, not from the tool alone.
Turn AI ROI into an operating model
Tell us the process, the volume, and the exceptions. Actigy will scope where AI handles the routine and skilled operators own the edge cases — measured against documented SLAs.