Grow Enterprise AI Maturity for Bottom-Line Impact (MIT CISR, August 2025)
TL;DR
An MIT CISR research briefing by Stephanie Woerner, Ina M. Sebastian, Peter Weill and Evgeny Káganer. It re-measures the four-stage enterprise AI maturity model that CISR introduced in December 2024, with a new and much smaller survey.
- New distribution, same scoring rule. Stage 1 13%, Stage 2 23%, Stage 3 46%, Stage 4 18% in 2025, against 28 / 34 / 31 / 7% in 2022 (figure, “Four Stages of Enterprise AI Maturity, 2022–2025”).
- Different and much smaller sample. The 2025 figures come from the “MIT CISR 2025 Real-Time Business Survey (N=152)”; the 2022 figures from the 2022 Future Ready Survey (N=721) (footnote 2; figure source note).
- Stage 4 performs best. Stage 4 enterprises are +13.9 pp on growth and +9.9 pp on profit against their industry average; Stage 1 is −26.5 pp and −15.1 pp. The measures are from 2024 (figure source note).
- The headline claim is that “the greatest financial impact is achieved in progressing from stage 2 … to stage 3”. Four challenges stand in the way: Strategy, Systems, Synchronization, Stewardship. Guardian Life and Italgas illustrate them.
- Not a page about 2026. CISR’s March 2026 follow-up, a session presentation called “Scaling Enterprise AI Maturity for Bottom-Line Impact”, is “available only to logged-in members and attendees of the event”. No public 2026 distribution exists.
Key data
| Stage | 2022 share (N=721) | 2025 share (N=152) | Growth vs industry, 2022 | Growth vs industry, 2024 | Profit vs industry, 2022 | Profit vs industry, 2024 |
|---|---|---|---|---|---|---|
| 1 Experiment and prepare | 28% | 13% | −12.6 pp | −26.5 pp | −9.6 pp | −15.1 pp |
| 2 Build pilots and capabilities | 34% | 23% | −3.5 pp | −6.8 pp | −2.2 pp | −1.4 pp |
| 3 Develop AI ways of working | 31% | 46% | +11.3 pp | +4.7 pp | +8.7 pp | +0.8 pp |
| 4 Become AI future ready | 7% | 18% | +17.1 pp | +13.9 pp | +10.4 pp | +9.9 pp |
The 2022 performance columns come from the December 2024 briefing, “Building Enterprise AI Maturity” (No. XXIV-12, Figure 1: “Measures of growth and profit are from 2022”). The 2024 columns come from this briefing. Its figure labels them “Growth (2025)” and “Profit (2025)”, but its source note says “Measures of growth and profit are from 2024”.
Method and definitions
- Scoring rule (unchanged). Respondents are grouped “using a measure of Total AI Effectiveness, the equally weighted combination of three measures: effectiveness of AI to (i) improve operations, (ii) improve customer experience, and (iii) support and develop ecosystems.” On a 0–100% scale, Stage 1 = 0–49%, Stage 2 = 50–74%, Stage 3 = 75–99%, Stage 4 = 100%. The December 2024 briefing uses the same wording and the same cut-offs.
- “Mature” here means Stage 4: a self-rated Total AI Effectiveness score at the top of the scale. It is not a separate outcome measure. Growth and profit are reported per stage as percentage points above or below industry average.
- Samples. 2022: Future Ready Survey, fielded October 2022, N=721 (XXIV-12). 2025: Real-Time Business Survey, N=152. The briefing does not say who was invited, who answered, or when the 2025 survey ran. It says only that it ran “during a period marked by the widespread adoption of generative AI and experimentation with agentic AI.”
- Interviews. Footnote 3 reports “Nineteen interviews with executives in eleven enterprises from 2024 to 2025.” The figure’s source note says “twenty executives in nine enterprises.” The two counts do not match.
- Sponsor. CISR is funded by its member organisations. The briefing is written for them (“published monthly to update the center’s member organizations”).
Scope and reliability
- Same rule, different population. The cut-offs did not change, but the sample did: from 721 to 152 respondents, from a periodic survey to a “real-time” one. The briefing presents the two distributions side by side without discussing comparability. A share of 18% in a sample of 152 has a 95% margin of roughly ±6 points.
- Self-report. Stages come from respondents rating how effective their own AI is. Financial performance is benchmarked against industry average, but the briefing does not say whether it is self-reported or taken from accounts.
- The headline claim and the figure. The text says the 2→3 step brings the greatest financial impact. On the 2025 figure’s own numbers, the steps in growth are 19.7 pp (1→2), 11.5 pp (2→3) and 9.2 pp (3→4); in profit 13.7, 2.2 and 9.1 pp. In the 2022 data the 2→3 step was the largest on both measures (14.8 and 10.9 pp). The briefing does not say how it measures “greatest financial impact”.
- Stage 3’s lead shrank. Stage 3’s advantage fell from +11.3 / +8.7 pp (2022) to +4.7 / +0.8 pp (2024). Stage 3 is now the largest group (46%), so it sits close to the sample average.
- Small print. The figure labels the 2024 measures as “(2025)”; the interview counts disagree (above).
Dynamic-capabilities reading
digital-transforming/improving-digital-maturity: the whole briefing is a staged maturity model with a measured distribution; it is a direct reading of how far enterprises have come in improving digital maturity.digital-seizing/balancing-digital-portfolios: the Stage 2→3 transition is about moving from a portfolio of pilots to scaled platforms. Guardian’s three-phase value-tracking framework (hypothesis, test and business case, scale plan) is a portfolio filter.digital-transforming/redesigning-internal-structures: the Systems and Stewardship challenges are structural. Guardian reorganised technology around products and platforms with small cross-functional teams. Italgas redefined its CHRO role as a chief people, innovation and transformation officer and created an AI director and a group AI office.
How it connects
- 2026-04-28-mit-sloan-ai-maturity: the MIT Sloan article of 7 October 2025 summarises this briefing (“according to a new research briefing”; same four authors, four challenges and cases). Its figure, however, is the ”© MIT CISR 2024” graphic with the 2022 distribution and “interviews in 2024 with 16 executives at nine enterprises”. That is how the 7% reached the wiki two months after the 18% was published.
- ai-maturity-measurement-comparison: the thread lists “an updated CISR Future Ready Survey wave” under What’s missing. This is that update, with a different survey.
- enterprise-ai-adoption: the concept page quotes CISR’s Stage 4 at 7%.
- 2026-05-02-dutt-chatterji-ai-experimentation-to-transformation: Dutt, Chatterji et al. describe the same pilot-to-scale step from the outside, as the “micro-productivity trap”. See also micro-productivity-trap.
- responsible-ai: Guardian’s two-track architecture review (a formal board and a fast-track board with risk, privacy and cyber representatives) is the Stewardship challenge in practice.
What was actually ingested
The full public web text of Research Briefing XXV-8 (cisr.mit.edu, no login), including the figure as an HTML table. The PDF download is reserved for registered users and was not used. To get the 2022 performance numbers, the December 2024 briefing XXIV-12 was also read (raw: raw/articles/2024-12-19-mit-cisr-building-enterprise-ai-maturity.md). The March 2026 session presentation was checked; only its title, date and abstract are public.
Linked entities and concepts
Entities: MIT CISR, Stephanie Woerner, Peter Weill.
Concepts and threads: enterprise-ai-adoption, ai-maturity-measurement-comparison, micro-productivity-trap, responsible-ai.
Dangling (single-source mention, deferred): Evgeny Káganer (named in research_by: on 2026-04-28-mit-sloan-ai-maturity; first author: listing here; Ina Sebastian now has an entity page), Guardian Life Insurance Company of America, Italgas, Bludigit, IESE Business School, Andrew McMahon, Steve Rullo, Mike Prestileo, Paolo Gallo.