AI Maturity Model · Adoption Stages
AI Maturity Model
What is an AI maturity model?
- Stages from individual use to work redesign
- Each stage has a different bottleneck
- Assess each department separately to be accurate
The moment you bring a maturity model into a meeting, the unit of conversation changes. The for-or-against debate of 'we must adopt AI' versus 'it's too early' becomes a conversation about ownership: 'Sales is at Stage 2, so this year's goal is reaching Stage 3.' The budget changes too. A single line item for license costs splits into account rollout, foundational training, work-standard revisions, and security review, and each item comes with an explanation of which stage's bottleneck the money is meant to clear. The practical value of a maturity model is that it shows where to spend money and training hours first.
What separates the stages is not the number of tools but whether the method survives a change of owner. The line between tool rollout and real work use is set not by account coverage but by whether the AI procedure is written into that task’s standard document or checklist. In practice, scoring on just three measures—account coverage, share of weekly active users (many organizations use 30% as an internal baseline), and whether someone is named as the accountable reviewer of AI output—is usually accurate enough. Early-stage bottlenecks are access and basic skills; later-stage bottlenecks are policy, data, and review accountability.
The assessment is meaningful at the department level, not as a company-wide average. It is common for Marketing to be at Stage 3 while Production is at Stage 1, and a plan built on a single average of Stage 2 produces training that fits neither side. Also, this model is different from DX maturity or data maturity. Even with a weak data foundation, Stage 3 is possible in document work; conversely, even with well-accumulated data, it is Stage 1 if nobody uses it.
The most common failure is when the assessment sheet turns into a department scorecard. Once a team is labeled Stage 1, it reads as a ‘lagging department,’ so responses get inflated in the next survey and the assessment loses its value. State up front, in the very first explanation, that a stage is not a report card but an assignment sheet for the next task. Setting stages from executive self-assessment alone is also risky. A gap is common where leadership sees Stage 3 but a practitioner survey says Stage 1, and a plan built without knowing that gap ends up buying yet another tool nobody on the floor opens.
John KimInstructor John Kim, author of B2B Marketing Design and CEO of Nanum Consulting
Seolyeong ParkGenerative AI and AX training specialist with 17 years of hands-on experience and 200+ lectures, helping non-technical staff put AI to work right awayIt applies the structure of maturity models long used in IT and data analytics (the CMM family) to AI adoption. The number of stages and their names vary slightly among consulting firms and research institutes, but the backbone is the same: moving from individual use to redesigned work.
Consider a hypothetical case. Company A, a Korean automotive parts maker, has 380 employees and annual revenue of about KRW 90 billion (≈ US$67,335,000). Last year it ran two rounds of prompt training for all staff, but six months later only about ten people were using AI. The training team decided to stop asking ‘why aren’t they using it?’ and start by asking ‘where do we stand today?’
- When the training planning lead met 24 practitioners across 8 departments, the marketing team was already drafting product pages with AI, while a production management foreman said, ‘I don’t even know what an account is.’
- Scoring on just three measures—account coverage, share of weekly active users, and whether AI use is written into work standard documents—placed Marketing at Stage 3, Management Support at Stage 2, and Production & Quality at Stage 1.
- The head of production said, 'We do use it,' but on checking, the only use was a single team lead’s personal account, and on the spot they accepted the Stage 1 rating as it stood.
- Teams at Stage 1 get a 90-minute hands-on course, while Stage 3 Marketing gets a workshop on review criteria and updating its standard documents, replacing the old one-course-for-everyone approach.
- After six months, the share of users active at least once a week rose from 18% to 47%, drafting a marketing product page dropped from 4 hours to 1.5, and the production division also reached Stage 2.