Organisational AI Adoption Maturity

Measure organisational AI adoption maturity across strategy, leadership, people, data, governance, and value — and agree where to invest next.

Run in TeamRetro

Assess the maturity of your organisation's AI adoption

Adopting AI well across an organisation is rarely about the technology alone — it depends on strategy, leadership, capability, data, governance, and a clear line of sight to value. The Organisational AI Adoption Maturity model gives leadership teams a structured way to assess how mature their AI adoption really is, function by function and capability by capability. Spanning seven areas — from AI strategy and executive sponsorship through to adoption breadth, people and capability, data foundations, governance, and value realisation — it surfaces where AI is genuinely embedded, where it is improvised, and where attention is overdue. Each dimension is rated against a five-stage maturity scale, from Ad Hoc through to Optimized, so teams can have an honest, shared conversation about their current state and agree where to invest next. Use it to baseline AI maturity, align leaders on ambition versus risk, and track progress as adoption deepens over time.

Dimensions

AI Strategy & Vision

Whether the organisation has a clear, shared direction for AI — why it matters, how far to push it, and which use cases to back.

  • Strategic Direction on AI

    The organisation has a clear, articulated view of why AI matters, where it fits, and what it is meant to change.

    1. Ad HocAI is talked about reactively; no shared view of why it matters or where it fits.
    2. EmergingA few leaders have views on AI; they don't add up to an organisational direction.
    3. DefinedA written AI direction exists and most leaders can articulate it consistently.
    4. ManagedThe AI direction shapes annual planning and investment; it is reviewed as the landscape shifts.
    5. OptimizedAI is part of the organisation's core strategy; every function understands its role in delivering it.

  • Ambition vs Risk Posture

    The organisation has settled how far it is willing to push AI adoption and how much risk it accepts in doing so.

    1. Ad HocAmbition and caution swing case by case; no stable posture.
    2. EmergingLeadership has views on appetite and risk but they aren't reconciled across the business.
    3. DefinedA clear stance exists on how aggressive the organisation will be with AI and what risks it accepts.
    4. ManagedThe posture is revisited as the technology and regulation move; trade-offs are made explicitly.
    5. OptimizedAmbition and risk are managed as one decision; teams act with confidence about what is and isn't in bounds.

  • Use-Case Portfolio

    The organisation knows which AI use cases it is investing in across functions and why those cases were chosen.

    1. Ad HocUse cases happen wherever someone champions them; no portfolio view.
    2. EmergingA list of AI initiatives exists; criteria for picking them are inconsistent.
    3. DefinedA portfolio of AI use cases is maintained with a shared rationale for inclusion and priority.
    4. ManagedThe portfolio is balanced across quick wins, strategic bets, and foundational work; it is reviewed regularly.
    5. OptimizedUse cases are added, paused, and retired on evidence; the portfolio is a deliberate management tool.

Leadership & Sponsorship

Whether senior leaders own AI, fund it credibly, use it themselves, and coordinate it across the business.

  • Executive Ownership

    A senior leader is accountable for AI as a whole, with the authority to make decisions across functions.

    1. Ad HocNo-one owns AI; whoever shouts loudest moves things along.
    2. EmergingOwnership sits with one function (often IT) and doesn't cover the breadth of AI work.
    3. DefinedA named senior leader is accountable for AI across the organisation.
    4. ManagedThe owner has the authority, budget, and reach to make cross-function decisions stick.
    5. OptimizedAI ownership is built into how the organisation runs; succession and continuity are planned for.

  • Investment Commitment

    The organisation invests in AI at a level matched to its stated ambition, and the investment is funded predictably.

    1. Ad HocAI investment is whatever budget can be scraped together case by case.
    2. EmergingSome funding exists; it doesn't match the ambition leaders talk about.
    3. DefinedA funded AI plan exists with a predictable budget envelope.
    4. ManagedInvestment is calibrated against outcomes; reallocation between initiatives is normal.
    5. OptimizedFunding scales up and down on evidence; AI investment behaves like any other strategic commitment.

  • Visible Leadership Use

    Leaders use AI themselves and talk about it openly; AI adoption isn't something they ask of others without doing.

    1. Ad HocLeaders ask the rest of the business to adopt AI but don't use it themselves.
    2. EmergingSome leaders use AI personally; others see it as someone else's job.
    3. DefinedMost leaders use AI in their own work and talk about it openly.
    4. ManagedLeadership use of AI is visible to teams and shapes expectations.
    5. OptimizedLeaders model AI use as a baseline professional skill; the tone from the top is consistent and credible.

  • Cross-Function Coordination

    AI work is coordinated across functions rather than each team buying tools, writing policy, and running pilots on its own.

    1. Ad HocFunctions run AI work independently; duplication, conflicting tools, and conflicting policy are common.
    2. EmergingSome coordination exists; alignment is informal and depends on relationships.
    3. DefinedA standing forum or function coordinates AI work across the business.
    4. ManagedCoordination drives concrete decisions — shared tools, shared standards, shared learning across functions.
    5. OptimizedCross-function coordination on AI is how the organisation operates; no function is left to figure it out alone.

Adoption Across Functions

How broadly and deeply AI is actually used across the business — and whether adoption is balanced across customer-facing and internal work.

  • Breadth of Function Coverage

    AI is in active use across the major functions of the business (people, finance, operations, sales, marketing, support, product) — not just one or two.

    1. Ad HocOne or two functions experiment with AI; the rest of the organisation does not.
    2. EmergingSeveral functions have started; coverage and depth vary widely.
    3. DefinedMost major functions have at least one meaningful AI use in production.
    4. ManagedAll major functions use AI in ways matched to their work; gaps are visible and addressed.
    5. OptimizedAI is part of how every function operates; the question is no longer whether but how well.

  • Depth Within Functions

    Where AI is used, it is used substantively — embedded in workflows, not bolted on at the edges.

    1. Ad HocAI use is shallow — a chatbot here, a draft tool there; little impact on daily work.
    2. EmergingSome functions use AI in real workflows; many treat it as an add-on.
    3. DefinedFunctions that use AI use it in meaningful, workflow-embedded ways.
    4. ManagedDepth is measured; functions are pushed past surface-level adoption to real change.
    5. OptimizedAI is woven into how work gets done in every adopting function; the line between AI and non-AI workflow is gone.

  • Front-Office vs Back-Office Balance

    AI adoption is balanced between customer-facing work and internal operations, rather than concentrated in one side of the business.

    1. Ad HocAI is concentrated on one side of the business; the other is largely untouched.
    2. EmergingSome balance exists; major gaps remain on one side.
    3. DefinedBoth front-office and back-office have meaningful AI adoption.
    4. ManagedThe balance is intentional and matched to where AI creates the most value for the organisation.
    5. OptimizedAI is applied wherever it produces value, regardless of where in the business that sits.

People & Capability

Whether the organisation has the AI literacy, specialist skills, shared learning, and talent pathways to do AI well and keep getting better.

  • Baseline AI Literacy

    Employees across the organisation have the baseline AI literacy expected of their role — they can use AI tools sensibly and know their limits.

    1. Ad HocAI literacy is whatever people pick up on their own; gaps are wide.
    2. EmergingSome training has been offered; uptake and quality are uneven.
    3. DefinedA baseline AI literacy expectation exists for relevant roles and is being met.
    4. ManagedLiteracy is assessed and refreshed; new joiners come up to standard quickly.
    5. OptimizedAI literacy is a baseline professional capability across the organisation; the bar is calibrated and rising.

  • Specialist Capability

    The organisation has the specialist AI capability it needs (data, ML/ops, prompt design, governance) to do AI well in the areas that warrant it.

    1. Ad HocSpecialist AI capability is whatever the team accidentally has; gaps are filled by hope.
    2. EmergingSome specialist skill exists; it doesn't cover the work the organisation has taken on.
    3. DefinedThe organisation has identified the specialist capabilities it needs and has them in place.
    4. ManagedSpecialist capacity is matched to the AI portfolio; gaps are filled deliberately.
    5. OptimizedSpecialist AI capability scales with ambition; the organisation can take on new AI work without surprises.

  • Internal Community & Knowledge Sharing

    People learning about AI find each other, share what works, and lift the organisation's collective skill faster than individuals could alone.

    1. Ad HocLearning happens in silos; the same lessons are re-learned in different corners.
    2. EmergingSome sharing happens informally; it depends on individual energy.
    3. DefinedA community of practice, forum, or channel exists where AI learning is shared across functions.
    4. ManagedSharing produces visible lift — patterns spread, mistakes don't repeat.
    5. OptimizedLearning compounds across the organisation; new AI capability is absorbed at the speed of the technology.

  • Hiring & Career Pathways

    The organisation hires for AI capability where it matters and offers career pathways that retain the people it develops.

    1. Ad HocAI capability is not part of how the organisation hires or develops people.
    2. EmergingHiring sometimes considers AI capability; career paths for AI-capable people are unclear.
    3. DefinedAI capability is part of hiring criteria and development plans for roles where it matters.
    4. ManagedCareer pathways exist for AI-capable people; retention is monitored and acted on.
    5. OptimizedThe organisation is a place AI-capable people want to be; it grows and keeps the talent it develops.

Data & Foundations

Whether the data, platforms, and vendor estate AI depends on are ready, shared, and managed at the organisation level.

  • Data Readiness

    The data AI needs to do useful work — accurate, accessible, well-governed — is in a state to support the AI use cases the organisation has chosen.

    1. Ad HocData is messy, scattered, and hard to use; AI work is held back by it.
    2. EmergingSome data is in shape for AI use; large gaps remain in quality, access, or coverage.
    3. DefinedThe data underpinning priority AI use cases is fit for purpose and reliably available.
    4. ManagedData readiness is improved deliberately as new AI use cases come into scope.
    5. OptimizedData is treated as a strategic asset for AI; readiness keeps pace with ambition.

  • Technology Foundations

    The platforms, integrations, and tooling AI relies on are in place — not improvised under each new initiative.

    1. Ad HocEach AI initiative builds its own scaffolding; nothing reuses.
    2. EmergingSome shared foundations exist; many initiatives still go it alone.
    3. DefinedA shared set of AI platforms and integrations is in place and reused.
    4. ManagedFoundations evolve to meet new needs; teams build on top, not from scratch.
    5. OptimizedAI foundations are a competitive advantage; new use cases launch faster because of them.

  • Vendor & Tool Management

    AI vendors and tools are chosen, integrated, and rationalised at the organisation level — not bought one team at a time.

    1. Ad HocEvery team buys its own AI tools; sprawl, overlap, and surprise spend are constant.
    2. EmergingSome procurement coordination exists; many buys still happen below the radar.
    3. DefinedAI tool selection is coordinated; major vendors are chosen at the organisational level.
    4. ManagedThe AI tool estate is reviewed and rationalised; spend matches use.
    5. OptimizedVendor and tool decisions on AI are made strategically; the estate is lean, well-understood, and easy to change.

Governance, Risk & Compliance

Whether AI is governed responsibly — clear policy, managed risk, regulatory compliance, and ethics applied in real decisions.

  • AI Policy & Acceptable Use

    The organisation has a clear, current, and known policy on what AI tools people can use, on what data, and for what purposes.

    1. Ad HocThere is no AI policy, or one exists but no-one knows it.
    2. EmergingA draft policy exists; coverage and awareness are patchy.
    3. DefinedA current AI acceptable-use policy is published and known across the organisation.
    4. ManagedPolicy is reviewed as tools and risks evolve; changes are communicated and absorbed.
    5. OptimizedPolicy is treated as a living document that people genuinely use to make decisions.

  • Risk Management

    AI-specific risks (hallucination, bias, data leakage, model drift, vendor dependency) are recognised, assessed, and managed alongside other organisational risks.

    1. Ad HocAI risks are not separately identified or assessed; surprises are common.
    2. EmergingSome AI risks are recognised; assessment is informal.
    3. DefinedAI risks are catalogued and assessed alongside other operational risks.
    4. ManagedRisk treatments are in place and reviewed; ownership of each major AI risk is clear.
    5. OptimizedAI risk management is mature, anticipatory, and credible to internal and external scrutiny.

  • Regulatory & Compliance Posture

    The organisation tracks the regulations that apply to its AI use, complies with them, and adapts as the rules change.

    1. Ad HocRegulatory exposure on AI is unexamined; compliance is hoped for, not managed.
    2. EmergingSome regulations are known and addressed; coverage is uneven across markets.
    3. DefinedApplicable AI regulations are documented and the organisation operates within them.
    4. ManagedCompliance is monitored as rules evolve; updates are timely and communicated.
    5. OptimizedRegulatory posture is defensible, transparent, and ready for external scrutiny.

  • Ethics & Responsible Use

    The organisation has a working position on AI ethics — fairness, transparency, human oversight, impact on people — and applies it in actual decisions.

    1. Ad HocEthical considerations on AI are raised reactively, if at all.
    2. EmergingSome principles are written down; they don't reliably influence decisions.
    3. DefinedA working position on AI ethics exists and shapes decisions on use cases and tooling.
    4. ManagedEthical review is part of how AI initiatives are scoped, approved, and operated.
    5. OptimizedResponsible-use thinking is built into AI work end-to-end; the organisation can defend its choices to staff, customers, and regulators.

Value Realisation

Whether the organisation measures what AI actually delivers, understands its cost and ROI, learns from failure, and keeps improving its AI practice.

  • Outcome Measurement

    The organisation measures what AI is actually delivering — productivity, quality, revenue, cost, customer experience — not just adoption metrics.

    1. Ad HocAI value is anecdotal; vanity metrics (users, prompts) stand in for outcomes.
    2. EmergingSome outcomes are tracked; coverage is partial and inconsistent.
    3. DefinedEach major AI initiative has outcome metrics matched to what it is meant to change.
    4. ManagedOutcomes are reviewed against expectations; underperforming work is adjusted or stopped.
    5. OptimizedOutcome measurement is honest, comparable across initiatives, and trusted as a basis for decisions.

  • Cost & ROI Visibility

    The organisation knows what it is spending on AI in total and what it is getting back, well enough to make sensible scale-up and scale-down decisions.

    1. Ad HocTotal AI spend is unknown; ROI is a hope.
    2. EmergingSpend is partially visible; ROI is mostly anecdotal.
    3. DefinedAI spend is tracked at an organisational level and tied to the outcomes it produces.
    4. ManagedSpend and ROI drive portfolio decisions; expensive work is scrutinised; cheap wins are scaled.
    5. OptimizedAI economics are part of how the organisation plans; investment decisions are evidence-based at every level.

  • Failure Capture & Learning

    When AI initiatives underperform or fail, the organisation captures what happened and uses it to improve later work.

    1. Ad HocFailed AI work is buried; lessons aren't captured.
    2. EmergingSome post-mortems happen; they don't change how the next initiative runs.
    3. DefinedA standing process captures and shares lessons from AI work across the organisation.
    4. ManagedLessons drive observable improvements over time; the organisation can name what it has learned.
    5. OptimizedLearning compounds across initiatives; the organisation is visibly better at AI work each year.

  • Continuous Improvement of AI Practice

    The way the organisation does AI work — strategy, governance, capability, delivery — is itself reviewed and improved as the field changes.

    1. Ad HocHow the organisation does AI is whatever it was when it started; nothing about the approach changes.
    2. EmergingOccasional adjustments happen, usually reactively after something goes wrong.
    3. DefinedThe AI operating model is reviewed periodically and updated when warranted.
    4. ManagedImprovement of AI practice is itself a programme — measure, adjust, measure again.
    5. OptimizedContinuous improvement of how the organisation does AI is built into the operating model; the approach evolves as fast as the technology.

When to use this health check

  • Baselining how mature your organisation's AI adoption is before setting an AI strategy or roadmap
  • Aligning the leadership team on AI ambition versus risk appetite
  • Identifying where AI adoption is shallow, siloed, or ahead of governance
  • Tracking AI maturity progress over successive quarters or years
  • Preparing a board or executive update on enterprise AI readiness

Tips & tricks

  • Run it with a cross-functional leadership group so adoption across every function is represented, not just IT or one champion team.
  • Treat the Ad Hoc to Optimized scale as a shared language — discuss why ratings differ before averaging them.
  • Pair low scores in Governance, Risk & Compliance with high scores in Adoption to surface where the organisation has outrun its guardrails.
  • Re-run the assessment on a regular cadence to make maturity progress visible and hold investment decisions to account.
  • Focus the debrief on the two or three dimensions with the widest spread of scores — that is where alignment is weakest.

Frequently asked questions

What is an AI adoption maturity model?
It is a structured framework that rates how mature an organisation's AI adoption is across several dimensions — such as strategy, leadership, capability, data, governance, and value. Each dimension is scored on a staged scale, here from Ad Hoc through Emerging, Defined, and Managed to Optimized, giving a clear picture of current state and where to improve.
Who should take part in this assessment?
A cross-functional group of leaders works best — executives, function heads, and those accountable for AI strategy, data, governance, and delivery. Broad participation ensures adoption across the whole business is reflected rather than the view of a single team.
How is each dimension rated?
Every dimension is rated on a five-stage maturity scale from Ad Hoc to Optimized, with a written descriptor for each level so participants share a common understanding of what each rating means.
How often should we run it?
Most organisations baseline once and then re-run every six to twelve months. AI capability and the surrounding technology move quickly, so a regular cadence keeps the picture current and makes progress measurable.
What should we do with the results?
Use the spread of scores to prioritise. Dimensions where ratings are low or where participants disagree most are usually the best places to focus next — for example closing a governance gap that has fallen behind rapid adoption, or investing in capability where ambition outstrips skill.