Why most readiness assessments fail.

The standard format is a 1-5 maturity score across a handful of dimensions, presented back to leadership with a heat map. Leaders nod. Nothing changes. The score is too abstract to argue with and too generic to act on.

A useful assessment does three things a scorecard does not. It names specific decisions the organization is dodging. It assigns those decisions to specific people. And it ends in a 90-day plan small enough that the next review can tell whether it happened. The pattern is consistent with MIT Sloan Management Review's multi-year research on AI in the enterprise, which finds that organizations capturing AI value tend to act on cultural and process change, not just technical readiness.

The six dimensions worth measuring.

Skip any framework that gives you fewer than four or more than seven. Below four, the assessment misses the human side. Above seven, the team stops reading. For comparison, Cisco's AI Readiness Index uses a six-pillar structure and consistently finds that fewer than one in seven organizations rate as fully ready - usually because the people and process pillars trail the technology one.

01 - Dimension

Business strategy.

The question: Is there a clear business strategy that AI is meant to serve?
What ready looks like: Leadership can name the top three business outcomes AI should move in the next year, in plain language, without mentioning a tool.
Common failure mode: AI initiatives are framed as 'we should be doing more AI' rather than 'AI should help us win X customer in Y way'.

02 - Dimension

AI strategy.

The question: Is there a deliberate AI strategy, or just a list of pilots?
What ready looks like: There is a written point of view on which AI bets the organization is making, which it is not, and why. Build-vs-buy choices are explicit.
Common failure mode: Every team is running its own pilot. No one can describe the portfolio in one sentence. Vendor decks are filling the strategy gap.

03 - Dimension

Customer strategy.

The question: Is AI tied to customer value, or only to internal productivity?
What ready looks like: At least one named AI initiative is connected to a measurable customer outcome (retention, conversion, satisfaction, time-to-value).
Common failure mode: All AI value is framed as internal efficiency. No one can describe what changes for the customer.

04 - Dimension

People.

The question: Do the people who need to use AI actually know how, want to, and trust it?
What ready looks like: Role-specific AI fluency is measured. Leaders are early users themselves. Skeptics are heard, not silenced.
Common failure mode: AI training is one-off and generic. Leadership talks about AI without using it. Concerns get coded as resistance.

05 - Dimension

Process.

The question: Have workflows been redesigned for AI, or just bolted on?
What ready looks like: At least one core workflow has been rebuilt with AI in the loop - not just a faster version of the old process.
Common failure mode: AI is layered on top of unchanged processes. Time saved disappears into more output of the same kind, not better outcomes.

06 - Dimension

Platform.

The question: Is the platform stack coherent, governed, and observable?
What ready looks like: There is one named owner for the AI stack. Tools are inventoried. Spend, usage, and outcomes are visible in one place.
Common failure mode: Shadow tools outnumber sanctioned ones. No one can produce a current AI tool list in under a day. Spend is unknown.

How to run one in four weeks.

Week one - frame. Agree the business outcomes AI should move and the named owners on the leadership team. No data collection yet.
Week two - listen. 12-20 conversations across leaders, practitioners, and skeptics. A short workforce pulse is optional but useful.
Week three - read. Score the six dimensions, write the diagnosis in prose, and identify the three decisions the organization is avoiding.
Week four - sequence. Walk leadership through the read, get owners for each gap, and lock a 90-day plan with no more than five named moves.

Free vs paid assessments.

Vendor-led free assessments are usually qualification calls in disguise. Useful as a forcing function for an internal conversation; rarely useful as a real diagnosis.

Paid assessments run from a few thousand dollars (a structured workshop and short read) to six figures (a full enterprise diagnostic from a global SI). For most mid-market organizations, the right spend is in the low five figures - enough to get external eyes and a written report, small enough that you spend the rest on the work the report names. BCG's research on AI value reaches a similar conclusion: only a small minority of companies are capturing real value from AI, and the gap is rarely about the size of the diagnostic.

The free AI Alignment Snapshot is the lightest version of this - a self-serve read across the same six dimensions, useful before you commission anything paid. Teams that want the longer argument can read our book, The Elephant in the Algorithm, which sits behind the assessment method.

Questions people ask.

What is an AI readiness assessment?

A structured read of whether your organization can adopt AI without breaking - across business strategy, AI strategy, customer value, people, process, and platform. The output is a written diagnosis of where you are ready, where you are not, and what to do about the gaps. Good assessments end in a sequenced plan; weak ones end in a maturity-model score and nothing actionable.

How is it different from an AI maturity model?

A maturity model rates you on a scale (level 1 to 5, ad-hoc to optimized). An assessment tells you what to do next. They are complementary, but a maturity score on its own rarely changes behavior - leaders see the number, agree it is roughly right, and carry on. An assessment forces specific decisions about people, process, and platform.

Who in the organization should commission it?

Whoever owns the outcome AI is supposed to move - usually a CMO, COO, Chief Customer Officer, or Chief Strategy Officer. CIO/CTO-led assessments tend to over-index on platforms and under-index on the business strategy that AI is supposed to serve. The best assessments are joint, with a clear business owner.

How long should an AI readiness assessment take?

Two to six weeks for a mid-market organization. Anything shorter is a checklist; anything longer is a consulting project pretending to be an assessment. The point is to get to a sequenced plan fast enough that the plan is still relevant when you finish.

What should the deliverable look like?

A written read (10-30 pages), a one-page diagnostic summary, a prioritized list of gaps with named owners, and a 90-day starting plan. Slides alone are not enough - the writing is what forces clarity. A scorecard with no narrative is not an assessment.

AI readiness assessment.