What does an AI readiness assessment actually cost in 2026?

If you are a CEO of a mid-market company in Europe and you typed "AI readiness assessment" into a search engine this week, you got three different answers.

One said it costs around €10,000 and takes eight days. That was the Bpifrance diagnostic, the French public investment bank's subsidized program for SMEs. A consultant goes in, maps the situation, delivers a report. Eight days. Ten thousand euros. The government covers most of it.

Another said €50,000 to €150,000 and takes six to ten weeks. That was a Big Four or large consultancy proposal. A team. A methodology. A deck. A roadmap that nobody will execute.

A third said something in between, from a boutique firm or a solo consultant. €15,000 to €40,000. Two to four weeks. A named person, not a brand, doing the actual work.

All three are real. All three are called the same thing. None of them are the same product.

The Bpifrance model is the closest thing to a public utility for AI diagnostics. A standardized framework, a vetted consultant, a fixed price, a fixed timeline. It is designed to get a company from zero to informed. The report tells you where you stand, what your risks are, and what your options look like. It does not tell you what to build on Monday morning.

That scope produces a diagnostic rather than an implementation plan. Eight days allows the consultant to survey, identify, and prioritize. A company examining AI systematically for the first time may find that useful.

The limitation is what happens after. The report lands on the CEO's desk. The recommendations are clear. The execution is not. Most companies stop there. The diagnostic becomes a shelf document. Something they can point to when someone asks if they are working on AI.

Consulting outputs often stop at the map. Execution still has to begin.

The Big Four model is a different animal entirely. The price reflects the brand, the team size, the methodology, the liability coverage, and the overhead of a global firm. What you get is thoroughness and a network. What you do not necessarily get is speed or a person who will hold accountability for the result.

The deliverable tends to be a comprehensive assessment with a multi-year roadmap, maturity scores, and recommendations grouped by priority. The work is real. The people doing it are competent. The gap is in what comes after the engagement ends. The team leaves. The report stays. The company is left with a plan and no builder.

There is a structural reason for this. Large consultancies are designed to diagnose and recommend. Implementation is a different business model with different margins and different risk profiles. Most of them are not set up to do both on the same engagement, at the same price, with the same team.

The industry is structured this way. The Big Four model makes sense when you need the brand, governance framework, or multi-year transformation architecture. Direct accountability for mapping and building calls for a different model.

The boutique model sits between these two poles. A smaller team, sometimes a single senior consultant, a fixed scope, a defined deliverable. The price varies widely because the quality varies widely. Some boutiques produce the same shelf documents as the large firms, just with fewer slides. Others produce something closer to an actual blueprint for what to build.

The differentiator is whether the person doing the assessment has built and deployed systems, found flaws in the plan, and fixed them.

There is a difference between a consultant who knows what AI can do and a consultant who has made it work inside a company. The first can write a report. The second can tell you where the data is actually stored, which process has the most undocumented steps, which person in the organization is the only one who understands how something works, and what will break if you change it without talking to them.

Solve IT does not force every company through the same diagnostic. A diffuse cross-functional problem may need a War Map. An already-defined workflow may go directly to a bounded Feasibility Sprint.

Scope, timing, price, evidence requirements, and risk controls are agreed for the actual case before work begins.

The useful question is not whether the assessment has an aggressive guarantee. It is whether it produces a decision the company can own: what should change, what should stay, what must be tested first, and who carries the risk.

The question you should ask before paying anyone for an AI readiness assessment is not what is the price. It is what happens on day one after the report lands.

Does the person who wrote it stay to build the first thing? Do they know where the data actually lives, not where the IT diagram says it lives? Have they deployed something similar in a company your size, with your mix of legacy systems and informal processes? Can they show you a specific example of a process they mapped, a system they designed, and a result they held themselves accountable for?

If the answer to any of these is no, you are not buying an assessment. You are buying peace of mind. There is nothing wrong with that, as long as you know what you are buying.

The Bpifrance diagnostic gives you informed awareness. The Big Four assessment gives you a comprehensive framework and board-level confidence. A boutique engagement can give you a focused perspective and, when the scope fits, a path from diagnosis to a controlled first implementation.

None of these is wrong. They are different products for different stages of readiness.

The cost of getting it wrong is not the fee. It is the six to eighteen months of inaction that follows a diagnostic that produced analysis but no next step. That is the real price. And it is the most expensive one.

If you are evaluating options, ask what happens after the assessment.

That is the thing worth knowing before you sign anything.