How to run an AI readiness assessment for a client (and what to charge)

Somewhere in your last few client conversations the question showed up: what should we be doing about AI? An AI readiness assessment is the working answer, a short fixed-fee engagement that replaces a shrug or a hot take with a scored baseline and a plan. The client's real problem is rarely picking a tool. It's knowing whether their workflows, data, skills, and rules can carry one.

The gap between buying AI and getting value from it is now the best-documented finding in the field. MIT's NANDA initiative put it bluntly in its 2025 State of AI in Business report: about 95 percent of enterprise generative AI pilots deliver no measurable P&L impact, and the cause is a learning gap between tools and the organizations using them, not model quality. Cisco's 2025 AI Readiness Index, drawn from 8,000 senior leaders across 30 markets, found only 13 percent of organizations fully ready. Most clients aren't in the 13 percent, and most of them suspect it.

Plenty of readiness work deserves its make-work reputation, and the skeptics have a point. Analyst Charlene Li argues that the standard AI readiness assessment traps clients, because an exercise designed to surface gaps always finds them, and the report becomes a reason to wait rather than a decision. She's right about the bad version. The fix is to end every assessment with a named first project, a ranked workflow list, and a date, so the score serves a decision instead of replacing one.

For consultants and fractional execs, this is also the cheapest positioning move on the menu. Your clients are asking anyway, and we've argued that being the one with answers wins the relationship. The assessment is how the answer becomes an engagement with a fee attached.

Keep the scoring simple enough to defend in a boardroom. Our free AI readiness scorecard is a workable spine: eight questions across leadership, skills, workflows, and guardrails, each answered 0, 1, or 2, normalized to a 0-100 score with four tiers from Exploring to Compounding. Run it with the client's leadership team in the room, not as emailed homework. The argument over whether a question deserves a 1 or a 2 is where the assessment earns its fee, because that argument is the client discovering their own gaps out loud.

Then back the headline number with the five-part inventory. The scorecard says skills sit at 38 percent; the inventory names the six people who can't get a usable draft out of AI and the two who quietly automate half their week. One warning from the field: don't let the data dimension become a stall. Messy SMB data is mostly fine for the document-and-conversation work modern AI does best, so score data on access, not elegance.

The deliverable separates an AI readiness assessment from consultant theater. Ours runs about ten pages: a one-page scored baseline by dimension, the three to five workflows to rebuild first with an hours-saved estimate for each, a short fix list for tooling and policy gaps, and a 90-day plan with owners and dates. Build the hours-saved estimates the way you'd defend them later, in hours recovered against a baseline rather than vibes. Close with a live walkthrough. The client should be able to act on the document without hiring you again, which is exactly why most of them hire you again.

Published 2026 pricing guides agree on the shape more than the decimals. Aries Consulting Group puts a narrow small-business audit at $2,000 to $8,000 and a full mid-market assessment at $5,000 to $15,000 on a two-to-three-week timeline, and flags anything past four weeks as a red flag. ConsultKit's market roundup lands on the same bands: $1,500 to $3,000 for a light discovery tier, $5,000 to $15,000 for a comprehensive assessment, and $15,000 and up once an implementation roadmap and vendor evaluation ride along. Treat these as market context, not a rate card. Your number depends on the client's size, the compliance load, and your standing with them.

Three structural rules matter more than the exact figure. Price it fixed-fee, not hourly: the scope is knowable in advance, and a fixed price aligns you with finishing rather than billing. Scale by inventory, not ambition: more functions interviewed and heavier regulatory exposure justify the higher band, and a ten-person client doesn't need the $15,000 version. And anchor the fee against the follow-on. ConsultKit pegs downstream implementation work at up to 15 times the assessment fee, which makes the assessment the cheapest qualified pipeline a consultant can build.

For most SMB clients the natural follow-on is training rather than a build, because skills is usually the lowest score on the page and the one no software purchase fixes. Design for that from the start: the assessment names which roles need hands-on AI fluency work, and you either deliver that training or broker it and stay the advisor of record. It helps to be visibly fluent yourself first; that's the case we make in getting certified before your next engagement. Either way, the consultant who measured the gap is the one trusted to close it.