How to choose an AI readiness assessment provider
Hiring someone to assess your team is a strange purchase: you are paying for a judgement, and the people selling it also sell the work it recommends. That conflict is not disqualifying, but it does mean the questions you ask before signing matter more than usual. Here is what to ask, which answers should end the conversation, and when you should skip the provider entirely.
First, decide whether you need one
If you already know which task takes up most of your team's time, you do not need an assessment. Go fix that task. Assessments earn their place in two situations: your work is spread across many workflows and the team cannot agree which to tackle first, or you need a defensible number to get time and budget from someone above you.
If neither is true, a provider will sell you a diagnosis for a problem you have already diagnosed. Run the free assessment instead and keep the money.
The questions to ask before you sign
- What exactly do I get at the end? Push for the artefact, not the adjective. "A roadmap" means nothing. "A ranked list of your workflows with a score and an hours estimate against each" means something.
- Who do you talk to? An assessment that only interviews the leader measures what the leader believes. The difference between what a manager thinks the team's week looks like and what the team reports is usually the most useful thing in the whole exercise.
- How long, and what happens during it? Ask what the weeks are spent on. If most of the calendar is analysis rather than waiting on responses, ask what is being analysed.
- What have you scored before that looks like us? Not logos. Ask for a workflow they assessed at a team of your size and what the recommendation was.
- What would make you tell us not to hire you? A provider who cannot answer this has never turned work down, which tells you what the assessment will conclude.
Answers that should end the conversation
Some responses are a reliable signal to walk. A proposal where the assessment runs for months before anything improves is selling duration. A methodology built on maturity quadrants and industry benchmarks is selling market research; your team's position on someone's grid does not tell you what to do on Monday.
Be careful too with anyone who scores you on infrastructure you do not own. A small team assessed on model governance and data platform maturity will score badly on things it was never going to buy, and the recommendation that follows will be to buy them.
The conflict of interest, handled honestly
Almost every provider who assesses you also sells implementation. That is not automatically bad; the people who understand your workflows best after an assessment are the ones who just studied them. It becomes a problem when the assessment cannot produce any conclusion other than "hire us for phase two".
The test is whether the deliverable is useful if you never speak to them again. Ask directly: if we take this report and do the work in-house, is it complete enough to act on? A provider who hesitates is selling a lead magnet. I run assessments and I also sell the build, so treat that as disclosure rather than neutrality.
What good looks like
It surveys the team, not just the leader
The distance between what leadership estimates and what the team reports is where the real finding usually sits.
It scores your workflows by name
You should see your own run of show, your own budget process, your own survey analysis in the output, not a generic list of business functions.
It puts hours against each one
Hours are what turn a score into a business case. Without them you cannot argue for the time to act.
It names a first build
The week after the assessment should look different from the week before. If the report ends at insight, it ended early.
It tells you what not to automate
An assessment that finds every workflow suitable did not assess anything. Some tasks should stay human, and a good provider says which.
Frequently asked questions
- How do I choose an AI readiness assessment provider?
- Ask what artefact you get at the end, who they interview, and what they have scored for a team your size. Then ask what would make them tell you not to hire them. Choose the provider whose deliverable would still be useful if you never spoke to them again.
- Do I need an AI readiness assessment provider at all?
- Only in two cases: your work spans many workflows and the team cannot agree which to fix first, or you need a defensible number to secure time and budget. If you already know which task is costing you the most hours, skip the assessment and go fix it.
- Is it a problem that assessment providers also sell implementation?
- Not by itself, since the people who just studied your workflows understand them best. It is a problem when the assessment cannot reach any conclusion other than hiring them for the next phase. Test it by asking whether the report is complete enough to act on in-house.
- What are the warning signs in an assessment proposal?
- A multi-month timeline before anything improves, a methodology built on maturity quadrants and industry benchmarks, and scoring a small team on enterprise infrastructure it does not own. All three point to a recommendation to buy more rather than to change how the work runs.
Want to see how I run one?
The Team AI Readiness Audit surveys the whole team, scores your named workflows, and ends with an hours estimate and a first build.
See the audit