Hiring an AI adoption consultant: 8 questions to ask

"AI adoption consultant" is a new title, and it covers very different profiles: trainers, technical experts, change management specialists, strategy firms. All of them can be useful. But if your problem is that the licences are bought and the tool is barely used, you need someone who knows how to change work habits, and who proves it with numbers.

Here are the eight questions I would ask in your place. I am a consultant myself, so read them also as the grid you can judge me on.

1. "How will we know it worked?"

This is the most important question. A good answer names specific usage measures: weekly active users, tasks covered, quality of outputs, drop-off. A weak answer talks about the number of people trained or session satisfaction scores. Those measure activity, not adoption.

2. "Tell me about a rollout that did not work."

Someone who has really led rollouts has failures to talk about, and what they learned from them. Be wary of a track record without a single setback.

3. "Where would you start with us?"

The right answer starts with your teams, not the tool: which tasks, which roles, which data. If the first week on offer is a generic training plan, the rest of the engagement will be generic too.

4. "Who does the work?"

In a firm, the person who sells is not always the person who delivers. Ask who will be there every week, and meet them before you sign. With an independent, the question becomes: what happens if they are away?

5. "What is left when you leave?"

Good support makes itself unnecessary. At the end you should have trained champions, written rules, a dashboard your teams can update, and a clear decision on what comes next. If the answer is "a phase two", ask why.

6. "How do you handle data and the legal side?"

The consultant is not your data protection officer, but they should know what Article 4 of the AI Act asks on AI literacy, what the GDPR means for data typed into the tool, and when employee representatives need to be involved. If they brush the question aside, take note.

7. "What do you think the tool’s limits are?"

An honest consultant will tell you where AI gets things wrong, which tasks are not worth it, and when it is better not to roll out. One who promises gains everywhere is selling the tool, not its adoption.

8. "How do you work remotely?"

A lot of this work is done remotely now, and it works when the rhythm is clear: fixed check-ins, well-prepared video workshops, an open channel for questions. Ask for an example of a typical week.

Format and price

A short first engagement with a clear deliverable lets you judge on evidence before committing to a programme. That is how I work: a diagnostic of a few weeks, then more only if the numbers justify it. Formats and rates are on the services page, and the first call is free.

If you are writing a job ad rather than hiring a consultant, the same questions apply. I analysed 146 AI adoption job ads: most of them skip the first one.