How to Vet an AI Consultant Before You Hire One
The red flags to watch for, and the specific questions that surface them before you pay anything.
One story that circulates a lot in small-business circles: a florist paid roughly £2,000 for an "AI marketing package" from an agency, and ended up with a chatbot on her website that nobody ever used. No process change, no time saved, no measurable outcome — just an invoice and a widget nobody clicked.
That story sticks around because it's not rare. AI consulting is a genuinely useful service done badly by a lot of people right now, and the bar for "sounding credible" is currently much lower than the bar for "actually helping." Here's how to tell the difference before you pay for anything.
1. Do they guarantee a number?
Red flag: "Guaranteed 40% more leads" or "double your revenue with AI," without a clear explanation of the mechanism. Nobody can promise a business outcome from a single tool — too many other variables are involved, and anyone claiming otherwise is selling confidence, not analysis.
Ask instead: "What specifically will this save me, and how will we measure it?" A real answer sounds like "roughly 4 hours a week on quote drafting, checked by comparing your time-per-quote before and after." A fake one sounds like a percentage with no mechanism attached.
2. Do they recommend the same stack to every client?
Red flag: The identical platform or tool pitched regardless of your industry, size, or existing systems. That's optimizing for the consultant's commission or convenience, not your actual fit.
Ask instead: "Why this tool over the alternatives, for my specific situation?" If the answer doesn't reference anything specific about your business, that's the tell.
3. Is there a written scope before you pay for anything beyond an initial assessment?
Red flag: Vague promises, no defined deliverable, no timeline, no clear definition of "done." This is exactly the shape of the florist's chatbot — a fuzzy "AI package" instead of one named, scoped thing.
Ask instead: "What exactly gets delivered, and when?" You should get a specific answer in writing before a larger payment, not after.
4. Can they give you a straight answer about your data?
Red flag: A vague or evasive answer to "what happens to my customer data," or no answer until you ask twice.
Ask instead: "Does this tool train on my data, and what's the retention policy?" A consultant who actually knows what they're doing has this answer ready, not researched on the spot.
5. Do you own what gets built?
Red flag: Prompts, configurations, and accounts that live only in the consultant's head or their own login, with no handoff if you want to leave.
Ask instead: "If I stopped working with you tomorrow, what would I be able to keep and run myself?" If the honest answer is "nothing," that's dependency by design.
6. Do they ever say no?
Red flag: Enthusiastic yes to every idea you bring, including the ones that are premature or poorly scoped. A consultant optimizing for the sale says yes to everything; one optimizing for your outcome says no sometimes.
Ask instead: "What would you tell me not to build yet?" If they can't name anything, they haven't actually thought about your business — they've thought about closing the deal.
Run this on anyone you're considering
These six questions take about ten minutes in an email and will surface almost every version of the florist's story before you're the one telling it. A longer, more detailed version of this checklist — including how one specific practice grades against every item on it — is at is your AI consultant legit. Worth running on anyone, including whoever wrote this.