Questions to ask an AI consultancy before you sign
Eleven questions that separate a firm with judgement from a firm with a deck, and what a good answer to each one sounds like.
The questions worth asking are not about capability, because every firm will say yes to all of them. They are about accountability. What would you tell us not to build, who specifically does the work, what happens if the number does not move, and what does this cost. A firm with judgement answers those quickly and with examples. A firm without one changes the subject to what it can build, which is the answer.
Every firm on your shortlist can do the work. That is not the thing in doubt, and asking about capability produces an hour of yes.
What separates them is what happens when the work is harder than the proposal assumed, and you can find most of that out before signing by asking eleven things.
About judgement
1. What did you last talk a client out of building?
The single most useful question available. A firm that sells judgement has an answer ready and will give you a specific example with a reason. A firm that sells capacity will struggle, because talking a client out of work is a thing it has never had a commercial reason to do.
Listen for whether the example cost them money. Anyone can decline work they did not want.
2. What would you need to see before telling us this is a bad idea?
Follow-on to the first. You are asking whether "no" is an available output. If the engagement structurally cannot end in a recommendation to stop, then the assessment is a formality and you are buying a build with extra steps.
3. What is the most likely reason this fails?
A firm that has shipped things has a ready list and it is usually dull: access to a system takes six weeks, the data is worse than described, the person who knows how the process works leaves. A firm that answers with adoption risk and change management is reciting.
About the people
4. Who specifically does the work, and how often will we see them?
The oldest problem in consulting is a quote priced on a senior name and delivered by whoever is free. Ask for names. Ask whether the person in the room today is in the room in week three.
5. What is your team doing when they are not on our project?
You are asking about capacity honestly. A firm running four engagements with three people has a problem that becomes your problem in month two.
About the money
6. What does this cost?
Obvious, and worth asking early rather than late, because the answer is often a sales call. When we read seven AI consultancies' websites in September 2026, six published no price at all. If a firm will not give you a band on the first call, you will be comparing shortlists on something other than money.
7. What is not in the estimate?
Better than asking what is. Every estimate has exclusions and the good ones are explicit. Listen for integration, data cleanup, the second approach after the first one fails, and who pays for the rebuild if an assumption was wrong.
8. What are the payment terms, and what is the notice period?
A retainer with a twelve-month minimum is a firm protecting its revenue. One month's notice says they expect to keep earning it.
About the result
9. What number will this move, and who agrees it beforehand?
The measurement question. Ask when the baseline gets taken, because a baseline taken after work starts is not a baseline.
10. What does your closing document say if the number does not move?
This is the one most firms have no answer to. It is also the one that tells you whether the engagement is designed to produce a result or a report. There is a version of this where the honest answer is that nothing moved, and a firm that has never written that document has never been in a position where it had to.
11. Can we talk to a client where it went badly?
You will often get a no, and the no is informative. A firm that can put you in front of a client whose project was difficult is a firm that handled it well enough to still be on speaking terms.
What the answers look like side by side
A firm with judgement gives short answers with examples and says "I do not know" at least once. A firm with a deck gives long answers, moves toward capability, and treats every question as an objection to handle.
The tell is the length. People who have done the work answer quickly because they are recalling rather than constructing.
One limit worth stating
Eleven questions is more than an hour will hold, and a shortlist call where you work through a list reads as an interrogation rather than a conversation, which gets you worse answers. If you have thirty minutes, ask 1, 4, 7 and 10. Those four cover judgement, delivery, honesty about scope, and accountability for the outcome, and the rest of the list is mostly elaboration on them.
Before the calls
Ask the same questions of each firm in the same week and write down the answers. Memory flatters whoever you spoke to most recently.
It also helps to have written the brief first, because a vague brief produces answers you cannot compare. How to brief an AI project so the quotes are comparable sets out what to put in it. What the work costs is in what AI consulting actually costs, and the wider framework for choosing is in how to choose an AI consultancy.
We answer all eleven on the first call, including the ones that do not flatter us. Thirty minutes, no deck.
If you have a budget, a deadline, and no clear answer on which AI project deserves either, that is the conversation to have.
Book the callWritten by
Radwan Altaf
Radwan runs AISynq. Before that he delivered software inside enterprise programmes at DHL, AT&T, DirecTV and Accenture, which is where the habit of measuring a result against its baseline came from. More about the firm.
Read next
- The AI project brief template: fill this in before you ask for a quoteA one-page template with the nine fields that make three quotes comparable, what to write in each, and the two answers that are allowed to be "we do not know".
- The AI vendor scorecard: compare three firms on one pageA scoring template for comparing AI consultancies on the twelve things that predict whether an engagement works, with what each score means.
- How to brief an AI project so the quotes are comparableFive things to write down before you contact anyone, and why a vague brief produces three quotes that differ threefold for reasons nobody can explain.
Work with us
Have a budget, a deadline, and no clear answer on which AI project deserves either?
Book the callGet the next one
New writing in Buying AI help as it goes up, roughly twice a month. One article per email and nothing else in it.