How should you use these questions?
Put the same questions to every consultant you speak to and write down the answers. The aim is not to catch anyone out. It is to find out, before you sign, what you are actually buying. Many engagements go wrong because the firm assumed it was buying a working system and the consultant assumed it was selling advice.
This guide is a question list. If you are still deciding what kind of consultancy you need at all, such as a strategy house, a software reseller or an implementation specialist, start with how to compare AI consultancies, which sets the types side by side. If you are weighing outside help against hiring, see consultant or in-house.
What will we have at the end?
1. What exactly will exist when the engagement finishes?
A good answer names a working thing: this process, running in your firm, used by your team on live work. A weak answer describes a report, a roadmap, a set of recommendations or a pilot. Those can have value, but know that you are buying a document.
2. Which process will you change first, and why that one?
A good answer is specific to your firm and ties the choice to hours lost and how often the job runs. A weak answer proposes to look at everything.
3. How will we know it worked?
Listen for a measure you can check yourself, such as the time a job takes before and after, and whether the team is using it on every piece of live work. Be wary of success defined as delivering the agreed documents.
How is the fee set?
4. Is the fee fixed, and what could make it change?
A good answer is a fixed fee agreed in writing before work starts, with clear reasons for how it was set. Daily rates and open ended time and materials put the risk of overrun on you. Our guide to how AI consultancy is priced explains the common models, and fixed fee or daily rates sets out the trade off.
5. What is not included?
Ask directly about software licences, changes after go-live, extra processes and support. A good consultant answers this without hesitation.
6. What happens if it is late or does not work?
This is the most revealing question on the list. A good answer carries a real commitment, such as a fee that cannot move and a sign-off by your team before anything goes live. A weak answer talks about shared responsibility and change requests.
Who will do the work?
7. Who will we actually work with day to day?
You want to meet the people who will map and build the process, not only the people who sell it. Ask whether any of the work is subcontracted.
8. What have you built that is running now?
A good answer describes systems in use today, in enough detail that you understand what they do. Slides, prototypes and demonstrations are not the same thing.
9. Do you understand how firms like ours work?
Listen for the vocabulary of your sector: matters, engagements, schemes, placements, fee earners, charge-out rates. A consultant who needs your working day explained from the ground up will take longer to get anything right.
How will our data and clients be protected?
10. Where will our client data go, and who can see it?
A good answer names the tools and providers involved, where data is processed, whether inputs are used for training, and how access is controlled. See data protection and AI on client work for the questions behind this.
11. Will you work under our confidentiality terms?
The right answer is yes, under the same agreement your clients expect of any supplier, with nothing you share used for anyone else.
12. How is output checked before it reaches a client?
Listen for a named reviewer built into the process. Anything suggesting AI output goes straight to clients unchecked is a warning sign.
What happens to our team?
13. Will our people need to learn new software?
The best answer is that the process runs in the tools your team already uses. Every new platform adds training, licences and resistance.
14. How will the team be trained, and on what?
Good training uses your own live work, not generic examples. See getting a team to adopt AI.
What happens after go-live?
15. What support do we get once it is live?
Ask how long support lasts, what it covers and how quickly issues are answered. A period of support while the new process beds in is a sign the consultant expects to stand behind the result.
16. Who owns what you build, and can we run it without you?
A good answer is that the process is built on your documents, templates and systems, documented, and runnable by your team. Ask what any ongoing fee covers, and what happens to your documents, data and configuration if you stop.
How do our own answers compare?
It is fair to put these questions to us. We rebuild one process at a time. The fee is fixed and agreed in writing before work starts, and nothing goes live until your team signs it off. Changes outside the agreed scope are quoted in days before any work starts. We scope the job with your people, build it on your own systems and documents, then run it in parallel with the current process so your team can check every result. It goes live only once they sign it off, and support carries on under a subscription that also covers hosting, monitoring, a monthly performance report and a quarterly review. Timing is agreed at scoping. A named person in your team signs off every decision the system makes.
How we work sets out each stage in detail, and the audit takes about three minutes and names the process to start with.