Part of our guide: AI strategy
Your business needs AI strategy support and you have seen the pitches, all of them promising transformation and all of them carrying a framework. Most of them are interchangeable, so the question that actually matters is how you choose between them with any confidence.
The consulting market for AI is noisy, spanning generalist consultancies that do AI the same way they do digital transformation and big tech vendors that call themselves consultants. It also holds boutiques and fractional executives, and they cannot all be good, so most of them are not right for you.
What actually matters when choosing a consulting partner:
Sector experience. This is non-negotiable, because insurance is not financial services is not legal, and the regulatory environment, the competitive pressures and the operational constraints are all different. A consultant who understands your sector can translate between your business language and AI language, and a generalist has to learn your sector from scratch, so you are paying for education, not expertise.
Ask directly: "How many businesses like mine have you worked with?" If they say "we've done work in insurance," ask for a reference in a similar-sized insurance business and ask what they learned. A good consultant can tell you three things that surprised them about your sector.
Senior delivery. This matters more than you think, because the initial strategy engagement often gets staffed with promising 30-somethings fresh from big tech and that is fine for some work. When you are making multi-million pound decisions about technology direction, you need someone who has actually lived through technology transformations. You need someone who can spot what your board will actually buy into and what will collapse under the weight of reality.
Ask who will actually be in your meetings, and if the answer is a partner at the consulting firm, that is a good sign. If it is a "subject matter expert" who reports to a partner who reports to a director, that is a warning flag. You want the senior person in the room, not a delegation.
Governance capability. Most consultants talk about AI governance like it is a checkbox: "You'll need a governance framework. Here's our RACI matrix." Real governance is harder: it is about decision-making under uncertainty, about knowing when to push back on business demand and about building a structure where smart people can move fast and regulators are satisfied.
Ask: "Walk me through how you'd set up governance for my business." A good answer includes: who owns what decision, what escalation looks like, how you measure whether governance is working and what happens when the model breaks. A generic answer tells you they have not thought about it.
Reference clients and evidence of outcomes. Testimonials are nice and evidence that clients actually used your recommendations and got results is better. Ask your consultant: "Can I speak to a client you worked with two years ago? I want to know whether they followed through on your recommendations and whether it worked."
That question is uncomfortable for consultants who oversell, and good consultants have a list of clients who will answer it credibly and tell you straight: "They helped us avoid three major mistakes and speed up our vendor selection process."
Red flags in consulting pitches:
Vendor-led recommendations. If a consulting partner keeps steering you towards a particular technology or vendor, you need to understand why. The reason is either that the vendor is actually the right tool for your problem or that the consultant gets a commercial arrangement with that vendor. That arrangement is more common than you would think. The consultancy makes money whether you succeed or not and the vendor makes money if you buy, so the incentives worth checking are the ones that align with yours.
Ask directly: "Do you have a commercial relationship with any of the vendors you're recommending?" If the answer is yes, that does not automatically disqualify them, and the information is something you need to weight.
Junior-heavy teams. Consulting firms love proposing a team with one senior person and five junior consultants, and that shape is profitable for them. It is expensive for you because you are paying for people who do not yet know your sector. If a proposal says "led by Partner X, supported by a team of analysts," ask for their CVs, because work experience matters.
No measurement framework. A good consulting partner will say: "Here's how we'll measure whether this engagement has been successful." Bad ones will sell you a 10-week strategy programme with no clear success criteria, and strategy work is valuable, though it has to be tethered to outcomes. Those outcomes are whether we are shipping faster, whether we are making better decisions and whether we are avoiding disasters.
Asking the right questions before you engage:
Ask: "Can you describe your approach to AI strategy in our sector specifically?" A good answer is detailed and specific, acknowledging the constraints in your industry and showing that they have worked through similar problems. A generic answer that sounds like it applies to any sector is a warning flag.
Ask: "How do you handle situations where your recommendation conflicts with what the business is asking for?" A good consultant will tell you about a time when they pushed back on a plan they thought was wrong, and they should be comfortable being the voice of reason, not just the person who rubber-stamps what leadership wants.
Ask: "What does success look like at the end of your engagement?" You should be able to point to something tangible: a prioritised investment roadmap, a governance framework that is actually in use, a vendor selection that has been validated and a set of quick wins that have started delivering value.
Ask plainly: "Who will we actually be working with?" Get their names and their background, and if most of the team is under 3 years' experience, question whether that is right for your complexity.
Evaluating proposals:
A good proposal is specific to your business, acknowledging the constraints you have mentioned. It includes a clear scope, a realistic timeline, a named team with relevant experience and success criteria you can measure.
A bad proposal is generic, could apply to any company, makes promises without evidence and carries a long list of deliverables with an unclear connection to your business outcomes.
A proposal that spends more time talking about the consulting firm's credentials than about your specific situation is selling consulting, not solving your problem.
The partnership that works:
The best consulting engagements are partnerships where you are not waiting for consultants to tell you what to do and you are working with them instead. They bring pattern recognition from other sectors and businesses and you bring intimate knowledge of your constraints and your culture, so together you move faster and make better decisions.
Get in touch to talk about whether an AI consulting partnership would help you move forward.