# How to brief an AI consultancy: a guide for business leaders How to brief an AI consultancy effectively — a practical guide for business leaders on scoping, data, success criteria and choosing the right firm. Published: 2024-12-26 Author: Rodan Analytics Most AI engagements fail before they start. Not because the technology is wrong, not because the consultancy is incompetent, but because the brief was vague. The business leader walked in with a problem they had not fully defined, the consultancy shaped the engagement around what they could sell, and six months later nobody could agree on what success looked like. If you are a senior operator evaluating AI consultancies right now, the most valuable thing you can do is not find the right firm. It is learn how to brief one. A well-constructed brief will expose weak consultancies immediately, sharpen your own thinking, and compress the time between first conversation and first result. This article will show you how to write a brief that commands serious attention, the questions you should be able to answer before any external firm gets involved, and the signals that tell you whether you are talking to people who will challenge you or people who will simply agree with you. ## Start with the commercial problem, not the technology The most common mistake business leaders make when approaching an AI consultancy is leading with the solution. "We want to build a chatbot." "We want to use machine learning on our data." "We need an AI strategy." None of those are briefs. They are technology preferences dressed up as business problems. They immediately constrain the consultancy's thinking and, more importantly, they let a weak consultancy off the hook. If you tell them what to build, they will build it, bill you, and leave. A proper brief starts one level up. What is the commercial problem you are trying to solve? Where is the money leaking, the decision-making slow, or the customer experience breaking down? A mid-market logistics business, for example, might frame it like this: "Our operations team spends 40 percent of their time manually reconciling carrier data across four systems. This delays invoicing by an average of eleven days and costs us roughly £800k annually in working capital." That is a brief. It tells you the domain, the pain, the scale and the measurable cost. It invites a consultancy to bring genuine thinking rather than a pre-packaged response. Before you contact anyone, write one paragraph that describes the commercial problem in those terms. If you cannot, the work is not ready to be outsourced. ## Define what you actually own before the engagement begins One of the most frequent causes of failed engagements is a data reality that only surfaces three weeks in. The consultancy assumed clean, accessible data. The client assumed they had it. Neither checked. Before you brief anyone, conduct an honest internal audit across four dimensions: - **Data availability** — does the data you need to solve this problem actually exist, and where does it live? - **Data quality** — is it clean, consistent and current, or is it partial, duplicated and manually maintained? - **Access and governance** — who controls the data, what legal or compliance constraints apply, and how quickly can access be granted? - **Internal capability** — do you have someone who can act as an intelligent client throughout the engagement, or will the consultancy be working in a vacuum? A retail business preparing to brief a consultancy on demand forecasting might discover that their sales data is split across three ERPs following an acquisition, that two years of records were migrated incorrectly, and that their IT function requires a twelve-week procurement cycle for any new data access. That discovery, made before the brief goes out, saves everyone months of frustration. Made afterwards, it kills the project. Be honest about what you have. A credible consultancy will not walk away from a messy data environment — they will price it correctly and scope accordingly. A consultancy that promises results without asking these questions is the one to avoid. ## Be explicit about scope, constraints and what good looks like Vague briefs produce vague proposals. If you want sharp, comparable responses from consultancies, you need to give them the same set of constraints to work within. Your brief should cover five things: - **Scope** — what is in and what is explicitly out? Name the systems, teams, geographies or processes involved. - **Timeline** — when do you need an outcome, and are there hard deadlines driving that? A private equity portfolio company preparing for an exit in eighteen months has a very different sense of urgency than a corporate running a long-cycle transformation. - **Budget range** — you do not need to name an exact figure, but giving a range prevents both parties wasting time. A £30k diagnostic and a £300k transformation programme require completely different proposals. - **Internal stakeholders** — who will be involved, who has sign-off authority and who has the power to block progress? - **Definition of success** — what does a good outcome look like in twelve months? Try to express this in commercial terms: cost reduced, revenue influenced, decision speed improved, risk mitigated. That last point is the one most leaders skip. Without a shared definition of success, you cannot evaluate the engagement honestly, and the consultancy has no accountability. It is worth spending time on. ## Know what kind of engagement you are commissioning Not all AI work is the same, and conflating different types of engagement is a reliable route to disappointment. There are broadly three categories of work. The first is diagnostic and strategy — understanding the landscape, identifying the highest-value opportunities and building a roadmap. This is typically lower cost, shorter duration and the right place to start if you are uncertain about priorities. The second is build and deploy — turning a defined opportunity into a working system. This requires committed internal resource, clear ownership and a production environment to deploy into. The third is ongoing managed capability — where the consultancy operates or augments a function over time, rather than delivering a project and leaving. Most organisations need to start with a diagnostic before committing to the second or third category. A structured, paid diagnostic — typically costing between £1,000 and £2,000 — will surface the genuine constraints, validate the commercial case and tell you whether a larger engagement is warranted. It also tells you a great deal about how a consultancy works: do they challenge your assumptions, or do they confirm them? If a consultancy skips straight to proposing a large transformation programme without first understanding your data, systems and internal capability, that is a signal worth heeding. ## What a strong brief signals to the right consultancy A well-written brief does more than organise your thinking. It qualifies the responses you receive. Send the same brief to three consultancies. One will respond with a generic deck that could have been written for anyone. One will ask for a call before responding — which may be appropriate, but watch whether they use that call to understand your problem or to pitch their credentials. One will ask sharp, specific follow-up questions: about the data environment, the stakeholder map, the constraints you did not mention but probably have. The third response tells you who is actually going to help you. A consultancy that challenges your brief — that says "you have framed this as a forecasting problem but it sounds like it might be a data integration problem first" — is a consultancy worth continuing the conversation with. The brief is not just a procurement document. It is a test. Use it as one. ## What to do next The organisations that get value from AI engagements are not the ones with the biggest budgets. They are the ones that arrive with clear thinking, honest constraints and a commercial outcome they genuinely care about. The brief is where that clarity gets built. Write the one-paragraph commercial problem statement before you speak to anyone. Audit your data honestly. Define what success looks like in terms you can measure. If you have done that and you are ready to talk to a consultancy that will challenge your assumptions rather than validate them, start with a diagnostic. It is a low-cost way to find out whether the opportunity is real, whether the data supports it and whether the engagement is worth commissioning at scale. Rodan runs structured diagnostic engagements designed for exactly this stage. If you are a business leader with a defined problem and an honest sense of your constraints, [book a diagnostic conversation](https://rodan.io) and we will tell you plainly what we think. --- **Meta description:** How to brief an AI consultancy effectively — a practical guide for business leaders on scoping, data, success criteria and choosing the right firm. HTML: https://rodan.io/insights/how-to-brief-an-ai-consultancy-a-guide-for-business-leaders