The Brief Is the Deliverable: What to Demand From Consultancy in 2026

Somewhere between the pilot and the invoice, the brief lost its edges.

The pattern is now familiar. A UK client approaches a consultancy with a genuine problem: margin erosion in a service line, a supply chain that no longer behaves, a finance function stretched by growth. The consultancy responds with a proposal of considerable polish. It references AI-enabled analytics, accelerated discovery, intelligent workflow. It is beautifully formatted and almost entirely unfalsifiable.

This is not new. Vague consultancy has existed for as long as consultancy has. What has changed is that AI has made vagueness cheap to produce and difficult to detect, because the language surrounding it carries an implicit promise of technical rigour. The result is a market in which the least precise proposals often look the most sophisticated.

The argument of this article is straightforward: in 2026, the competitive advantage for a buyer lies not in understanding the technology but in refusing to accept a brief that cannot be tested.

A new source of imprecision

A capable supplier can now generate a forty-page proposal in an afternoon. Sector context, a maturity model, a phased roadmap, a benefits case. The generative tools do not invent the structure; they assemble it from the conventions of the genre. What they cannot do is establish whether any of it applies to the client in front of them.

Buyers, meanwhile, are reading volume as diligence. A thick document signals effort, and effort signals seriousness. Procurement teams under time pressure skim the executive summary and compare day rates. The middle of the proposal, where the actual mechanism of change should be described, goes unexamined.

Why inflated promises survive scrutiny

Three structural reasons. First, many procurement processes are designed to compare inputs rather than outcomes, because inputs are easier to price. Second, framework agreements create an assumption of quality that discourages challenge. Third, and most corrosive, clients often lack a clear view of what success would look like, and are quietly relieved when the supplier supplies one on their behalf.

That relief is the root of most disappointment. The success definition then belongs to the supplier, and it is usually drawn to be achieved: a maturity score improved, a roadmap delivered, a set of workshops completed. None of these is a business result.

Seven questions to ask before signing

1. What decision will this engagement allow us to take that we cannot take today?
2. Which of these deliverables would you drop if our budget were reduced by a third?
3. Who, by name and job title, is accountable on your side for the outcome, and how much of their time is committed?
4. What would cause you to recommend that we stop?
5. How will we measure success in plain English, without reference to a maturity model?
6. What are you assuming about our data, our people and our capacity to absorb change?
7. What does the final fortnight of the engagement look like, and who is doing what?

The answers are revealing. A supplier who can answer question two without hesitation understands the difference between deliverables and decoration. A supplier who cannot answer question four has not considered failure, or does not wish to discuss it.

Success measures that survive contact with reality

A useful measure has four components: a baseline, a direction, a date and an owner. For example, reducing average invoice query resolution from nine working days to five by 30 September 2026, measured from the finance system, owned by the head of shared services. That sentence can be audited. It can also be missed, which is precisely why it is worth writing down.

Composite indices, capability scores and maturity levels are not measures. They are translations of measures, and the translation is done by the party being measured. Where a supplier insists on such a framework, ask for the underlying operational metrics it is derived from, and then contract on those instead.

Commercial terms worth the argument

Payment should be tied to stages, not to elapsed time. A first phase that produces a findings report, followed by a gate at which the client decides whether to proceed, protects both parties from a slow drift into an engagement neither would have chosen. The second phase should be priced separately, even at a modest premium, so the first decision remains genuinely open.

Three clauses deserve attention. Exit provisions, including what is handed over and in what format if the engagement ends early. Data and IP, particularly where AI tools are used and client data is processed. And change control, because scope creep in a vague brief is not an exception; it is the expected outcome.

The counter-argument, considered

There is a fair objection. Some genuinely exploratory work resists precise scoping, and over-specifying it can strangle the insight that made the engagement worthwhile. That is true, and it is an argument for labelling such work honestly rather than for leaving everything vague.

Exploration can be bounded: a fixed number of weeks, a fixed fee, a defined decision to be taken at the end, and an explicit statement of what the client will do with the findings. What it should not be is an open-ended retainer described as a transformation programme.

What good buying looks like in 2026

The buyer's role is closer to that of an editor than a customer. An editor does not write the piece, but insists that it says something, that the argument holds, and that the ending is earned. Applied to consultancy, that means asking what the work will change, how the change will be recognised, and who will answer for it if it does not arrive.

None of this requires hostility towards suppliers, and none of it is a reason to avoid AI-assisted methods. It is simply a refusal to accept polish in place of precision. A sharper brief costs an afternoon to write and is the cheapest form of risk management available to any UK organisation commissioning advisory work this year.