Diagnosis Is the First Step. The Proposal Rarely Is.

A consulting engagement typically begins with a competitive process the client controls: an RFP, a shortlist, a scored pitch. To win that process, a firm has to say something specific about what it will do, because a specific claim reads as more credible than a vague one on a scoring sheet. So the document that wins the work usually already names the shape of the fix — a target operating model, a new pricing architecture, a technology migration, a restructured reporting line — months before any diagnostic work has been billed.

This is not unique to consulting, but consulting makes it visible because the proposal is a written artifact that survives the sale. A pitch deck that says "we will help you think about your operating model" loses to one that says "we will consolidate four regional operating models into one, cutting approval layers from six to three." The second version is more likely to win, and it is also a commitment the firm has made in public, to the client's own selection committee, before a single interview has been conducted.

The diagnosis inherits a public answer

The diagnosis phase that follows a signed contract is genuine work. It is staffed, scoped, and timeboxed like any other phase of the engagement, and the people doing it are not being asked to fabricate a finding. But the phase inherits a constraint the client rarely sees written down anywhere: the shape of the answer is already public, printed in the document that won the business. A firm that named the solution to win the work has already spent its credibility on that answer before the diagnosis begins. Concluding something the pitch didn't say is not just intellectually harder. It is a commercial event — a firm telling the client the thing it paid to hear was wrong.

This produces a recognizable pattern once enough engagements are compared side by side, which is the value of a case-pattern view over any single case. It is not that any one diagnosis is dishonest. It is that a firm's attention, staffing, and library of prior work are built around one category of remedy, and a diagnosis run through that lens tends to surface the problem the lens is built to see.

• A firm whose practice is operating-model design tends to find that the operating model is the constraint, and proposes a redesign.
• A firm whose practice is technology implementation tends to find that the systems are the constraint, and proposes a migration.
• A firm whose practice is pricing and commercial strategy tends to find that the pricing architecture is the constraint, and proposes a repricing.

Run the same organization through all three, and each diagnosis would likely produce a different primary cause — not because two of the three are wrong, but because each firm's diagnostic method is built to notice the category of problem it is staffed to solve. The choice of method decides what a diagnosis is capable of finding before any evidence is gathered, and the firm's commercial specialty is itself an unstated choice of method, made before the client ever describes the problem.

The shape of the answer is already public, printed in the document that won the business.

The incentive is not subtle once it is named. A team under pressure to show early value on a signed engagement reaches for the finding that supports the plan the client already approved, because a diagnosis that contradicts the sold scope forces an uncomfortable conversation about fees, timeline, and who was wrong. It is far easier, and faster, for the diagnosis to arrive at a version of the problem the proposed solution already answers. The interviews get conducted, the data gets pulled, the workshop gets held — and all of it converges, with unusual reliability, on the conclusion that was printed in the proposal.

This is different from the diagnosis being skipped entirely, which is the more familiar failure. Here the phase exists, is staffed correctly, and produces a real document. What it lacks is the one property a diagnosis needs to be worth running: the live possibility that it concludes something the client did not already pay to hear.

The tell is rarely a fabricated finding. It is a diagnosis that takes longer to describe the problem than to explain why the problem happens to require exactly the solution already on the table, and a final workshop that spends more time getting buy-in for the plan than testing whether the plan is still the right one.

The structural fix is not more rigor inside the same sequence — better interview protocols do not solve a problem created by commercial timing. The fix is sequencing: diagnosis has to be sold, scoped, and paid for as a phase that can end in a finding the firm did not propose, before any solution-shaped deliverable exists to defend. That means the diagnosis is priced and contracted separately from the implementation work, with an explicit decision point between the two where the client can take the diagnosis and hire someone else, or do nothing, without the firm having lost the diagnosis fee for saying so.

A diagnosis with no exit is not a diagnosis; it is confirmation with better production values. The same discipline applies to what a diagnosis is built to describe in the first place — a description of the business, not a recommendation, and a description written by a firm that has already sold the recommendation is no longer free to be only that.

In practice this is a line in the contract, not a philosophy: a fixed fee for the diagnosis with its own deliverable, a defined gap before any implementation statement of work is signed, and a client who was told, before signing, that the diagnosis might recommend a different firm, a smaller project, or no project. A firm unwilling to write that gap into the contract is telling the client something about how confident it is in what the diagnosis will find.

None of this requires distrusting consultants generally. It requires noticing that the commercial process which selects a consulting firm and the diagnostic process that firm then runs are not independent, unless someone deliberately makes them so. The proposal that wins a pitch is written to be persuasive before the work starts; the diagnosis that follows is credible only to the extent it was never asked to agree with it.

This is close to how we structure consulting engagements ourselves: the diagnostic phase is scoped and priced as a standalone piece of work with its own stopping point, before any solution-shaped proposal is written, so the finding is free to be inconvenient. The broader discipline this sits inside is the same one running through every note here — that solving the wrong problem well is still a loss, and the only way to know the problem is right is to let the diagnosis end somewhere the pitch did not already decide.