Why Consulting Engagements Fail — and the One Diagnostic That Fixes It

The engagement that nobody wanted

The most common failure mode in consulting is not bad work. It is good work on the wrong question. A team of six spends fourteen weeks on a rigorous, well-sourced, beautifully-formatted answer to a question leadership was not actually asking. The deliverable is accepted, the invoice is paid, and the recommendations are never implemented.

Everyone is vaguely dissatisfied. The client blames the firm for being "not practical." The firm blames the client for being "not ready." The same pattern repeats on the next engagement. We now believe the entire failure mode can be diagnosed in the first meeting, with one question.

The question

"What decision will be made, by whom, on what date, that this work is meant to inform?"

Every word matters. Let me walk through why.

"What decision"

Not "what problem," not "what question." A decision — a specific fork in the road that leadership is facing and that, in the absence of new information, will be made in a specific way. The word forces the client to name a binary or a ranked set of options. "Should we build or buy this?" is a decision. "Help us understand our data strategy" is a topic. Topics are where engagements go to die.

"Will be made"

The decision is definitely happening. If the decision is not happening — if it is hypothetical — the engagement is research, not consulting, and should be priced as research.

"By whom"

A specific person or committee. Not "the executive team" — a name. If the client cannot name the decision-maker, the decision is not ready. Doing the engagement first is a waste of everyone's money.

"On what date"

A calendar date. Not "sometime in Q3." The date is the forcing function; it determines scope, depth, and format. Most failed engagements agreed on scope without agreeing on a date, and the date then revealed itself too late.

"That this work is meant to inform"

Inform, not "drive." Our work is an input into a decision the client owns. We produce better work when we remember whose decision this actually is.

Why the question does not get asked

Ask any partner at a large firm and you get three answers. First: the client cannot answer it. Partially true, completely beside the point. If the client cannot answer, the engagement is not yet ready to scope; the honest next step is a short diagnostic, not a long engagement that pretends the question has been answered. Second: answering would narrow the engagement. Also true, and why the question is taboo in firms whose revenue depends on scope expansion. A decision-centered engagement has a natural stopping point; a topic-centered one does not. Third: the decision-maker does not want their decision constrained. The most interesting one. Some executives prefer topic-scoped engagements because it preserves optionality to reject or reframe findings after the fact.

We would rather lose an engagement at the first meeting than take on work for a decision-maker who will not commit to what the decision is. Clients who will not answer the question will not implement the work either.

The practical effect

When the question is asked and answered, three things change immediately.

Scope collapses. A "six-month digital transformation diagnostic" becomes "a four-week decision memo on whether to consolidate the two CRM instances by end of Q3." The work is smaller, faster, and more useful.

Format changes. The deliverable is not a 140-page deck. It is a 12-page decision memo with a recommendation, an options comparison, a reversibility analysis, and a clear answer to the question. Harvard Business School case-study format, not management-consulting hairball format.

Price changes. The engagement costs less in absolute dollars and more per day, because the work is concentrated. Clients get a better answer for less money. The firm earns a better margin for less time. Both sides win. The only loser is engagement length, which is the firm's revenue metric but the client's cost metric.

The diagnostic we use in first meetings

We run a short exercise in every first meeting. On a single page: the decision (one sentence, binary or ranked options), the decision-maker (named), the date (specific), the three to five pieces of information that would change the answer, and what we would do in the next 72 hours to start gathering them. If we can complete that page together, we have an engagement. If we cannot, we stop. Either way, the first meeting ends with either real work or real clarity.

Henry Mintzberg has a line we return to often: "Strategy is not about planning; it is about the pattern that emerges from the decisions you actually make." Work that does not connect to a specific, named, dated decision is not strategy. It is content. The firms worth working with produce less content and more decisions. The question above is the test.

At Cadence Advisors Group, we do not start engagements we cannot name the decision behind. If you have an AI decision coming up and need it informed well rather than decorated, [that is the work we do](/contact).