How Consultants Diagnose a Business Problem
The named problem is one observation about the business, not the finding. Diagnosis is the discipline of not stopping there.
A diagnosis begins by treating the reported problem as one observation, not the case.
A client calls with a named problem. Churn is up in a segment, a launch missed its window, a region is short of its number. That sentence is where the engagement starts, and it is almost never where the finding ends up. The named problem is real — the number moved, the date slipped — but it is a symptom read from one seat in the building, and the seat determines what gets seen.
What gets named first
A sales lead watching pipeline names a pipeline problem. A product lead watching usage names an adoption problem. Each is reporting an honest reading of the instrument they carry, and each instrument only shows part of the system. The first move in diagnosis is holding the named problem loosely long enough to ask what else could produce the same signal, rather than accepting the first plausible cause because it arrived first and matches what the reporting seat already suspected.
This is not scepticism toward the client. It is the same discipline applied before a brief is even written: the visible complaint and the underlying condition are frequently two different things wearing the same name, and the diagnostic work is the process of telling them apart before a solution gets built for the wrong one.
What people say in an interview and what the organisation actually did rarely agree completely.
Interviews are where a diagnosis starts, and they are also where it is most likely to be misled, because every person interviewed has a theory of the problem before the conversation begins, and the theory usually protects their own part of the system. This is not dishonesty. It is what it looks like when someone has lived inside a problem long enough to have a story about it.
The correction is not to distrust the interview. It is to treat it as one kind of evidence among several, and to weigh it against records nobody produced with this argument in mind:
- Calendars and where senior time actually goes, against where the strategy says it should go.
- Usage or transaction data, against the version of customer behaviour repeated in the room.
- The gap between a written policy and what a spot-check of five recent cases shows people actually doing.
None of these sources settles the question alone. A pattern that shows up in the interviews, the calendar, and the data at the same time is a finding; a pattern that shows up in only one of them is a lead that needs the other two before it earns a place in the diagnosis. This is the same triangulation a diagnostic run before a consultant is hired is built to produce internally, done here from the other side of the engagement.
A working hypothesis earns its place by surviving an attempt to disprove it.
By the second or third week, a diagnosis usually has a candidate cause — a hypothesis about what is actually producing the symptom, distinct from what was originally named. The temptation at this point is to start building toward a recommendation, because a plausible cause feels like an answer and clients are, reasonably, waiting for one.
The discipline is to spend deliberate time trying to break the hypothesis rather than confirm it. What would we expect to see in the data if this cause were wrong? Is there a segment, a region, or a time period where the pattern should hold and does not? A hypothesis that only gets tested against evidence that already supports it is not a finding — it is the first guess wearing a diagnosis's clothing.
The evidence that reaches a diagnosis was gathered by people who wanted it to say something. The test is what happens when it is asked to say the opposite.
This is closest to the concern behind reading the business before framing a major decision: a description assembled to confirm a starting position is not a description, whatever its author intended. The difference here is that the discipline is applied earlier, before a decision is even on the table — to the cause itself, not yet to what should be done about it.
Diagnosis has a stopping point, and where it stops decides what gets proposed.
A diagnosis could, in principle, keep widening — every cause has its own causes, and a determined enough inquiry can trace churn back to a hiring decision made three years earlier. The method includes deciding where the diagnosis stops, and that decision shapes the recommendation as much as any finding inside it does.
Where the line gets drawn
Stop too early and the recommendation treats a symptom as though it were the cause — the same failure the diagnosis exists to prevent, just relocated one level down. Stop too late and the finding is technically correct and practically unusable, because it names a cause the organisation has no near-term ability to act on. The judgment call is not whether to stop, but where a cause becomes something the client can actually own and change within the timeframe the engagement is meant to serve.
That judgment is where a consultant's experience shows up most directly, more than in any framework or interview technique. Two competent people can run the same interviews and read the same data and draw the stopping line in different places, and the diagnosis that results will point toward genuinely different work.
Naming the actual cause is where an engagement's real cost is spent, before a proposal exists.
Clients often experience the diagnostic phase as overhead standing between them and the recommendation they came for. In practice it is the opposite: the diagnosis is where the expensive thinking happens, and the recommendation that follows a sound one is usually straightforward by comparison, because the hard question — what is actually producing this — has already been answered.
This is the shape of the work we do before a scope is written: separating the reported problem from its cause, weighting interviews against records nobody wrote for this argument, and testing the resulting hypothesis against its own best attempt to disprove it. It sits alongside the broader discipline of diagnosis before solution that runs through how we scope any engagement, and it is available as a shared method a team can run itself, for the cases that do not need an outside diagnosis at all.