Diagnose the problem before buying the solution

A useful solution follows evidence about the cause of the problem; buying first can make the visible symptom more expensive without changing its source.

The team says sales are weak, so somebody proposes advertising. Another person recommends a new salesperson. A third wants a cheaper price. All three suggestions may be reasonable, but none establishes why sales are weak. Before choosing a remedy, define what is happening precisely enough for evidence to distinguish the possibilities.

Diagnosis is not a ceremonial delay before action. It is the work that makes action relevant. A business can move quickly and still spend a little time checking whether the proposed solution addresses the actual constraint.

Replace the complaint with an observable gap

Sales are poor is a complaint. Qualified enquiries fell over the last several weeks while the proportion of enquiries becoming orders remained similar is a more useful observation. It points towards a different investigation from enquiries stayed steady but fewer proposals were accepted.

State the measure, period and comparison. Identify where the information comes from. If reliable records are missing, make that absence part of the problem statement. A precise sounding number reconstructed from memory should not be treated as stronger evidence than it is.

Avoid starting with a person's name as the explanation. The sales manager may have made mistakes, but the investigation should examine lead quality, response time, stock, pricing, offer clarity, incentives and customer objections. Blame narrows the search before the cause is understood.

Separate observations from explanations

Create two short lists. The first contains things that were directly observed. The second contains possible explanations. A customer said the delivery date was too late belongs in the first list. Customers think we are too expensive belongs in the second unless the evidence supports it.

Keep competing explanations alive long enough to compare them. If the preferred answer is a new website, ask what evidence would show that the website is not the main constraint. Perhaps most enquiries arrive by referral and the delay happens after a proposal is sent. In that situation, the proposal process deserves attention.

This approach does not require certainty. It requires an honest distinction between what is known and what is being inferred. A useful decision can be made with incomplete information if the uncertainty and consequences are visible.

Follow the work where it happens

Look at a small sample of actual transactions or tasks. Trace what happened from the initial request to the result. Note waiting, missing information, repeated work, unclear responsibility and exceptions. Ask the people doing the work to show the process rather than only describing the official version.

In a fictional service business, a complaint about slow staff might conceal a queue of work awaiting owner approval. Adding another employee would increase the queue if the approval rule remained unchanged. The appropriate first action could be a clearer decision boundary rather than more headcount.

Do not generalise from one unusual incident. Compare ordinary cases with failures and successes. Ask what differs. The sample should help identify a pattern worth testing, not become an excuse to declare a universal cause after one conversation.

Choose a test that can change the decision

A useful test distinguishes explanations. If response time is suspected, record how long qualified enquiries wait and whether a faster response changes the next step. If offer clarity is suspected, test a clearer proposal with comparable prospects. Define what you expect to learn before observing the result.

A test does not have to be statistically elaborate to be useful, but its limits matter. A handful of conversations can reveal confusion without proving a market wide percentage. An improvement during a busy season may not be caused entirely by the change you made.

The strongest question is often what evidence would make us change our minds.

Put the decision in order

Some actions are sensible regardless of the final explanation. Correcting missing records, clarifying ownership and removing an obvious error may be reversible improvements. Other actions, such as a major hire or a long contract, deserve stronger evidence because reversing them is harder.

Rank decisions by consequence and dependency. If the business cannot explain its current offer economics, a large advertising commitment may be premature. If customers cannot receive an essential service, immediate operational action may be necessary while the deeper investigation continues.

End the diagnosis with a plain problem statement, the evidence available, the most plausible explanations, the next test and a named decision owner. That is more useful than a long report whose conclusion is that everything needs improvement.

How do you compare competing explanations?

Write at least two plausible explanations for the observed problem. If deliveries are late, one explanation may be insufficient transport and another may be orders reaching dispatch after the cutoff. Each suggests different evidence. Examine actual timestamps, workload and exceptions rather than asking people only whether they feel understaffed. A good diagnostic question makes it possible for the preferred explanation to be wrong. That is how the investigation earns the right to recommend spending.

Look for cases where the problem did not occur. An order delivered on time under similar conditions can reveal what was different. Perhaps the information was complete, the approval arrived earlier or the job followed a simpler route. The successful case is not proof by itself, but it can help narrow the inquiry. Comparing only failures may hide the operating condition that already allows the business to perform well when it is present.

What makes evidence useful?

Evidence needs a clear source, period and definition. A complaint that everything is always late is a starting signal, not a measured description. Identify the promised date, actual date and scope of the sample. Distinguish records from recollection and estimates from direct observations. You do not need perfect information before acting, but you do need to know how uncertain the important parts remain. Otherwise the confidence of the recommendation can exceed the strength of its basis.

Ask the people doing the work to demonstrate a recent example. Their explanation can expose informal steps that the official procedure omits. Avoid turning the observation into an exercise in catching someone out. The immediate purpose is to understand how the process actually operates. Accountability still matters, but the manager needs to distinguish a missing system condition from a person ignoring a clear and workable responsibility before choosing the corrective action.

How do you test the proposed cause?

Choose a bounded change that should affect the result if the explanation is correct. If incomplete order information is suspected, test a clear acceptance check for a defined group of orders. Observe whether dispatch delay changes and whether the check creates another burden elsewhere. If the result does not move, revisit the explanation rather than defending the intervention because time has already been invested in it.

Record what the test supports and what it leaves unresolved. A small improvement may show that one cause matters without proving that it explains the entire problem. A result can also be distorted by demand, staffing or an unusual event during the test. Keep the conclusion proportionate. The practical aim is to improve the next decision with the evidence available, then continue learning. Before approving a purchase, state the causal claim it depends on and the observation that would make you reconsider it.

Take one recurring complaint from your business. Rewrite it without blame, jargon or a proposed solution. Then list the evidence that could prove your favourite explanation wrong.