The cause is unclear. What should you check first?

When the cause is unclear, separate the bounded observation, available records, and possible explanations. An issue tree helps organize several investigation paths. For a specific event with a sequence of records, 5 Whys helps ask successive questions. At each step, check which records the explanation covers and which remain unexplained.

This fictional order case moves from 20 late orders to eight stock-confirmation orders, then to six with further records. Keep the other two open. Explore the evidence path beside the explanation, then paste your own analysis, select one explanation to investigate, and retain the known facts, hypotheses, unknowns, and next check.

What a classic example teaches

The Lean Enterprise Institute presents Taiichi Ohno's example of a stopped machine. The investigation follows a blown fuse through lubrication problems and pump wear to metal scraps entering a pump without a strainer. Replacing a failed part would leave the conditions behind the failure unresolved. The source also explains that five is not a mandatory number of questions. Read the example.

That teaching example presents a completed chain. Your investigation may begin with a symptom and several untested explanations. Five plausible sentences from a meeting are not equivalent to five established causal links.

Use the same evidence for both methods

Consider this fictional practice case. A small store promised to dispatch 100 orders during a week. Twenty missed their promised dispatch date. Each late order is assigned to its first recorded blocker.

This issue tree separates four investigation paths. The counts add up to twenty because the classification assigns each order once. That does not mean each order had only one real-world cause.

Stock confirmation affects the most orders, which makes it a reasonable starting point. Volume is only one consideration. A small group of highly time-sensitive orders might deserve attention first.

Open diagram at full size

Keep the scope of each answer visible

Investigating the stock branch produces this record:

The second answer narrows the investigation from eight orders to six. Keep the other two on the open-questions list. Applying the six-order explanation to all eight would overstate what the evidence supports.

Now compare update timestamps, reserved stock and warehouse counts. If both hypotheses contribute, branch at that point. A 5 Whys investigation does not require every problem to have a single linear cause.

At this stage, “the system did not synchronize” is premature as a final conclusion. A more useful next step identifies which records to inspect, what result would support each explanation, and who will perform the check.

Choose a starting point from your evidence

An issue tree can also organize possible solutions. Keep a tree asking why something happened separate from one asking how to change it. Switching questions halfway down a branch makes the reasoning difficult to follow.

Fictional order investigation; not a business finding

20 late orders: branch out

First recorded blockers: stock 8, address 5, cutoff 4, pickup 3; total 20.

Order and process records support this classification; each order is assigned once.

A first blocker is not the only real-world cause. Volume does not establish impact or priority.

Inspect records, impact, and deadlines before choosing a branch; retain the other branches.

Scope: 100 promised, 20 late; stock 8, address 5, cutoff 4, pickup 3. Further records cover 6 of the stock 8; 2 remain unexplained.

Personal investigation; supplied material has not been verified by this site

Full original material:

Problem scope and period: [what happened, to whom, and record location]
Investigation branches and counting rule: [branches and unexplained records]
Current explanation: [select one explanation to investigate]
Sample scope: [which records it covers]
Supporting material and status: [records available or still needed]
Contrary observations: [what would change my explanation]
Next check: [records to inspect and how to interpret them]
Owner and review date: [actual agreement or unconfirmed]
Open questions: [retain unexplained records; a hypothesis is not a root cause]

Explanation selected for this check:

Not selected

Evidence status marked by you:

Supporting records not yet identified

Review items checked for this version (not factual verification): 0/2

Retain hypotheses, unknowns, and evidence checks. This record does not establish a root cause.

Turn the analysis into your next investigation

Write down a bounded problem first: what happened, to whom, during which period, and how you know. Sketch the main branches and inspect one with enough evidence to justify further work. At every step, record the explanation, its evidence and the remaining uncertainty.

Use the logic tree guide when several explanations need organizing. If you already have a specific event and a sequence of records, go directly to the 5 Whys guide. You do not need to draw a tree simply to use both methods.

For each of this example's final hypotheses—update timing and stock reservations—record which evidence to inspect and what would support or contradict it. Add an owner and review date, or mark them as unconfirmed. Keep the two unexplained orders separate. The next output is a check result, rather than a combined label such as “synchronization failure.”

You can practice with the order counts and evidence record. Before extending the diagram, check that the counts reconcile, the scope stays consistent, and the final question leads to an observable check. Those checks make the analysis useful even when a root cause has not yet been established.

For AI-assisted organization, provide the problem scope, original records, current explanations, and unknowns. Ask it to flag connections lacking evidence. This exercise retains scope and investigation status; a complete causal chain or checked review item cannot replace actual evidence collection.