NEW / EVIDENCE LAB_
Ready for a longer investigation?
These self-checks are quick prompts. In the Evidence Lab, follow a complete simulated investigation: examine records, compare alternatives and reveal the consequences of your choices.
Open the Evidence Lab →
THE SELF-CHECKS_
Choose the next defensible step.
CHALLENGE 01 / REASONING
An incident is restored, but the same failure has happened six times this quarter. What should happen next?
The operations lead says the most recent engineer caused it. What evidence would distinguish a human action from a fragile underlying process?
Reveal worked explanation
Suggested reasoning: Open a separate problem investigation while maintaining incident recovery. Frame recurrence, gather comparable event records and test systemic causes. A person's last action is not a sufficient cause statement.
CHALLENGE 02 / REASONING
A Pareto chart shows 45% of reported defects in one category. Has the root cause been found?
What does a category distribution tell us—and what does it leave unanswered?
Reveal worked explanation
Suggested reasoning: No. The largest category is a prioritization signal. Evaluate comparable severity and exposure, then investigate that category using tests such as Fishbone and deeper causal analysis.
CHALLENGE 03 / REASONING
A Fishbone workshop produces 18 branches. Which branch is the root cause?
Could a team vote be enough to settle the question?
Reveal worked explanation
Suggested reasoning: None is proven by the diagram. Identify discriminatory evidence, test leading hypotheses, and keep weak branches recorded as possibilities rather than facts.
CHALLENGE 04 / REASONING
The team reaches the fifth Why but has no evidence for steps three and four. Is the chain complete?
What would make those two transitions defensible?
Reveal worked explanation
Suggested reasoning: No. Five is not a target. Check the causal links using observations, logs or controlled tests, and stop when further explanation becomes unsupported speculation.
CHALLENGE 05 / REASONING
An outage affects one region after a change; others run normally. What analysis is useful?
Identify the IS and IS NOT distinctions and what changed.
Reveal worked explanation
Suggested reasoning: A Kepner–Tregoe Problem Analysis comparison can clarify what is unique to the affected region and narrow tests. A change correlation is not proof of the cause.
CHALLENGE 06 / REASONING
A proposed migration meets most preferences but violates a legally required data-residency condition. Can it win with a high score?
Should weighted wants be allowed to compensate for non-negotiable constraints?
Reveal worked explanation
Suggested reasoning: No. Apply mandatory MUST conditions first. An alternative that fails an essential requirement is excluded regardless of its weighted WANT score.
CHALLENGE 07 / REASONING
The workaround stops failures for two days, yet the issue usually recurs weekly. Can the problem be closed as resolved?
Consider the baseline frequency and observation window.
Reveal worked explanation
Suggested reasoning: Not on this evidence alone. Keep containment distinct from permanent correction, and observe at least a representative recurrence cycle with agreed outcome measures.
CHALLENGE 08 / REASONING
A proposed cause fits every observed fact, but no one has tried to disprove it. What is the next investigation step?
Find an observation that would be unlikely if the proposed cause were true.
Reveal worked explanation
Suggested reasoning: Design a falsifiable test, examine contradictory cases and compare alternative explanations. Confirmation alone risks reinforcing a favored story.
REFLECTION_Which answer changed when you asked what would contradict your favorite explanation? Which proposed fix still needs a baseline and verification period? Those questions are often more useful than remembering the name of a diagram.
Keep learningReturn to the cases, or learn the terms for your next investigation.
Open glossary →