Verifying the Root Cause

After using Five Whys to identify what you think caused a problem, you need to verify it actually produces the issue. Instead of assuming your explanation is correct, test it by removing or changing the supposed cause and watching what happens. For example, if you think a specific ingredient is giving you headaches, try eating the same meal without that ingredient to see if the headache still occurs. Check this: The next time you identify a root cause, ask what would happen if you were wrong.

Verifying a root cause means treating your explanation like a scientific hypothesis that must be tested before being accepted. The process follows a simple pattern: first, identify a suspected cause (perhaps using the Five Whys technique). Then, predict what should happen if this cause is truly responsible. Finally, change or remove the cause and observe whether the problem disappears as expected.

In healthcare, a doctor might suspect a patient's rash is caused by a new medication. Before stopping treatment entirely, they might have the patient pause the medication for a few days while monitoring for improvement. In manufacturing, engineers testing whether a machine vibration causes defects might temporarily reduce the machine's speed and check if defect rates drop accordingly. In software, a developer might reproduce a bug by undoing their recent code changes rather than assuming those changes are innocent.

This verification step protects us from confirmation bias—the tendency to notice evidence that supports our beliefs while ignoring what contradicts them. When we skip verification, we risk investing time and resources into solutions that don't address the real problem. The test itself can be as simple as an if-then prediction: "If this cause is real, then making this specific change should produce this specific result." If the predicted result doesn't occur, it's time to reconsider your root cause analysis.

Verification transforms root cause analysis from speculation into evidence-based problem-solving. Research on causal reasoning shows that people systematically overestimate the quality of their own causal judgments and fail to seek disconfirming evidence. Studies of confirmation bias demonstrate that people naturally seek information that confirms their hypotheses while neglecting tests that could prove them wrong—a tendency that spans across domains from medical diagnosis to forensic analysis.

• Nickerson (1998): Confirmation bias operates as a "ubiquitous phenomenon in many guises," leading people to seek, interpret, and remember information in ways that confirm pre-existing beliefs [1].
• Popper (2002): The principle of falsification holds that scientific theories cannot be verified through confirming observations alone; instead, they must be subjected to rigorous attempts at refutation, with theories that withstand such tests achieving "corroboration" rather than confirmation [2].
• Klein (2007): The premortem technique, which asks teams to imagine that a plan has failed and work backward to identify causes, increases the ability to identify genuine threats by 30% compared to traditional planning methods [3].
• IHI (2015): Root Cause Analysis and Actions (RCA2) framework emphasizes that verification is not optional—it is essential for distinguishing genuine root causes from symptoms and avoiding wasted implementation efforts [4].

Verification isn't always practical or ethical. In medicine, you cannot withdraw a life-saving treatment to test whether it's causing side effects. In complex systems, multiple causes may interact, making it difficult to isolate any single factor. Some problems involve rare events that are difficult to reproduce on demand. Additionally, verification tests themselves can be flawed—your observation might be wrong, or the test might not isolate the cause cleanly. Time constraints and limited resources sometimes force action before perfect verification is possible. Finally, some organizations create a culture where admitting error is punished, making people hesitant to verify their initial conclusions and potentially discover they were wrong.