Investigating Beyond “That’s How We Do It”

The Problem

The investigation starts before it officially begins.

A senior tech walks past the failed pump, glances at the shaft, and says “Oh, I know what this is — happens every time the humidity drops.” The room nods. Someone opens the RCA template and starts filling in the cause chain. The logic tree gets built around the conclusion, not toward it.

This is tribal knowledge in action. It isn’t malicious and it isn’t lazy. It’s what happens when unwritten, undocumented experience fills the space that objective evidence should occupy. The veteran’s pattern recognition is real and often valuable. The problem is that it goes untested, and the investigation quietly becomes a formality.

When the program relies on gut feel instead of data, solutions fix symptoms. The underlying cause — shaped by decade-old assumptions that no longer match current technology, personnel, or process variables — stays untouched.

Getting to the Root Cause

The reliance on tribal knowledge persists because the program never created a reason to challenge it.

When an RCA process doesn’t mandate physical evidence, sensor logs, or maintenance records before the team starts forming hypotheses, the vacuum gets filled by whoever speaks first and speaks with the most confidence. That is not a people failure. It is a program design failure.

Two conditions keep this pattern locked in:

1. No evidence standard at the start. Most RCA programs begin with a meeting. Someone asks what happened, someone answers, and the investigation follows that answer. Without a requirement to anchor the conversation in hard data before hypotheses are floated, the first credible narrative wins.
2. Expert bias goes unchallenged. In many plants, questioning a veteran’s read on a failure is socially costly. It reads as disrespect rather than rigor. So nobody asks “how do we know that?” and the assumption travels through the logic tree unchallenged until a recurrence forces the question.

The organization hasn’t built a culture where “show me the data” is the default. So the investigation defaults to the easiest narrative rather than the most accurate one.

Corrective Action (You Can Do This Week)

You don’t need to sideline your veterans. You need two structural rules that make evidence the starting point for everyone.

• The Evidence First Rule. Before the first hypothesis gets floated, require the team to put three pieces of hard data on the table. Sensor logs, failure photos, maintenance records, condition monitoring trends — anything physical and documented. This is not a bureaucratic step. It anchors the conversation in reality before opinions can take over and makes it much harder for any single narrative to dominate unchallenged.
• The Naive Questioner Role. Assign one person in every RCA meeting to play the outsider. Their only job is to ask “how do we know that for a fact?” whenever an assumption gets stated as a conclusion. This role is not adversarial. It is structural protection against expert bias, and it makes challenging assumptions a function of the process rather than a personal confrontation.

Run both on your next investigation and watch how many early conclusions survive contact with the evidence.

What Do You Think?

Have you ever watched an investigation get derailed because a subject matter expert insisted they knew the answer before the data was in? How does your team balance respecting experience while keeping the analysis honest?

Help If You Need It

If your investigations consistently default to folklore over facts, the fix is structural, not cultural.

Root cause analysis training builds the evidence-gathering discipline and facilitation skills your team needs to hold the line when experience tries to shortcut the process. And RCA software like EasyRCA creates a documented trail that ties every conclusion to verified data, so gut feel never quietly becomes the official cause of record.