Free tool · 5–10 minutes
What to check before you accept “human error” as a root cause. Five questions, six areas to brainstorm, and a verification table that turns potential causes into confirmed ones.
The rule
A person’s mistake is where an investigation starts. It’s rare that the person’s behavior was the root cause.
You need to check the system around the person — and document what you found.
Most events have more than one contributing cause. You want to find the weaknesses that allowed the error, and the gaps that let it through undetected. Only then can you identify root causes.
In a hurry
Before the checklist
The six areas only work if you’re asking these first.
If the problem statement already assumes a cause — “operator failed to follow procedure, resulting in defective components” — the checks will just confirm the assumption instead of testing it. Fix the problem statement first.
“The technician was probably rushing” is an assumption. “The technician didn’t log into the system and documented results outside of the system” is a fact. Only facts belong in the evidence you collect for your investigation.
A plausible explanation is not a confirmed one. Each of the six areas suggests places to look into for evidence. Having no evidence to back up your theory means you’re still guessing.
As part of your investigation, check whether the same equipment, procedure, or material shows up on another line, another product, or another shift. Otherwise you might overlook a bigger risk.
A labeling error identified before product release may not need the same depth as an issue that reached the patient. Match your investigation scope and how many evidence layers you’ll review to the risk level, not to how easy or hard it is to retrieve them.
Brainstorm
These six areas come from the Ishikawa (fishbone) diagram: Man, Machine, Method, Material, Measurement, Environment. Here we use them as a brainstorming aid: six places to look before the investigation stops at “human error”.
Go through each area and tick every check that could be part of your event. Write anything else in the box under it. The verifying comes after the six areas. Everything you type stays in your browser — nothing is sent anywhere.
Area 1 of 6
The record says: “operator error”, “technician mistake”, “did not follow procedure”.
Evidence to pull
Area 2 of 6
The event involves production equipment, tooling, fixtures, or software.
Evidence to pull
Area 3 of 6
The procedure, work instruction, batch record, or process flow could be part of the event.
Evidence to pull
Area 4 of 6
The event could be connected to raw materials, components, packaging, labels, or consumables.
Evidence to pull
Area 5 of 6
The event depends on a test, an inspection, or recorded data.
Evidence to pull
Area 6 of 6
The task depends on working conditions.
Evidence to pull
Verify
Every check you ticked and every line you wrote is a row here. Give each one an action that will verify or disprove it: which record you’ll pull, what you’ll compare, whom you’ll ask. A cause without a verification action is still a guess.
Example: a wrong-label nonconformance, three potential causes into verification. The scenario is invented; the mechanic is the tool.
Now yours. Everything you type stays in your browser — nothing is sent anywhere.
Nothing here yet. Tick the checks that could be part of your event in the six areas, or add a row.
The honest part
Sometimes every system cause you test comes back ruled out: current procedure, verified competence, capable equipment, calibrated measurement, suitable material, controlled conditions.
Then an individual lapse can be your documented conclusion. Write it with your verification table attached: what you checked, what you found, why the system is judged capable. An auditor can accept that.
What they won’t accept is “human error” with nothing behind it.
If you’ve done everything right, you’ve narrowed your potential causes down to the most relevant ones. What’s left is picking the ones that make the most sense to tackle, and then choosing actions that are not “retraining”.
Two patterns in the toolkit’s weak-pattern library deal with this: “human error” as the root cause and retraining as the default action, each with a weak and an improved example. Subscribers get:
Delivered by email, along with the MedTech CAPA Journal: frameworks for NCs and CAPAs that solve the problem and hold up in an audit. No fluff. Unsubscribe anytime.
Your team’s CAPAs keep ending in “human error” and retraining? See the in-house training