A lot of incident investigations end at the same sentence: "worker did not follow procedure." Box checked. Corrective action: retrain the worker. File closed.
That sentence is almost never wrong. It is just almost never the answer.
The Report Stops Where the Real Work Starts
Here's what most incident investigations actually are: a description of what happened, followed by a guess at who is responsible for it happening. That is not investigation. That is narration with a name attached.
A real investigation asks a harder question than "what did the worker do." It asks: what had to be true about this site, this schedule, this equipment, and this supervision for that worker's decision to make sense in the moment? Workers rarely wake up planning to take a shortcut that hurts them. They take the shortcut that everyone around them has been quietly taking for months, the one nobody flagged because the job kept getting done.
Stop the investigation at "worker error" and you fix nothing. You retrain someone into repeating the exact same decision, because the conditions that produced it are still sitting there, unexamined.
Three Places Investigations Almost Never Look
The gap between the written procedure and the site procedure. Every company has a procedure on paper. Fewer companies know what workers are actually doing when nobody is watching. If your investigation does not ask "was this drift normal here," you are only auditing the document, not the job.
What supervision actually looked like that day. Not whether a supervisor was assigned on paper, but whether they were physically present, whether they had time to supervise this task specifically, and whether they had the authority to stop the work if something looked wrong. A supervisor stretched across three tasks at once is not providing the supervision your org chart says exists.
What got normalized over time. Most serious incidents are not the first time something unsafe happened. They are the first time it went wrong. If your investigation does not ask "how long has this been happening," you are treating a pattern like a one-off, and the pattern will produce another incident.
The Test We Use
Here's a test to run on every investigation before you sign off on it: if you hand the corrective action to a different worker doing the same job next week, does it actually change what they experience on site? "Retrain the worker" fails this test almost every time, because the new worker walks into the exact same schedule pressure, the exact same normalized shortcut, the exact same thin supervision. Nothing about their day changed. You just replaced the name on the file.
A corrective action that passes the test looks different. It changes the schedule that made the shortcut attractive. It changes what the supervisor is responsible for covering in a shift. It changes the site norm that let the drift go unflagged for months. That is harder to write down, and it is the only version that actually prevents a repeat.
Why This Gets Skipped
Not because investigators don't care. Most genuinely want to get it right. It gets skipped because going one level deeper means asking questions that point at scheduling decisions, staffing decisions, and supervision decisions, not just the worker on the ground. That is an uncomfortable direction for an investigation to go, especially when the person reviewing the report is the one who made those decisions.
We want to be direct about something. A company that only ever finds worker error is not running clean sites. It is running an investigation process that stops exactly where it becomes inconvenient.
What to Do Monday Morning
Pull your last three closed incident reports. Read only the corrective actions. If every single one says some version of "retrain," "remind," or "reinforce," your investigation process is stopping at the worker every time, and you have a blind spot, not a clean record.
Then pick one of those three and ask it the harder question: what had to be true about this site for that decision to make sense? Follow that question as far back as it goes. That is where your next real fix is sitting, and it has been sitting there the whole time.
WorkSafe Sounds helps Ontario contractors build investigations that find the actual cause, not just the closest name. If your corrective actions keep saying "retrain the worker," that is worth a second look before your next incident, not after.