Where the Blaming Stops

Where the Blaming Stops (And Why Your Post-Mortems Keep Missing the Real Fix)

A shipment goes out three days late. The client is furious. Somebody calls a post-mortem.

Twenty minutes in, the room has landed on an answer: Sarah in fulfilment missed the cutoff. She's new, she was covering for someone, she should have flagged it sooner. Case closed, action item assigned, meeting adjourned.

Except Sarah missed the cutoff because the roster had her covering two roles with no handover notes. The roster looked like that because scheduling runs off a spreadsheet nobody's owned since the last reorg. The spreadsheet is like that because nobody's had time to fix it, because everyone's been firefighting shipments that go out late.

You could keep walking that chain backwards for a very long time. Every cause has a cause. That part isn't controversial because it's just how causation works, all the way down to conditions nobody in the room chose or controlled. But nobody walks the chain that far. Somewhere between "Sarah" and "the entire operating model," the room stops. And the stopping has nothing to do with where the truth actually ran out.

Blame isn't an explanation. It's a stopping rule.

Causes recede indefinitely. Blame does not. Blame picks a node in the chain and declares, arbitrarily, "this is where the story ends", and it picks that node based on what the story needs to do, not on what's actually true.

Research on how people trace responsibility through spreading events (Kardos and Lickel, 2026) found that most of us stop tracing origin at roughly the second step back, then treat that as the beginning. Not the true cause - just deep enough to feel resolved. And when a small initial failure snowballs into a bigger one, we load extra blame onto whoever started it, well past their actual share of the outcome. That's post-mortem culture in a sentence: punish the first name on the timeline, not the structure that let one missed cutoff become a client crisis.

Why "Sarah" is always the answer

Three forces reliably pull a room toward a person instead of a system:

Findability. The human closest to the failure is always locatable. The scheduling spreadsheet is not going to sit in the meeting and take questions. Charles Perrow, writing about industrial accidents, noted that once an investigation reaches "the operator failed," it very often just stops there, not because that's the deepest cause, but because it's the available one.

Control. We reserve real blame for someone who could have known and could have done otherwise. That's not a bias, it's a reasonable instinct, but it means every distal cause (understaffing, a broken process, a decision made two reorgs ago) feels less blameworthy than the person standing in front of you, even when it's doing more of the actual causal work.

Actionability. This is the one that costs you the most money. Rooms stop at the cause they can do something about today. Retraining Sarah is a task you can assign before lunch. Rebuilding the scheduling system is a project nobody wants to own. So the fix that gets assigned is almost never the fix that would actually prevent the next failure. It's just the fix that's easiest to close out.

The test I use on myself

One question to ask before you let the room move on: if we kept walking one more step back, would the story fall apart, or would we find an actual lever?

If going one step further would implicate the process, the incentives, or a decision made above Sarah's pay grade, and the room quietly doesn't want to go there, that's not the true cause. That's just the exit everyone found comfortable. If one more step back gets you to something you could genuinely redesign (a role, a handover process, a piece of software nobody owns) that's usually where the real fix is hiding, one link past where the meeting wanted to stop.

The forgiveness problem

There's a reason "blame the last person who touched it" is so hard to kill as a habit, and it isn't laziness. You can't discipline, retrain, or forgive a broken process. You can only do those things to a person. A system doesn't have a face you can hold accountable and then move past. So organisations keep reaching for a name, even when the name isn't the answer, because a name is the only kind of stopping point that lets everyone actually resolve the thing and get back to work.

Which is fine, as far as it goes — accountability matters, and pretending nobody is ever responsible for anything is also dishonest. The mistake isn't holding Sarah accountable. It's mistaking that for having fixed the problem.

Next post-mortem, get your name. Then ask one more question before you close the file, it's usually where the real fix is hiding.