How do you tell that a checklist prevents mistakes instead of only creating a sense of control?

This is a painful question, because it is easy to buy a sense of order — and very expensive to pay for it later.

A bad checklist gives false confidence: boxes are checked — so everything was verified.

A good checklist does something else: it helps you run a procedure in a way that forgets fewer important things and does not lose state.

The scene that reveals it

The checklist is checked.
The problem still happens.
A postmortem starts.
And you discover that half the items were “verified/ready”, with no definition of what was verified.

That is the sense of control.

Three signs a checklist is real

1) Sometimes it can be not passed

If a checklist never leads to stop/escalate/postpone — it’s an alarm bell.
Reality always gives reasons to say “we are not ready”; you can either see them or cover them with a tick.

2) There are verifiable gates

Items like “ready” don’t work. Items like these work:

“how do we know it’s ready”
“what signal do we verify against”
“what fact do we record”

There don’t have to be many — but a few critical gates must exist.

3) The checklist changes after misses

If after a miss the checklist doesn’t change, it doesn’t accumulate experience. It turns into a ritual.

In a working version, something simple happens:

“here is where we got confused” → fix 1 item;
“here is what we forgot” → add 1 gate;
“here is what turned out to be useless” → delete 1 item.

A quick test (5 minutes)

Take 10 items and honestly mark:

how many have a verification criterion?
how many can truly lead to “stop/not ready”?

If almost none — you don’t have control, you have a habit.

Honest limits

For low‑stakes processes, a “sense of order” may be OK.

But if stakes are high (incident, launch, safety, access), the checklist must be about running and verifiable gates — otherwise it starts to cause harm.