Industry
Part of A practical reference for building operational checklists that hold up
5 things to weigh up on operational checklists mistakes
Operational checklist mistakes: unobservable items, missing stop rules, wrong list for the reader, growth without subtraction, and findings that go nowhere.
A checklist that is not working looks exactly like one that is. The boxes are ticked either way. That is what makes the failures below expensive: they are invisible from the paperwork, and only visible from watching the work.
The failures are grouped by where they start. Content and structure are in the operational checklists overview.
What to take away
- Ticking without checking is a design problem, not a discipline problem.
- The single most common defect is a list that has grown past the time available to run it.
- If nothing ever happens when a list records a problem, the list stops recording problems.
Mistakes in what goes on the list
Items that name a topic instead of a state. "Check the fridge" cannot be got wrong and cannot be got right. "Temperature within [range], reading written down" can.
Weak vs Strong Checklist Items
Weak item
- Topic vs state
- Check the fridge
- Two things in one box
- Inspect and record
- Unobservable
- Feel confident
- No stop rule
- Continue anyway
- Escalation
- Escalate to a person
Strong item
- Topic vs state
- Temperature within range
- Two things in one box
- Inspect; record separately
- Unobservable
- Reading written down
- No stop rule
- Written permission to halt
- Escalation
- Escalate to a role
Items containing "and". Two things in one box means one of them is optional and nobody knows which.
Items nobody can observe. Anything requiring the reader to remember, intend, or feel confident is not a checklist item. It is a hope with a box next to it.
A list built from a task inventory. Lists built from everything a job involves include the items that have never mattered, and those are what push the list past the time it has. Build from what has actually gone wrong, as set out in how to create operational checklists.
No stop rule. Without written permission to halt, someone facing an unconfirmed item will continue, because continuing is what everything else in the situation is pushing them toward.
Escalation to a person rather than a role. People leave. The list does not.
Mistakes in when the list is run
No named pause point. "During the process" is not a moment. A do-and-confirm list without a stated event gets run whenever, which is usually after the point where it could have helped.
Do-and-Confirm vs Read-and-Do
Do-and-confirm
- Best reader
- Expert
- Content
- Confirm states
- Risk
- Ignored by second run
- Example source
- Memory items
Read-and-do
- Best reader
- Beginner
- Content
- Full steps
- Risk
- Leaves out nothing
- Example source
- Read items
Running it after the decision it was meant to inform. The most common version: a pre-work check completed once the work has started, because that is when there is time.
The wrong kind for the reader. A do-and-confirm list handed to a beginner leaves out everything they need. A read-and-do list handed to an expert gets ignored by the second run. The distinction is standard where the stakes are high, and material such as the FAA's pilot's handbook of aeronautical knowledge treats memory items and read items as separate categories on purpose.
Silent ticking where two people are needed. The value of a two-person check comes from one person reading aloud and the other stating what they see. Both people ticking their own copy produces two independent assumptions.
Mistakes that show up over months
| Symptom | What actually happened | The fix |
|---|---|---|
| Lists complete in half the expected time | Ticked in a batch at the end | Cut the list to the real time budget, and add readings that must be written |
| One item always ticked, never checked | The item is unobservable, or nobody knows what it means | Rewrite it as a state, or remove it |
| Handwritten additions in the margin | The list is missing something people need | Add it, and thank whoever wrote it |
| Different people run it differently | Items are ambiguous, or roles are unnamed | Watch two people run it and rewrite what differs |
| Nobody can say which version is current | No version or date on the face | Version and effective date, legible from where it is read |
| Problems recorded, nothing changes | No response loop | Two visible responses will do more than any amount of encouragement |
The last row is the one that ends the practice. A list whose findings go nowhere teaches everyone that ticking is the deliverable.
Symptoms, Causes, Fixes
Symptom
- Half-time completion
- Batch ticking
- Always ticked
- Unobservable item
- Margin additions
- Missing item
- Different runs
- Ambiguous items
- Unknown version
- No date
- Nothing changes
- No response loop
Cause
- Half-time completion
- Cut to time budget
- Always ticked
- Rewrite as state
- Margin additions
- Add and thank
- Different runs
- Watch and rewrite
- Unknown version
- Version and date
- Nothing changes
- Two visible responses
Fix
- Half-time completion
- Always ticked
- Margin additions
- Different runs
- Unknown version
- Nothing changes
Mistakes in maintenance
Growth without subtraction. Every incident adds an item; nothing ever comes off. Within a year the list exceeds its time budget and the ticking becomes routine. Add and remove in the same sitting.
Maintenance Rules
- Add and remove in the same sitting
- Laminate only after a month unchanged
- Version readable from across the room
- Review on triggers, not calendar
- Decide retention before use
Laminating too early. Lamination makes correction impossible, so it belongs on a list that has run unchanged for a month, not on a new one.
Old copies surviving. Printed lists outlive their versions. A laminated card from two years ago looks exactly like a current one, which is why the version has to be readable from across the room, a point covered further in operational checklists format.
Reviewing on a calendar rather than a trigger. Annual reviews find nothing when nothing has moved and arrive late when something has. Triggers: an equipment change, an incident, a new person struggling, any handwritten addition appearing twice.
No decision about completed lists. How long they are kept, who reads them, and what happens when one shows a problem should be decided before the list goes into use. Deciding retention in advance rather than discovering it later is the approach the National Archives sets out for records scheduling, and a one-page list is not exempt from the reasoning.
The mistake of using the wrong document entirely
Three common mismatches:
Checklist, Procedure, or Form?
Does the reader know how to do the work?
Checklist of confirmations
Written procedure
A checklist used to teach a task. If the reader does not know how to do the work, they need a written procedure, in the shape described in SOP templates. A list of confirmations will walk them into an error confidently.
A procedure used as a checklist. Long, complete, and never run at the moment it was meant for. People extract the three lines they need and ignore the rest, which is a checklist they wrote themselves without the benefit of review.
A form used as a checklist. A form collects information after the fact. A checklist changes what happens next. A document that does both usually does neither, and the tell is a signature block with no stop rule.
