Most safety inspections run on checklists. And sometimes the form designed to make sure nothing gets missed ends up training inspectors to stop looking at anything that isn’t on it.
This is a Field Intelligence series, we’ll talk to Liam Hook, HSE Manager at Marubeni Middle East and Africa Power Limited, with seventeen years across power, oil and gas, construction, and aviation, who argues that the checklist culture most companies have built is one of the quieter problems in modern field safety.
🎧 Listen Liam tells the story in full, along with his case for treating inspectors as judgement-makers rather than form-fillers. Go to the full episode →
The Yes-or-No Trap
A safety inspector arrives on site with a checklist of forty or fifty items. They walk the area, ticking yes or no. At the end they hand in a completed form with a numerical score attached. The system assumes that everything important is on the list and that everything on the list is important. Both assumptions are wrong more often than they are right.
“Checklists are good because it reminds you of something to look for and to look at, but then you become very blinkered on: is all of this OK, yes or no? Well, that one thing is wrong, so no.”
— Liam Hook, HSE Manager, Marubeni Middle East and Africa Power Limited
The blinkered problem has two halves. The first half is that anything not on the list gets filtered out. A trained inspector walking a site without a checklist might notice a faint burning smell coming from a contractor’s plumbing rig and stop the work before a fire starts. The same inspector walking the same site with a forty-item checklist is mentally occupied with finding items twelve through fifteen and may walk straight past it.
The second half is what happens at the bottom of the form. A score of 49 out of 50 sounds like a failure. But if the one item marked “no” is a single untied ladder on a site with five hundred properly secured ones, the numerical output tells the client, the regulator, and the safety committee something technically true and substantively misleading.
Checklist Fatigue and the Paper Push
The second problem is volume. A permit-to-work for a single piece of work in oil and gas or power now routinely runs to a method statement, a risk assessment, a toolbox talk record, a start card, a job safety analysis, and a permit form — a pack thicker than most novels, before anyone has picked up a tool. Liam recalls a valve internals replacement where the method statement alone ran to 146 pages, none of which addressed the valves themselves because the document had been written for the construction of the equipment, not its maintenance.
“Safety teams are becoming more paper pushers, bureaucratic guys trying to cover their own behinds, the bosses’ behinds, the company’s behinds. It’s important that we’re doing paperwork, but I think sometimes it’s a bit over the top.”
— Liam Hook
The volume creates two effects. First, it crowds out the time inspectors and supervisors would otherwise spend looking at the work. A site walk that should take an hour takes a morning because three quarters of it is documentation review. Second — and worse — it creates the conditions for performance over substance. The pack gets completed because the pack has to be completed. Whether the work inside the pack is actually being thought about is a separate question the system stops asking once the boxes are ticked.
This is what most teams now call checklist fatigue. It is not just tiredness. It is the slow erosion of attention that happens when the apparatus around safety becomes the work itself instead of a record of the work.
Checklists as Reminder Points
The reframe Liam offers is small but useful. Stop calling them checklists. Call them reminder points.
The shift sounds cosmetic. It isn’t. A checklist tells the inspector what to do. A reminder point asks the inspector a question. Is the scaffolding tied off properly? on a checklist is answered yes or no. The same line as a reminder point becomes: have I actually looked at the scaffolding? Do I know what tied-off properly looks like for this configuration? The inspector who treats the form this way ends up learning the regulations from the inside, instead of treating the form as a substitute for understanding them.
This is also why Liam — who is not a checklist person on a site visit — still uses checklists for lockout-tagout. When you are about to put your name on a permit that says it is safe for someone else to touch an electrical earthing rod, you want the discipline of every box. The high-consequence, high-sequence tasks are where checklists earn their place.
The point is not that checklists are bad. It is that most teams have stopped distinguishing between the contexts where a checklist is the right tool and the contexts where it is the wrong one. The same form gets applied to both, and the form ends up degrading judgement in the places where judgement matters most.
The Pattern Problem
The most important thing checklists are not good at: they count, but they do not see patterns.
Liam describes a safety committee meeting at one asset that worked line by line through every observation logged that month — hundreds of them. Halfway through, his colleague leaned over and said quietly: that’s a scaffolding issue. That’s a scaffolding issue. That’s a scaffolding issue. What is going on with the scaffolding contractor? The safety team running the meeting had not seen it. They were processing observations one at a time, and the underlying pattern — that one contractor was producing most of the volume — had been invisible the whole time.
Three hundred observations is not three hundred problems. It might be twelve problems, with one contractor responsible for half of them. Until the data is read across observations rather than within them, the inspection programme is collecting evidence about a problem it is not equipped to diagnose.
Fluix’s Take: How to Make Checklists Work for You
The checklist is not the problem. The expectation that the checklist will do the thinking is. Across the field operations teams we work with, the pattern is consistent: forms grow longer every year, observation volumes climb, dashboards multiply — and the underlying judgement in the field gets thinner, not sharper. The work of inspection management is not collecting more data. It is making sure the data sharpens decisions rather than substituting for them.
In practice, the difference between a checklist that helps and one that hurts comes down to a small number of habits.
Write items as questions, not commands. Scaffolding tied off gets a yes-or-no. Have you walked around the scaffolding and confirmed it is tied off correctly for this configuration? gets thought. Same coverage, very different effect on the inspector’s attention.
Weight items by consequence, not by count. A single untied ladder and a single missing fire extinguisher do not deserve the same numerical weight as a misplaced cone. If the scoring system cannot tell the difference, the score is misleading whoever reads it.
Trim the form quarterly. A reminder point that has not flagged a real issue in six months is no longer earning its place. Forms grow naturally over time; they almost never shrink unless someone is given explicit permission to cut.
Reserve full checklists for high-consequence, high-sequence tasks. Lockout-tagout, confined space entry, permit-to-work, isolation procedures — these deserve every box and every signature. General site walks deserve a shorter prompt sheet and a more experienced eye.
Read observations together, not one at a time. If your monthly safety committee meeting is going through three hundred observations line by line, the meeting is the bottleneck. Group them by contractor, by location, by category. The patterns are almost always there once the system is set up to surface them.
Train inspectors as inspectors, not as form-fillers. The one who looks at something and thinks I need to check what good looks like for that becomes a better inspector over time. The one who thinks that’s not on my form becomes a worse one.
None of this requires throwing out the checklists. It requires treating them as what they actually are: a structured prompt for an experienced human, not a substitute for one.
