Your Checklist Is All Green and the Equipment Still Broke: 5 Reasons Why

A checklist that's fully "green" tells you nothing if the boxes were ticked from habit. On an ExxonMobil subsidiary's site, every stage of production had checklists in place — and equipment was still failing far more often than the paperwork suggested. Nothing looked wrong on paper; the wrongness was in how the lists were filled.

We rolled out logsheet.ai there and, in the process, catalogued five failure modes that turn a checklist into theater. Here's each one and what we changed.

1. Ticking boxes as a ritual

Operators weren't comparing readings to reality — they were ticking to finish. Separator pressure had sat at 4.2 atm for several shifts; the spec is 4.0–4.5, so the operator marks "normal" without looking. Scrolling a long list and checking every value against limits simply takes too long, so habit wins.

The result was a shift log with zero signal: equipment running in emergency mode, two unexpected breakdowns in the same area in one week, emergency repairs replacing scheduled maintenance, and no way to pin down when degradation started. At that point the equipment is unsafe to be around.

Fix: kill the checkbox. Replace it with an actual numeric value (temperature, pressure) or a drop-down. Operators pushed back for a couple of weeks, then adapted. Buy-in only held once we also cut the volume of what they had to enter — which leads directly to #2.

2. Information overload

Depending on the unit, a checklist ran 20–40+ items. In practice people read the first few and last few lines and defaulted everything in between. Separator pressure — a critical indicator — lived on line 12 and got auto-marked "normal" forever.

Management resisted trimming: dropping any parameter felt like a safety risk. So instead of blunt deletion we made the list conditional in logsheet.ai:

• show only parameters relevant to the current task, hide the rest;
• force critical items first, mark the rest optional (an abnormal pressure reading auto-reveals a comment field and a photo upload);
• adapt to shift and section so only what's needed is visible;
• surface extra items only when a trigger value fires — hidden under normal conditions.

Net effect: 10–15 items instead of 40, and operators spent the freed-up attention actually filling them in. Reports got sharper.

3. No feedback loop

Operators had no visibility into whether their entries mattered. The job felt like "complete the checklist," not "make the entry true." Management didn't monitor completion or attach any consequence, so honesty quietly decayed.

We piped completed data into reports and dashboards visible to operators and managers. On one dynamic-equipment sound check we cross-referenced checklists against breakdown records:

• not a single checklist flagged a problem;
• the checklist data didn't match the breakdown history — it was false;
• we lined up failure dates against checklist dates and the person who signed off right before each failure;
• that surfaced exactly who was filling lists without reading the equipment.

Feedback like this makes verification feel inevitable, so lists get filled honestly. For managers it doubles as a per-employee KPI on both quality and timeliness.

4. Filling the list from memory at end of shift

The "experienced, too-busy" operator reconstructs the whole checklist from memory at 6 PM instead of logging at the point of measurement. You get inaccuracies, and you can't reconstruct when something went wrong or who owned it.

Fix: fill it on a mobile device. The app records location and entry time, so late/off-site entries get flagged. One operator — John Henderson — was entering data from the office at 8 PM that should have been logged at 2 PM on the floor. That analysis became a new KPI for logging at the point of measurement, plus a scheduled prompt for entries at set times and voice-message input for speed.

5. The checklist isn't wired into shift handover

The lists lived outside the handover, so the incoming crew re-solved problems the outgoing crew had already cracked. On a well pad, day shift found the electric actuator on the methanol dosing valve knocking and not seating — fixed by turning the valve three extra turns because the limit switch was sticking in the cold. Night shift got a vague verbal handoff. When it recurred, someone opened the valve too far, methanol had to be shut off, and 3.5 hours went into re-diagnosing what was already known. The logbook would have made it a few minutes.

Fix: the incoming shift starts by filling a checklist auto-generated from the previous shift's entries, plus a "traffic light" that rates the prior list in real time on:

• where the person physically was when they filled it;
• the gap between the event and the entry;
• how fast it was completed — thorough vs. rushed.

What "operational" looked like

Automating a few processes turned the system into something worth trusting: lists capped at 10–15 items and split into sections; real-time completion via mobile and voice; KPIs on quality and timeliness; statistics used to analyze results; and checklists wired into handover. The checklist stopped being a wall of checkboxes and became a tool that produces an informative shift report, catches deviations early, and triggers corrective action before the breakdown.