Where Your Safety Program Counts on Nobody Making a Mistake

An administrative control is any protection that works because a person does the right thing: a procedure, a permit, a checklist, a training requirement, a supervisor’s attention. Every workplace runs on them, and no workplace could function without them.

They’re also the weakest thing that still counts as a control, and they tend to fail as a group, because they all depend on the same thing.

Three cases from this fall make the point. At Shell Polymers Monaca, eleven administrative controls guarded a hazard the company had already identified as potentially fatal, and an engineer who’d never performed the task went through all eleven in one action (Two Valves, Six Minutes, $95 Million). At ModWash, energy control existed as an expectation because nobody had written the procedures down. At Araujo Construction, fall protection came down to whether a foreman decided to rig it that morning (both in OSHA’s FY2026 Top 10 Looks Exactly Like Last Year’s).

Even OSHA’s new overdose guidance is a version of the same problem. “Somebody will call 911” only works if somebody recognizes what they’re looking at.

Here’s how to find yours.

Inventory the places where one person is the last line

Walk your highest-consequence jobs and ask, at each step, what stops the event if this person gets it wrong. If the answer is “someone else notices,” you’ve got a chain of administrative controls rather than layers of protection. Write down every task where the final safeguard is somebody’s judgment in the moment.

Ask whether an engineered control is already sitting there

This is the sharpest lesson from Monaca. Shell owned a panel that could have blocked the gas and hadn’t configured it. Interlocks, double-block-and-bleed arrangements, key-transfer systems, guarding, physical isolation: these often exist on equipment you already own. Ask your maintenance people and your vendors what protective functions are available but switched off. The answers surprise people, and they’re sometimes free.

Look at your interfaces

Three valve numbers that differ by one digit is a design problem, not an operator problem. Go through your control screens, labels, tags, and equipment IDs looking for things that are distinguishable in theory but not at a glance under pressure. Where you can’t fix the interface, add a confirmation step or a second set of eyes on that specific action.

Do the inspections you already owe

Five of the ten most-cited standards specify their own recurring inspection, and these are requirements, not suggestions:

• Lockout/tagout: periodic inspection of energy control procedures, at least annually
• Scaffolding: inspection by a competent person before each shift
• Ladders: periodic inspection for defects
• Respiratory protection: user inspection before each use
• Powered industrial trucks: examination before each shift or after each use

If you can’t produce those records, that’s not a hypothetical gap. It’s a citation waiting for an inspector.

Then go watch the work

Reading a procedure tells you what’s supposed to happen. Watching the job tells you what does. Pick your three highest-consequence tasks and go see one of each this quarter, at normal working pace, without billing it as an audit. Whatever distance you find between the written control and the actual practice is the real measure of how protected you are.

None of this is an argument against procedures. It’s an argument for knowing which of your hazards are held back by paper alone and treating that list as work to be done rather than a description of how things are.

Feeling overwhelmed with OSHA’s expectations of site-specific plans and programs? NASP is available for comprehensive plan and program audits as well as mock OSHA inspections. Click here to find out more.