Introduction

Last week we focused on how design can make the safe way the easy way. This week we look at something that undermines even well-designed systems: the reality that procedures get bypassed, not by reckless workers, but by capable, experienced people under real production pressure. Understanding why this happens is the first step to fixing it.

Monday – The Gap Between Work as Imagined and Work as Done

This week's topic starts with a question that's worth sitting with: when the procedure says one thing and the work gets done another way, which one is real?

In safety research, there's a concept called the gap between 'work as imagined' and 'work as done.' Work as imagined is the procedure as written - the sequence of steps, the tools specified, the timing, the confirmations. Work as done is what happens on the floor, shaped by equipment that behaves differently than expected, time pressure, missing tools, ambient conditions the procedure didn't anticipate, and the accumulated experience of workers who have figured out what works.

In many facilities, these two things are significantly different, and everyone knows it. Workers know. Supervisors know. And in most cases, the 'work as done' version is getting the job done fine. Until the day it isn't.

The dangerous assumption is that the procedure is being followed because no one has said otherwise. Understanding the actual gap - not to punish the workers who've adapted, but to understand why the adaptation happened and whether it creates risk - is one of the most important things a safety team can do. You can't fix a gap you don't know exists.

Real-World Example

A food manufacturing facility in Georgia had a written sanitation procedure for a high-care processing line that specified 14 sequential steps, including a 10-minute dwell time for a sanitizing solution before rinsing. The procedure had been written by the quality team based on the chemical manufacturer's specifications.

A routine third-party audit found that workers were consistently reducing the dwell time to approximately four minutes. When auditors spoke with workers, they learned that the sanitation crew had discovered during a hot summer period two years earlier that the solution became significantly less effective if allowed to dry in place and in the warm facility, 10 minutes was enough time for drying to begin on certain surfaces. The four-minute dwell had been adopted informally because workers had found it produced better results than the 10-minute written version under actual conditions.

The workers weren't bypassing the procedure carelessly. They had adapted it based on real operational experience. But their adaptation had never been formally reviewed, documented, or validated by quality or safety. The facility updated the procedure to reflect temperature-adjusted dwell times and added a humidity check as a decision point. The quality director acknowledged: 'The workers had already solved the problem. We just hadn't been listening closely enough to know.'

Discussion Prompt

Is there a procedure in your area that you know gets done differently than it's written? Not to get anyone in trouble, but genuinely: what's the gap, and why does it exist? Is the adaptation safer, less safe, or just different? And who knows about it?

Tuesday – Normalization of Deviance - How 'Getting Away With It' Becomes the Standard

Yesterday we talked about the gap between how work is imagined and how it's actually done. Today we look at what happens when that gap persists without consequence, a process that researcher Diane Vaughan called 'normalization of deviance.'

Normalization of deviance describes what happens when an organization repeatedly accepts a deviation from a standard, because the deviation doesn't immediately produce a bad outcome, until the deviation stops feeling like a deviation at all. It becomes the new normal.

This isn't about reckless behavior. It's about how risk perception is shaped by experience. If a worker skips a step 50 times with no consequence, the subjective risk of skipping it begins to feel lower than the procedure implies. The procedure starts to seem overly cautious, written by people who didn't understand real conditions. And then, on attempt 51, the consequence that the procedure was written to prevent finally occurs.

Normalization of deviance is invisible from inside the system. It requires someone to look at current practice with fresh eyes and ask: compared to the standard, what are we actually accepting? That's a question SATs are uniquely positioned to ask.

Real-World Example

The most studied example of normalization of deviance in any field is the Space Shuttle Challenger disaster of 1986. Engineers had known that O-ring seals on the solid rocket boosters showed erosion in cold temperatures. They had flagged this concern multiple times. But each previous launch had succeeded, the O-rings had eroded but had held. Over time, decision-makers came to treat the erosion as an acceptable anomaly rather than an unresolved hazard. The standard for 'safe to launch' had shifted to accommodate a known deficiency because that deficiency hadn't yet caused a catastrophic failure.

On the morning of January 28, 1986, temperatures were significantly colder than any previous launch. The O-rings failed. All seven crew members were lost.

The lesson for manufacturing floors isn't about aerospace, it's about the mechanism. Getting away with something is not evidence that the something is safe. It may simply mean the consequence hasn't arrived yet. Every facility has its version of the erosion that's being treated as acceptable: the guard that's regularly bypassed, the lockout step that's skipped on quick maintenance tasks, the PPE that's not worn for 'short' exposures. The question every SAT should be asking is: what are we normalizing that we haven't labeled as normal yet?

Discussion Prompt

Can you think of something in your area that used to be flagged as a concern but has quietly become accepted because nothing bad has happened? What would it take for someone with fresh eyes to notice it? What would you tell them if they asked why, it's done that way?

Wednesday – Pressure, Shortcuts, and the Production Trap

We've looked at the gap between procedure and practice, and at how deviations get normalized over time. Today we look at the real-time force that drives most procedure bypasses in the moment: production pressure.

Production pressure is the felt urgency to keep output moving. It's real, it's legitimate, and it's present on every manufacturing floor. Machines need to run, orders need to ship, customers have expectations, and supervisors are tracking numbers. None of that is inherently wrong. The problem arises when the implicit or explicit message becomes production matters more than the procedure.

Workers are remarkably good at reading organizational priorities. When a supervisor rewards speed and asks questions about shortcuts only after an incident, workers learn what the organization values - regardless of what the safety poster says. When a worker stops a line for a safety concern and the primary response is frustration about downtime rather than appreciation for the catch, they learn. And they adjust their behavior accordingly.

The trap is that production and safety feel like they're in opposition. In a well-designed system, they're not. But it requires leadership to make that alignment explicit, through how they respond to stops, how they talk about near-misses, and whether the worker who slows down for safety feels valued or quietly penalized.

Real-World Example

A tier-two automotive parts supplier in Kentucky was investigating a recordable hand injury involving a press operator who had been manually clearing a jam without completing a lockout. The operator was experienced, trained on lockout procedures, and had completed lockout correctly hundreds of times before.

During the investigation, the EHS manager interviewed the operator, the team lead, and two coworkers. What emerged was a pattern that none of them had explicitly discussed before: on high-volume days, the team lead routinely walked the floor and verbally pushed for output. Workers had learned that stopping a press to complete a full lockout, which took approximately four minutes, during a jam on a high-volume day generated visible frustration from the team lead, even when no one said anything explicitly. Workers had started clearing jams 'quickly' to avoid that interaction.

The team lead was not a bad person or a reckless supervisor. He was responding to pressure from above him, and his behavior had been shaped by the same system. The corrective actions targeted the system, not the individual: a lockout completion rate metric was added to the shift performance dashboard alongside output metrics, and team leads were coached to explicitly acknowledge lockout completions during high-pressure periods. In the six months following, the team lead who had been the subject of the investigation became one of the most consistent advocates for the change on the floor.

Discussion Prompt

Have you ever felt pressure, spoken or unspoken, to skip a step or move faster than the procedure allows? What sent that message? Was it a person, a metric, a culture, or something else? What would it look like if this team made it genuinely clear that slowing down for safety is always the right call?

Thursday – Writing Procedures That Get Followed

We've talked about why procedures get bypassed. Today we look at what makes a procedure actually work, because not all the problem is culture and pressure. Some of it is the procedure itself.

Procedures that get bypassed share common traits: they're longer than they need to be. They include steps that workers know are irrelevant to their specific conditions. They use technical language that doesn't match how workers talk about the work. They were written by someone who doesn't regularly do the task. They haven't been updated since the equipment or process changed. And they've never been field-tested by the workers who are supposed to follow them.

A procedure that accurately captures how safe work is actually done - written with input from the people doing it, tested in real conditions, and updated when conditions change - is a fundamentally different document than one written from an office and posted on a wall.

The workers who know the most about what a procedure should say are the workers who do the job. Involving them in writing and reviewing procedures isn't just good for morale, it produces better procedures. Procedures that people trust because they recognize their own experience in them are procedures people follow.

Real-World Example

A specialty chemical manufacturer in Texas had a written procedure for a batch transfer operation that had not been updated in four years. Over that time, two of the three pumps used in the transfer had been replaced with different models, the control interface had been upgraded, and one of the intermediate valve positions described in the procedure no longer existed as a named position in the new system.

Workers had adapted by developing an informal verbal handoff, senior operators telling newer ones what the procedure 'actually means' in the current equipment configuration. The written procedure had become a document that nobody used as written, but that still existed as the official reference for training and compliance purposes.

After a near-miss in which a new operator followed the written procedure literally and created an unintended pressure condition by referencing a valve position that no longer existed, the facility launched a procedure verification project. Workers who performed each operation were paired with the EHS team to walk through every critical procedure step-by-step under actual conditions. Discrepancies were documented. Updated procedures were written by worker-EHS pairs and reviewed by engineering before being re-issued.

Of 23 critical procedures reviewed, 18 contained at least one step that no longer reflected actual conditions. The operations director said at the project close-out: 'We had been training people on procedures that were already wrong before they learned them. That's not a worker problem. That's on us.'

Discussion Prompt

Is there a procedure in your area that you think is outdated, inaccurate, or written by someone who hasn't actually done the job? What's different between what it says and what you do? And what would it take to get that fixed , who would need to be involved?

Friday – Building a Culture Where It's Safe to Stop

We've spent this week looking at why procedures get bypassed - the gap between imagined and actual work, normalization of deviance, production pressure, and poorly written procedures. Let's close by looking at the cultural foundation that all of this rests on: whether it's genuinely safe to stop.

In every facility, there are moments where a worker sees something that doesn't feel right, or reaches a step in a procedure that can't be completed as written, or is asked to do something they're not sure is safe. What happens in that moment, the actual response from the team and from leadership, determines whether that worker stops or continues.

If stopping results in appreciation, investigation, and resolution: workers learn to stop. If stopping results in frustration, pressure to continue, or the quiet sense that they've caused a problem: workers learn to continue and say nothing.

Building a culture where it's genuinely safe to stop is not primarily a policy question. It's a behavioral question — about what leaders do in the specific moments when a worker raises a concern. Every one of those moments is an opportunity to either reinforce or undermine the culture you say you want.

Real-World Example

A heavy industrial contractor in Alberta had a consistent problem: workers on high-consequence tasks - crane lifts, confined space entries, hot work near flammables - were sometimes proceeding despite having concerns they hadn't fully resolved. Post-incident investigations kept surfacing the same pattern: workers had sensed something wasn't right but hadn't felt confident enough to stop the work.

The company introduced a formal Stop Work Authority protocol, but they didn't just add a policy. They trained every supervisor on what to do in the first 90 seconds after a worker exercised it. The training was specific: thank the worker by name, ask what they observed, treat the stop as a success rather than a disruption, and debrief the team on what was found before restarting.

To reinforce the behavior, every exercised Stop Work Authority was logged and reviewed at the monthly safety meeting; not as an incident, but as a success. Workers who exercised the authority were recognized publicly. In the first year, 34 stop-work events were logged. In 11 of those cases, the formal post-stop review identified a genuine hazard that would have been encountered if work had continued.

The safety director presented this data to the executive team: 'Thirty-four stops. Eleven real hazards caught. Zero incidents from any of the 34. This program is working, and every supervisor who responded correctly made it work.' Stop Work Authority rates increased each subsequent year. Worker surveys showed a steady increase in the belief that raising a safety concern would be met with appreciation rather than frustration.

Discussion Prompt

Weekly Wrap-Up: Is there a situation on this floor right now where a worker - maybe you, maybe someone else - has a concern they haven't raised because they're not sure how it will be received? What would need to be different for that concern to surface? And what can each of us do - as team members, not just as leadership - to make this the kind of place where stopping is celebrated?

DOWNLOAD PDF VERSION 

Tags: safety topic , Safety Brief , bypassed procedures ,