High-reliability organization principles help public safety agencies catch dangerous patterns before they cause harm. This research, developed by studying aircraft carriers and air traffic control, explains why incident data scattered across disconnected systems lets those patterns go unnoticed until it’s too late. Agencies that ask where they’re vulnerable, treat errors as information rather than blame, and connect their data can improve continuously instead of only reacting after something goes wrong.

Consider this scenario from a police department in Anytown, USA. Sgt. Reyes raps almost apologetically on her lieutenant’s door one late afternoon. “I’ve noticed something that concerns me,” she says. He motions her inside.

She explains how she’d started digging after a rookie got hurt chasing a crook in a stolen car through a residential block. What Sgt. Reyes discovered took her three weeks to piece together because the data was stored all over the place: a pursuit in the CAD log, a use-of-force report in a different system, body camera footage nobody reviewed unless something obvious went wrong.

Once she laid it all out, the pattern seemed impossible to miss. Pursuits that should have been terminated weren’t. A single officer had been involved in a dozen of them, and six of the suspects arrested in the incidents had been hurt. “I’m not trying to cause problems,” she told her lieutenant. “I’m just really worried someone is going to get seriously hurt or killed.”

Her lieutenant’s first question was fair, and it’s the question every leader should be asking right now: How did we not know this was happening?

When a Rule Quietly Stops Mattering

The honest answer in most agencies is that nobody made a conscious decision to tolerate the problem Sgt. Reyes uncovered. Nobody rewrote the agency’s policy to allow officers to chase every fleeing vehicle regardless of the danger. What happened instead is something safety researchers call drift (or normalization of deviance): a slow, incremental slide in how work actually gets done, driven less by bad intent than by the absence of any negative consequence for ignoring the standard. A rule against unnecessary pursuits gets bent once, then again, and when nothing bad happens, the bend becomes the new normal.

Diane Vaughan documented this mechanism in her study of the Challenger disaster, where engineers gradually redefined a known and unresolved problem as an acceptable risk, one small rationalization at a time. The result, of course, was the loss of seven lives and a $2.8 billion spacecraft.

Drift is very good at hiding in agencies with records siloed in multiple systems that don’t talk to one another. And that’s precisely the condition found in a National Policing Institute review of 19 Department of Justice pattern-or-practice investigations: Insufficient or inappropriate policy was a contributing factor in 94.7% of those investigations, inadequate accountability mechanisms in 78.9%, and inadequate training in 73.7%.

Those aren’t 19 agencies that stopped caring. Rather, they’re 19 agencies where, over time, practice started to diverge from policy. And nobody had a reliable way to see the pattern of drift until it was serious enough to prompt a federal civil rights investigation.

“Every agency I’ve worked with wants to do right by its people and its community. Very few of them fail because their personnel stopped caring.”

What High-Reliability Organizations Get Right

There’s a name for the kind of organization that manages to avoid this phenomenon. They’re called high-reliability organizations (HROs).

Back in the 1980s, researchers studying aircraft carrier flight decks, air traffic control centers, and nuclear power plants found something that seemed to defy the odds. Though these organizations were running complex, round-the-clock operations, catastrophic failures were unbelievably rare. One research team noted these organizations “worked in practice, but not in theory.” Regardless of what the textbooks predicted about complicated, high-risk systems, the carriers and control towers kept not failing.

Over subsequent decades, researchers identified a handful of habits that set HROs apart:

• They obsessively think about the things that could go catastrophically wrong, rather than coasting along on the fact that nothing has gone wrong lately.
• They resist the temptation to oversimplify complicated situations.
• They keep leadership connected to what’s actually happening on the ground.
• They let expertise, not rank, determine who steps up to lead in a crisis.
• They build in the capacity to recognize and recover from mistakes quickly, because they assume mistakes will happen.

If all this sounds like common sense, that’s because it is. High-reliability organizations treat risk management as a permanent state of mind. At Lexipol, we call this Total Readiness. For a public safety agency, isn’t a program or initiative with an end date — it’s an ongoing habit of assuming the next Sgt. Reyes discovery is already out there, waiting to be found.

Ask the Uncomfortable Question First

If there’s one question that captures this way of thinking, it’s the one that should be asked in every command staff meeting, long before disaster rears its ugly head: Where can we go wrong? That question goes against the instinct of most organizations (including many public safety agencies), where the absence of a bad outcome gets mistaken for evidence of good performance. In many cases, this can be avoided by watching incidents that have gone bad in other agencies and asking, “Could that happen here?”

Firefighters, law enforcement officers, corrections officers, EMTs, and dispatchers operate in unpredictable, high-pressure conditions where a single missed detail can be the difference between success or tragedy. Agencies that wait for bad things to happen before asking where they might be vulnerable have already given up their best chance to prevent them. Asking the question early requires a leadership team willing to look for weaknesses intentionally, in the absence of any complaint or incident forcing the issue.

In practice, that means going beyond perfunctory annual audits and implementing regular and frequent analyses of trends relating to data such as pursuit and use-of-force incidents, response times, and refusal-of-transport patterns. Looking at the historical numbers and seeing what’s changed or changing can help prevent lawsuits or headlines in the future.

Errors and Mistakes Should Prompt Questions, Not Immediate Discipline

The Sgt. Reyes story raises another important question: What happens to personnel and supervisors when drift is allowed to take hold? What’s the appropriate response to errors and mistakes? [1]

The traditional answer is discipline — period. While that might often feel satisfying, it’s frequently the wrong call. Sidney Dekker, whose research on human error has shaped how the aviation and health care industries investigate their own mistakes, points out how outdated models of error identify the careless individual and then stop. A more productive way is to treat errors as symptoms — windows into what the system asked of that person, what they knew at the time, and why what they did made sense to them in the moment. When dealt with the old way, offending personnel get disciplined, morale takes a hit, and the underlying conditions that produced the drift stay exactly as they were. In an HRO model, the agency questions why the policy or training didn’t hold, and then fixes what it finds.

That’s not the same as excusing misconduct. Willful violations still warrant discipline. But a culture that leaps straight to blame without asking why teaches everyone a clear lesson: Don’t be the one who points out the next problem. A high-reliability agency depends on people bringing up exactly the kind of uncomfortable pattern Sgt. Reyes recognized and reported, and that only keeps happening if doing so doesn’t end careers.

Not Every Problem Needs the Same Response

None of this means every process deserves the same zero-tolerance scrutiny as a vehicle pursuit or an interior fire attack. Kaizen, the practice of continuous, incremental improvement, has its own place in a high-reliability agency, and it operates in an entirely different setting.

There are plenty of lower-risk areas where trial and error can be not just acceptable but useful, including adjusting how shifts are staffed, testing a new approach to non-emergency call response, and refining a community engagement tactic. It goes without saying that the people doing the work every day usually have the best ideas for doing it better.

High-risk operations — especially those that happen infrequently — don’t get that same latitude. As Lexipol cofounder Gordon Graham says, “High-risk, low-frequency events worry me in every occupation. Things that are very risky and done very rarely, the employee does not have the memory markers to deal with these. This is a universal concept.”

A pursuit, a high-risk stop, a use-of-force decision, an interior fire attack, a high-acuity EMS call — these leave way less room for experimentation because the cost of getting it wrong is measured in injuries and lawsuits rather than lost efficiency. Effectively handling the top-left quadrant of Graham’s risk/frequency matrix – where high-risk, low-frequency events reside – requires the discipline of a high-reliability organization: clear policy, rehearsed training, and close supervision, with little tolerance for coloring outside the lines. Agencies need the skill of recognizing, in the moment, which kind of problem they’re facing, and resisting the urge to run every decision through the same setting.

Connect the System Before You Ask It Tough Questions

Sgt. Reyes could answer the lieutenant’s question only because she’d already spent three weeks doing manual, tedious detective work most supervisors don’t have time for. The real obstacle preventing most agencies from becoming an HRO is infrastructure. Agency leaders can’t stay focused on their potential points of failure if the data revealing those points is scattered across disparate systems, both paper and digital.

That’s why performance excellence and interconnected systems are both critical components of Total Readiness. When policy, training, wellness resources, and incident reporting all live inside one system instead of five, supervisors don’t need three weeks to see the problematic patterns. They show up because someone built a system designed to show them, not because a sergeant had a hunch and went looking. That’s the practical promise behind an integrated approach like LexOne, built to connect these functions across law enforcement, fire, EMS, corrections, and dispatch. Solid data and good dashboards can help reveal problems lying in wait, and can actually reach the people positioned to act on them.

Every agency I’ve worked with wants to do right by its people and its community. Very few of them fail because their personnel stopped caring. Instead, they fail because nobody built a way to see small problems before they become big ones, and because the people who spot them don’t feel safe enough to call them out. A learning-driven agency is simply one that has decided to fix both of those things on purpose, and to keep asking where it’s weak long after the last time anything went wrong.

1. Psychologist James Reason distinguishes error, where the intention and plan are correct but execution fails (like when an officer means to draw a Taser and a firearm comes out instead), from mistake, where the plan itself was wrong due to a lack of knowledge. A violation is different still; it may be a deliberate, knowing choice to disregard a rule, or it may occur because of drift. The distinction matters because treating one as the other usually means fixing the wrong problem.

If you’d like to go deeper on the thinking behind this piece — the data problem, how drift takes hold, and where Kaizen and high-reliability principles each fit — check out my free Lexipol white paper, Performance Excellence in Law Enforcement.