[Mary] Hi there. Welcome to Safety Labs. When you implement changes in your organization, you do so because you think it will improve safety outcomes. Which specific outcomes are you trying to improve? And how do you know whether your systems are working? Our guest today believes that we need to be very clear about how we evaluate safety systems — what metrics we use, and whether they're giving us the right information.
Corrie Pitzer has worked in the safety industry for the last thirty years, first as group human resources manager and group risk manager in South Africa. He founded SafeMap in 1994. Corrie specializes in safety transformation in high-risk organizations and is regarded as a leading specialist in his field. Many corporations utilize his expertise on culture, risk management, and leadership, including Nike, De Beers, and Southern Company, among many others. He has a Bachelor of Arts in Industrial Psychology, an Honours in Business Administration, a Master's in Business Administration, and a Graduate Diploma in Higher Education. Corrie is also a registered psychometrist. He joins us from Vancouver, just across the Salish Sea from where I am. Welcome.
[Corrie] Thank you very much, Mary. Nice to be here with you.
[Mary] When you zoom out and look at the safety industry as a whole, what changes have you observed in the way organizations approach safety, on a scale of decades — from when you started until now?
[Corrie] That's a very broad question, and so many things have changed, mostly for the better. As you mentioned, I came from the human resource field, and I've always looked at safety from a human perspective in the broadest sense possible. I was the group human resource manager for a large platinum company at the time, and then there was a very big disaster, and I was placed into safety. I started a process in this organization — it had over sixty mines in South Africa, what is today known as BHP — and I introduced a week-long intervention, which was really a HOP intervention. I based it at the time on Deming's principles — the book had just been published. But outside this organization, in the broader world, there was still a very strong and growing focus on behavioral safety, BBS, which I thought was quite a narrow focus. And on the other side there was this continued engineering focus — safety as something to be engineered within the organization's operational processes. Those two things coexisted.
Over time, behavior-based safety became very popular. Then about ten years ago, the work of Erik Hollnagel, Todd Conklin, and David Woods started to broaden the field significantly — initiated, I think, by James Reason, who had a tremendous impact on modern safety thinking. And so now we are very much in an era where we look at safety from a very adaptive perspective, which I think is a great thing.
[Mary] How would you summarize your core argument, or your feeling about where we're still getting things wrong in safety today, or where we still need to push further?
[Corrie] I think generally we're getting things right — organizations are moving in the right direction, broadening the perspective and definition of what "safe" means, which is the starting point of everything. But there is just so much real estate in an organization, only so much budget — so where does it go? We have this almost internal conflict between what's generally viewed as the old view of safety — Safety I, BBS, engineering safety — and the new view. And I think it's almost a trade-off, because not many organizations can truly step into the new thinking. The reason is that we don't yet have clear tools and techniques in the new view of safety to actually make that work. The tools and techniques we are well-versed in are all old-view safety tools. We're stuck in a dual world — conceptually we know where we should be, but we're trapped in old thinking by virtue of the processes ingrained in the organization. Legislation and regulation are very strongly steeped in the Safety I focus too, so it's almost hard to escape. Unless we as an industry and profession can develop more sophisticated, advanced techniques and tools, I think we risk losing the battle, so to speak.
One example: in the utility industry in the United States, there's a growing focus on energy-based safety, which is actually a regressive approach — it comes from the seventies, from the way we used to think about safety as purely the release of energy in the workplace. The human factors dimension is where we should be going, but the tools and techniques aren't taking us there. So there's a duality and a dichotomy that's almost unresolvable at this stage.
[Mary] I want to get more specifically into measuring and metrics, but to lay the background — your bio mentions you have accreditation in psychometry, which I hadn't heard of. Can you share what that is, and how it informs your view of safety measurement?
[Corrie] Psychometry, or being a registered psychometrist — it's a qualification a little like registering as a psychologist, where you go through a period of practical exposure and then register with a medical board. In South Africa, this was a registration done at the medical board, with A, B, and C level registrations. The C level was at the advanced level — measurement of things like personality and culture. And that's what psychometry is all about: how do you measure essentially complex psychological constructs? That is always a challenge, and it has certainly driven my thinking around safety.
There are two key things in psychometry — how valid is the measurement, and as a result of that, how reliable is it? These are very basic concepts, but incredibly important, because we use metrics far beyond these requirements of what makes a valid measurement. Psychometry has had a significant impact on my approach to safety over the years.
[Mary] Injury rates have traditionally been the main way of "proving" — I'm doing air quotes — that safety is working. What's your view on using injury rates as the primary proof?
[Corrie] I'm fairly outspoken about this. You almost have to think about metrics from first principles — the reason for metrics is that you want information to see how effectively you're managing your business. Now, what typically happens in safety is we choose the metric and then decide what we want to know. But coming back to psychometry: what is the construct of "safe" in an organization? If you have injury rate metrics, you're actually measuring injury rates — you're not measuring safety. The question is: are injury rates a proxy for safety? And the answer is clearly no. It's a fraction of what "safe" means in an organization, and for many reasons.
One is that the information that eventually surfaces as injury rates has gone through various levels of distortion — not necessarily with bad intent, but naturally, as it filters through the organization. The person who makes a mistake and has a small injury already faces pressure not to report it. So you have all these distortions going through the organization. That's why I call injury data "dirty information" — what you see at the top as your metrics is by no means a reflection of what originally happened in the organization.
I've started to construct a little model of the five levels between what happens at the bottom of an organization and what needs to happen at the top. The first is the operational level — what I call the friction level, the gap between work-as-imagined and work-as-done. There are maybe six or seven sources of friction there: the design of processes and procedures built for conditions that don't quite exist; goal friction between costs, quality, scheduling, and safety, all being handed down to one pair of hands at the front line; resources — whether people have the right time, equipment, scheduling; information friction — the difference between what they know and what they've been given; and the interfaces between shifts, crews, contractors. All of this is turmoil at the front end, going on every single day. We know very little about it. It's almost never measured.
Out of this mass of friction come signals — mistakes people make, workarounds, clashes between people or departments. Still abundant, but we don't have the mechanism to actually measure that. Then it gets to data points — near misses, incidents — which are what we actually capture. That's already a very small fraction of the underlying friction. Then it's turned into metrics, defined by injury rates. So we've gone from a broad base of information, through these filters and layers, to a pinnacle point where we say this metric here measures the condition of safe or unsafe. It is actually bizarre to think that we do that.
I often hear, "but that's all we've got." You can't justify the use of invalid, unreliable data just because that's all you've got — that seems a bizarre argument to me. And then those metrics get charted, put into graphs, aggregated, brought into board meetings, turned into targets, and start bending backwards down the organization, creating incredible increases in friction at the front end. The whole cycle of measurement drives everything in the business. We have great measurements in the field of operational efficiency — we can measure outputs, quality of products. But safety is at the complete other end of that scale. The information we get in safety is not valid, and we make the same decisions about safety data as we do about operational data. While we can make valid decisions about operational data because the dataset is large, we cannot make valid decisions with safety information. Yet we do.
[Mary] You've borrowed a metaphor from Neil deGrasse Tyson — the bucket in the ocean — to describe safety metrics. Can you explain that, and then talk about how we might start to address this dilemma?
[Corrie] The bucket analogy comes from a video I was watching — Neil deGrasse Tyson wasn't thinking about safety at all when he said it, but what he said is: you can sit in an ocean on a boat, lower a bucket into the water, pull it out, and measure what's in it — you can do that millions of times with millions of boats for a million years, and you'll never measure the existence of whales in the ocean. So here you have an incredibly sophisticated, costly measurement system, but the bucket is wrong for the purpose. Injury metrics are the wrong bucket. We will never measure the concept of "safe" using them.
And yet we keep using them. I think the answer comes back to what I mentioned earlier: it's forced upon us. Bigger forces — investors, regulators, executives needing to tell shareholders we're doing well — that's the fundamental psychology behind it. And what saves the day for us, most of the time, is that what surfaces as metrics represents such a minute piece of information about the broad base of friction that we can effectively get away with it for long periods. Organizations can survive themselves, if I can put it that way.
I've done a fascinating activity for a large organization with over a hundred thousand people across forty or fifty sites around the world — a process we called "elimination of fatalities." We'd bring in a team of twelve to fifteen people from different parts of the company, meet on a Sunday for orientation, and then during the week, split into pairs and filter into workplaces to engage with frontline employees. We used what I call the "six Y lens" — asking: what risks do you see here that may be misjudged? What risks do we take because of shortcuts? What risks do we fail to see? What came out of this was remarkable — we literally identified thousands of SIF potential events, serious injury or fatal potential events, sitting in the organization. At the end of the week, feeding it back to management, they were astounded. During the two years this process happened, that large organization had one or two fatalities. But there were literally thousands of risk situations that we discovered — and had we gone through the same site the next week with another team, we'd have found hundreds more.
So the ocean we're sitting in, the times we actually see a whale's nose breaking the surface, is completely devoid of information about how many whales and sharks sit underneath that never come up. Why? We are lucky — or more scientifically, it's randomness. There's a whole soup of randomness inside every organization, and the chances of a catastrophic event surfacing as a metric are so small that we can go and tell our investors and shareholders, "here's a metric that says we're safe," because we haven't had any serious accidents, haven't had a fatality, and here's our injury rate that proves it. Therein lies the dilemma.
[Mary] If someone's trying to find answers, they have to think about the questions. If you could redesign the evaluation process for safety, what questions should an organization answer before claiming success with a given initiative or tool?
[Corrie] That's almost a qualitative question, and you have to go upstream in the organization to answer it. I essentially use seven dimensions or sets of criteria to judge any intervention.
The first is: how clearly is the purpose of this intervention defined? What is it actually for? The second: how sound and applicable is it for our organization, our context? Once you have those assessments, the third is: how well is it deployed? What is its reach in the organization — not only breadth, but vertically, how deep does it reach, and what is the quality of that deployment? These three things determine the fourth: what is the impact, the result? To me the most important question here is: does it increase efficiency or slow work down, from a safety perspective? Does it create more bureaucracy, or make the process leaner? This is where things normally fail, because safety interventions typically become additional processes taking additional time — and there's only so much real estate in an organization. The fifth: how well is it integrated? Is it bolted onto the work process, or is it integrated with it — a parallel process or woven in? Judge any intervention against this and you'll find they fail here almost always. The sixth: has it been improved over time? Was it installed and left to exist, or is it kept alive through constant improvement, evaluation, re-examination of its purpose and applicability? And the seventh, in the long term: how sustainable is it? Does it need to be propped up constantly, or does it sustain itself?
And we are still miles from the question of whether this makes a change to the metrics in the organization. I think we should never want to make that connection — never want to say this tool, this training program, this campaign "improved our safety performance" as measured by metrics, for the reasons I've already mentioned. What we're actually doing is limiting and curtailing the randomness of events — clearing friction at the bottom end of the organization, and as a result reducing the odds of the whale jumping through the surface and crashing on a boat. Not preventing it entirely — we can never do that — but reducing the odds. If we go upstream and do this qualitative analysis, we are giving ourselves a better chance of not having that catastrophic outcome.
[Mary] If a safety leader realizes that their metrics and the questions they care about are a poor match, what diagnostic questions should they start with?
[Corrie] To some degree I've already answered that — doing qualitative assessments upstream in the organization. But here I'd want to be very cautious, because what many listeners might then think I'm advocating is "leading indicators," upstream qualitative measurements. And there I want to be careful, coming back to validity and reliability. What does it actually measure?
For instance, if we say we're going to measure safety observations — peer-on-peer observations, or whatever form they take — the direct measurement of those observations is the number of observations. What it does not measure, and yet is often claimed to measure, is safety culture, the extent to which people are in a collaborative mode, wanting to help each other. It's got completely nothing to do with that. It measures the number of observations.
Another one I have strong feelings about is leadership walkabouts, which is the go-to process for most organizations. What can I do? Go on a gemba walk. I've seen this so many times — they have a schedule, often on a Tuesday because Mondays are too busy, and all the top leaders go out to engage with workers: field engagement, visible field leadership, all fancy language. But in the end it only measures how often they walk in the field. And there are even downsides — everybody knows the suits are coming out on a Tuesday, so you are actually eroding your own credibility through that process. People prepare, things look clean, you get a little ten-minute snapshot from a person presenting information to the boss, made to look as good as possible. It's fudged. On Tuesdays, everything goes well. So you have this delusion of what the workplace really looks like, and you think you've done a diagnostic. And I know this may sound very pessimistic about something that's supposed to be a good intervention when done well — but I'm still waiting to see one done well. When I posted about this on LinkedIn recently, there were something like thirty thousand views and the comments were confirmation, people saying, yes, this is really what's happening in the workplace.
We have to be so careful. Going back to those dimensions: what is the design of a leadership walkabout? Is its purpose clearly defined? Is it applicable for the organization? How well is it deployed? What is the impact? If we honestly answer those questions, we might be able to design a leadership walkabout that genuinely makes a positive impact — one where people actually want their leaders around. I spent the first twelve months of my career working underground on a mine to get a blasting ticket, because my manager insisted that as a human resource manager, if I didn't know what people actually do, I couldn't select people for the job. And I can tell you: the last thing we wanted underground was management coming to do their walkabouts, because they disrupt the workplace, ask questions that feel almost uninformed, and you have to perform theater. That hasn't changed much.
There are genuinely good diagnostic tools, though. I firmly believe that a well-designed, well-executed culture analysis can give you great information. I've done it with organizations that used the information constructively — we actually measured friction and turned it into data and metrics for decision-making at the top. One large utility company, around thirty thousand people, did this about ten years ago, took action on all those outcomes, and literally transformed their organization into a remarkable one. It was a valid, reliable diagnostic, used for informed decision-making. That process is possible. But if we try to get away with little snapshots, quick-and-dirty metrics — and I've heard that term before — there is no such thing as quick-and-dirty in safety. You have to do high-value, high-quality, valid measurements, and not think you can use symptoms of systems as your indicator.
[Mary] Before we go further, let's hear from the company behind the podcast.
In warehouse settings, safety competes for attention, and traditional box cutters can rust or dull, which creates laceration risks and productivity slowdowns. The Slice Box Cutters come in manual-retract and auto-retract options. They have an award-winning ergonomic design, blades that never rust, a finger-friendly blade edge that's safe to the touch, and blades that last up to eleven times longer than traditional blades. If you'd like to learn more about which tools will work for your materials, you can request free samples by going to safetylabspodcast.com/samples.
And now, back to our chat. You've said that evidence without context is just data. In safety, what are some of the crucial contextual factors that can completely change the meaning of the same number?
[Corrie] The first and most fundamental context for me is: how do we define "safe"? That is the be-all and end-all of everything we do in safety. If you define safety as the absence of injury, and that's your proxy for the concept of "safe," then you'll follow certain processes, install certain interventions to measure and prevent that — whether you're actually preventing it or just reducing the reporting of it, that's your focus.
The more modern definition is safe as the presence of capacity, which comes out of the new view of thinking. And then there's also safe as the presence of control. Now, those two — the presence of capacity and the presence of control — don't sit at the same level. Presence of capacity tells us about upstream capacity in the organization: the quality of leadership, the quality of systems. Presence of control is at the pointy end of the organization, the front end, where people work.
We've had many problems in the past by focusing on the front end and on what people do and fail to do. That's the reason we started blaming frontline employees for accidents — because they make the accidents. That's why behavior-based safety had such a devastating impact on organizational culture, by putting the worker in the crosshair of management as the focus. And it's the same if we define safety as the presence of control — because then, as with the energy-based safety model, we're focusing all the effort at the front end. Where does energy get released? Where the workers work. And why does it get released? Because most of the time the workers fail to keep the controls intact. So here we go again — we haven't made much progress from thirty, forty, fifty years ago if this remains the context.
My approach has always been to advocate a redefinition of "safe." Our definition is that "safe" is actually "safer" — you can never be safe, but you can be safer. And that definition is: the readiness to respond to risk, relentlessly. That definition has a whole series of implications. Readiness to respond to risk relentlessly should exist at all levels of the organization — strategically, tactically, operationally — and across all dimensions: people management, leadership, systems, equipment. You can actually build a whole advanced, progressive model of context in safety if you scope out this definition at all levels. At the corporate or strategic level, readiness is very much a capacity of culture, systems, leadership. At the front end, readiness means awareness, the capability to respond to risks, understanding where they exist — and "relentlessly" is how we actually focus on that every day, every job, every task. It's a definition that can run throughout the whole organization at all levels.
Organizations that use this definition of safer find it to be a very progressive focus. And there are metrics you can actually use within it — clear metrics that you can measure. Most organizations don't currently have the process, the signals, the data capture, or the systems to do it. But in our work with clients, we create those systems, those processes, that deliver these metrics so you can make informed decisions. Not completely perfect, but you can actually create what I call risk metrics or latent metrics.
Let me give two examples. We talk a lot about near misses. What is a near miss in practical terms? Something that nearly went wrong. Now, some near misses are not hideable — a derailment on a rail track, something fell over, there's visible damage. People report those readily because they can't hide them. But there are many near miss events that are hideable — there's no witness, nothing went wrong to the extent of visible damage, only you were there. I have a video clip I often use: a person in a warehouse with a forklift pushes a pallet up onto shelves, it hooks on the rack and starts to pull the whole shelf forward, but they stop just in time and release it. If there wasn't a camera, no one would have known. There's now a weakness in that shelving, but it stays unreported.
If that person actually came forward and reported that as a near miss, think about what that means — they were confident enough, psychologically safe enough, to report it, and now someone can go fix that weakened shelving. That is almost a metric of cultural maturity. When I re-analyzed all the data from the two-year fatality-elimination exercise for that corporation, I looked at the ratio of these hideable versus non-hideable events across more and less mature sites culturally. It was very clear: more mature sites had many more reports of hideable events. Immature sites, with distrust and fear, only get reports of things people couldn't hide at all.
So: look at your near miss reporting, and if you can find many reports of things people could have hidden, you're sitting in a good spot organizationally. That's a cultural metric sitting right there in your existing hard data.
A second example: we have all this focus on control failures — but it's virtually impossible to measure control failures honestly. The moment you start measuring through formalized audits or inspections, systems miraculously come alive because everyone knows an audit is coming. So you get a false measurement. But we have a mechanism in our toolkit where people can identify controls in a random, non-telegraphed way and measure whether they exist or not. I now have about thirteen to fifteen of these metrics that we've developed over time, which give you this upstream measurement of the readiness to respond to risk relentlessly.
[Mary] We're running a little low on time, but I want to get to the question of where is the evidence that HOP works. What do you think the real goals of HOP are, and how should we evaluate its success?
[Corrie] The last thing I'd want — and this has been a debate across the industry — is a demand that HOP should prove it works by reducing injury rates. For all the reasons we've discussed, injury rates are completely invalid, unreliable metrics, and claiming a connection to them is the wrong approach.
Here's what HOP actually does: it operates at that friction level in the organization. It clarifies the purposes for people at the front end. It enables people to work together in teams better. It creates a better relationship between management and supervisors. It eases and creates flexibility in the organization. And as we open up those layers of how friction converts into metrics — the willingness to report — we create an increased willingness to report. So if you have a successful HOP program, you should actually have an increase in reported incidents. Because you're measuring injuries, you should have an increase in reporting.
Ten years ago, with the utility I mentioned, one of the things we measured was that they had a zero target — a zero-harm mantra, extremely powerfully managed — and it was suppressing information. Significant injuries had been suppressed from reporting. My recommendation was to forewarn them: you're going to have an increase in your injury rates. They accepted that, managed it, and the organization started to become much more transparent. They started having weekly meetings — I've sat in one where contractors from all over report openly what went wrong that week. Now they know what they're dealing with. To this day, they have a continuing gradual increase in incident reporting, but their serious injury rates are significantly decreasing.
You can't demonstrate this on a small scale with insufficient metric volumes. But you can prove, in any organization, that a HOP approach is improving the quality of work, of teamwork, of operational work at the friction level of the organization — less workarounds, less frustration, a much more congenial way of working together. That is the evidence. And you cannot and should not be required to go beyond that as a standard of evidence.
[Mary] For the next generation of safety professionals, where would you focus training — what skills should be pushed further than they have been?
[Corrie] It comes back to the first question, really. The skill of understanding numbers, of understanding psychometry, of understanding measurement — because measurement is the centerpiece in organizations. What we measure drives the organization. The drive of the organization should determine the metrics, but that's not how it works in reality — we start measuring, the measurement goes up the organization, becomes a target, gets driven down, and starts to affect that bottom friction level. So understanding psychometrics and statistics at a very basic level is an essential skill.
The other area, closely related, is whether safety is being driven towards human psychology or towards engineering in the broadest terms. Most safety people come from an engineering perspective — they want to engineer hazards out of the workplace, that's just fundamental. So the other area I'd push is understanding safety psychology, or more importantly, risk psychology — understanding risk from a human perspective, not just from a physical hazard perspective. Energy in the organization — working at heights and all these things — that's at the center of the picture. But sitting around that center is the entire context of human behavior: how do we perceive those risks, to what extent are we overlooking or tolerating them, how do we judge and understand risk analytically? We can do as much as we want on the physical engineering side, but people interact with risk, and that interaction is where we need to drive our focus — not just on people's behavior, not just on the energy in the middle, but on the interaction between them. That is a skills area, a perspective, I think is being missed.
[Mary] For listeners who want to learn more about the topics in our discussion today — psychometry comes to mind, since I hadn't heard of it before today — are there resources you'd recommend?
[Corrie] My website, safemap.com, has resources and white papers on various concepts we use — and increasingly so going forward. I'm also the co-founder of a conference called Safety on the Edge, a global forum in partnership with the NSC. We've just had our most recent one, so come back next year — it's in May — because we're bringing together thinkers and thought leaders. It's not a run-of-the-mill conference where you look for gloves and boots. It's about concepts, ideas, and interventions. The two resources: safetyontheedge.com and safemap.com.
[Mary] And if listeners want to reach out to you, they can find you on LinkedIn as well?
[Corrie] Yes, they can Google me on LinkedIn.
[Mary] Well, that's our show for today. I hope listeners enjoyed it as much as I did. Thank you for joining me, Corrie.
[Corrie] It's been a pleasure to be with you. I think you've got a great podcast — I've been watching a few episodes since you asked me to be on, and it's a valuable resource.
[Mary] Thank you, and thanks to our listeners who make it valuable, and the Safety Labs team behind the scenes. Bye for now. This podcast is created by Safety Products Global, the world's leading manufacturer of safety knives. Through our trusted brands, Klever, Slice, and PHC, we empower companies to prevent injuries by providing safer cutting tools for every material and application. Until next time, stay safe.
