"How safe is 'safe enough'?"

Klik hier voor de Nederlandstalige versie.

is what someone asked me during the reception at the Aviation Symposium organized by the Dutch Public Prosecution Service, where yesterday I had the opportunity to talk about my vision on safety. I immediately bounced it back and asked: “Why that question?” The man answered me that as an inspector/auditor he visited many places where he sincerely wondered whether considerable efforts to maximise safety were worth the minimal gains. Pareto, you know? Dedicate 80% of your resources to get rid of that last 20% - for all I care you can freely choose your own percentages. How much money and effort do you still have to put into something that is already very safe or in other words what do you consider an acceptable level? What do you think is ALARP (As-Low-As-Reasonably-Practicable)?

“Good question,” replied a third interlocutor. “When I am asked where I want to set my safety target for the coming year, it is difficult for me to say: well, one death seems okay to me?!” I agree, that might be tough. In the end, we all want everyone to be able to go home at the end of the day and not have to call from the hospital? The only problem with a zero target is that we often drive incidents underground. After all, no one wants to reset the accident-free days counter at the entrance to his or her company. So things are not reported or worse 'tipp-exed', as Bart Vanraes so eloquently puts it.

But what should we do? What if zero is a noble and ethically sound objective, but we still want to offer people the freedom and psychological safety to report things? I think (part of) the solution lies in how we define safety. We currently still do this based on the number of incidents and accidents. The less of both, the safer we consider ourselves. This works fine if you have a lot of incidents, but as the number decreases, there is less to measure (the so-called safety paradox). We typically then desperately start looking for more data. More reports (“Don't be afraid, we have a Just Culture”), more automated data that we try to summarise in (often meaningless) dashboards and statistics. Because after all, we have to be able to demonstrate how safe we are to authorities, shareholders, customers, employees and - if things really go wrong - the judge. But do you really believe that you will convince a judge with those numbers alone? He - just like the others - will probably look for what you did to prevent the accident.

Perhaps we should therefore expand our definition of safety and not just rely on the absence of an unacceptable risk. After all, risk assessments often remain an 'educated guess' and are often less objective as they appear. Perhaps we should also define safety as the capacity (resilience) to deal with unexpected circumstances (Hollnagel, 2014). That sounds very abstract, because how on earth are you going to measure how your organisation deals with a crisis? Yet it is not that difficult. After all, your employees solve many small crises every day. They apply rules and procedures flexibly to get their job done. Understanding how your people achieve this gives you insight into what you can do to make their work easier, more efficient, of higher quality and safer. But not only that. You ensure that people throughout the entire organisation gain the confidence to not only report things that are not going well, but to actively look for solutions. After all, they are the experts in their field. My own research tells me that Learning Teams can be a useful tool in this regard.

If your employees at all levels actively look for improvements, and you tackle issues in a context of mutual trust, the question “how safe is safe enough?” gets an answer that is supported by the entire organisation which makes the judge's work a lot easier.

Hollnagel, E. (2014) - Safety-I and Safety–II: The Past and Future of Safety Management, Ashgate Publishing, Ltd