Compliance as Practice, Not a Once-a-Year Box-Tick

I fly a lot, and I stopped listening to the safety briefing years ago. The crew point to the exits and show how the life jacket inflates, and I carry on with whatever I’m doing, because I’ve heard it all many, many times before. Then it struck me that if an emergency ever happened, I would mess it up. I would forget most of what I’ve been told over and over, because I’ve never actually had to do any of it.

Most companies train people for emergencies the same way. Once a year everyone sits through a session, signs a sheet and goes back to work. The company has met its legal duty, and nobody has practised anything.

The research on how skills fade says many of those people could not do what they were trained for when it mattered. Compliance training works better as short, regular practice, with a pass-or-fail check kept for certification.

The stakes are not small. In 2024 the EU recorded 2.78 million non-fatal accidents at work and 3,367 deaths. In Great Britain, 680,000 workers were injured in 2024/25, and injuries and ill health from work cost an estimated £22.9 billion a year.

Aviation shows what repetition without practice does. In US research for the Federal Aviation Administration, the share of passengers paying attention to the safety briefing fell from about 70% in 1979 to 40% or less by 2007.

An Australian study for the Transport Safety Bureau found the briefing was the safety message passengers most often dismissed as irrelevant because they had heard it so many times.

In Australian experiments, even people who had just watched a briefing recalled only a third to a half of its key points.

Skills fade fast when they are not used. A review of decades of studies found that trained skills drop by more than a standard deviation after a year without use (Arthur and colleagues, 1998).

Most of us know this from school. Sit one of your old exams today, in basic maths, algebra or science, and very little of it would still be there.

CPR is the best-studied example. The American Heart Association’s guidelines say CPR skills can decay as early as three months after a course, and a review of advanced life support training found that practical skills fade within six to twelve months, faster than the knowledge behind them.

How people are trained matters as much as how often. A meta-analysis of 95 studies covering almost 21,000 workers found that hands-on training, where people practise the task and get feedback, produced more learning and fewer injuries than lectures, videos and leaflets (Burke and colleagues, 2006).

The gap was widest in the most dangerous jobs, and in low-hazard work passive training did about as well (Burke and colleagues, 2011). So the fire drill matters more on a building site than in an office.

Small amounts of frequent practice keep skills alive. In a Canadian trial with nurses, 58% of those who practised CPR for a few minutes every month performed it to a high standard a year later. Among those who practised every three, six or twelve months, the figures were 26%, 21% and 15%.

Spaced questions sent by email to doctors gave an advantage that was still there two years later (Kerfoot, 2009).

The law already asks for this. Spain’s workplace safety law requires training that is “teórica y práctica, suficiente y adecuada”, focused on the person’s own job and repeated periodically where needed.

The UK regulations ask for adequate training, repeated periodically where appropriate. Neither says once a year, and neither says a signature proves anything.

The UK Health and Safety Executive makes the same point when it separates training from competence and tells employers to check that training has worked.

I’ve taught from badly designed curriculums, the kind that must be followed rigidly. With no follow-up and no spaced repetition, most of my adult students had forgotten a class from two weeks earlier.

Some workplaces already train this way.

Hospitals are the clearest case. The American Heart Association’s RQI programme, run with Laerdal, replaces the usual CPR course every two years with short practice every three months, about ten minutes a session according to its makers. Staff practise on a manikin that gives feedback as they go, and keeping up the sessions keeps their certification current. The company says more than 2,400 hospitals use it.

Kerfoot’s spaced questions became a product. Qstream, which grew out of his research at Harvard, sends staff a few short questions at spaced intervals and is used mainly to train pharmaceutical sales teams.

Building sites already have the regular slot. “Toolbox talks”, short safety briefings on a single hazard, are a long-standing habit, often held at the start of a shift. They are talks rather than practice, but the time is already set aside. Turning it into practice needs no extra slot in the week.

A better design spreads the training across the year. Short practice tasks arrive every few weeks, each one asking people to do something rather than read something: put the steps of a lockout in order, spot the hazard in a photo of their own site, talk through what they would do if a colleague collapsed. The tasks are built on a map of how people usually get it wrong, such as assuming someone else has isolated the power, so each mistake brings back the right practice later. Nothing in this phase is pass or fail. Mistakes are flagged so they can be fixed.

Certification stays a gate. Resuscitation training already works this way: people practise freely, then must meet a set standard to pass. The company gets its certificate and its audit trail, and the people holding the certificate can still do what it says they can.