Working in the Dark
One of my favourite stories to share at conferences, webinars, and just about anywhere else involves a medical valve manufacturer, a near-doubling of defect rates, and a solution that cost a fraction of what anyone expected.
I was working with this organisation on selection planning when they asked me to take a quick look at a problem in their manufacturing arm. A single production line had seen its defect rate almost double in a matter of weeks, and whilst QA was catching these errors before anything left the building, the risk and cost implications were significant. Doubled defects meant doubled waste in materials, person power, and production capacity; it meant potential supply chain disruption and, ultimately, revenue loss.
The organisation had already invested heavily in addressing what they assumed was a skills gap: manager-led huddles, bespoke e-learning, and a half-day masterclass featuring their top-performing line operators. They were now exploring a VR solution that would allow engineers to practice in a safe, immersive environment. A colleague who had recently had some assumptions about immersive technologies challenged at a conference asked whether I might take a look before they committed further.
My analysis began with what I consider the obvious questions:
When did this start?
What changed around that time?
Have the personnel changed?
The answers were revealing. The same people who had previously achieved half the defect rate were now underperforming, and roughly two-thirds of the line had seen an increase in their individual error rates. This suggested something systematic; these were skilled engineers who had not suddenly forgotten how to do their jobs.
When I asked what else had changed, the list included:
A new shift manager who had redesigned the tooling layout trays.
A reduction in shift lengths following feedback about tiredness.
A significant building renovation that had moved the site to solar power with new LED lighting.
None of these immediately screamed causation. The new shift manager was a former line engineer who knew the work intimately, the shorter shifts had been well-received, and the new tooling trays had been co-designed with the workers to reduce repetitive strain injuries.
The renovation seemed like an unlikely culprit until we spoke directly with the line engineers, several of whom gave us our answer almost immediately.
To up your learning science game in 2026, consider attending the Evidence-Informed Practice Conference as we prepare to explore practical insights from the worlds of cognitive neuroscience, psychology, behavioural science and much more.
The new lights were brighter on paper, but they had been installed at a much greater height than the originals, meaning the effective illumination at the workstation was considerably dimmer. People were struggling to see what they were doing. When we checked data from a second site that had undergone the same renovation, we found a similar spike in defects within two weeks. This was not a training problem; it was a lighting problem.
The fix was straightforward: lower the lights, install adjustable lamps at each workstation, and pause the renovation rollout at remaining sites until the issue was resolved. Within weeks, defect rates returned to their previous levels. The cost was a fraction of what any training intervention would have required, and the solution scaled effortlessly across every affected workstation without the variability that inevitably accompanies any form of training.
We only found the root cause by getting into the data, by asking questions that seemed unrelated, and by being willing to consider that the answer might lie somewhere nobody was looking. That willingness to take a holistic view is what separates performance enablement from assumption-led intervention.