French telecoms operator Orange has changed how it trains engineers. When employees need to learn cloud technologies, the company doesn't send them on six-month training courses. Instead, workers join new teams immediately while they keep their salary and job security. They learn on the job with mentors who support them. "What makes the difference is not just the quality of the training but of the mentor—the feeling that as a learner you're supported," says Vincent Lecerf, Orange's executive vice-president of human resources.

This approach solves a common problem in many companies: traditional training often fails to produce results. Compliance courses may tick boxes, but they don't have much impact on actual work. On-demand learning libraries promise flexibility, however they struggle to attract users. "People are so busy that only if learning affects the job they're doing right now will they get involved," says Daniel Godfrey of training provider Hemsley Fraser.

Fortunately, humans are naturally good at learning from each other. Psychologist Matthew Lieberman's research shows that teaching others actually boosts our own learning. When we explain concepts to someone else, it forces the brain to work harder. This process reveals gaps in our knowledge and strengthens what we've already understood. Some companies are taking advantage of this principle. For example, fintech firm Infinox has dealers and relationship managers swap skills with each other. Meanwhile, media agency Spark Foundry holds lunchtime debates where staff members research and argue different perspectives on topics.

Brain science reveals several key factors that optimize learning. Sleep is crucial because the brain processes information and forms memories while we rest. Breaking training into smaller chunks works better than full-day sessions, as it allows time for practice and reflection between learning periods. Mixing learning with real tasks—such as practicing a new IT system while serving customers—helps the brain shift information from working memory into more readily accessible long-term storage.

Research by psychologists Robert and Elizabeth Bjork demonstrates that for knowledge to stick in our minds, learning must be effortful and allow mistakes. But what about high-risk roles where errors could cause serious harm? Technology provides practical solutions for this challenge. UK cyber resilience company Immersive thrusts clients into mock cyberattacks with simulated news reports and panicked social media responses. "By solving a problem you're going through the thought processes," explains founder James Hadley. "You're building that muscle memory because when a real incident happens, you need to respond automatically."

The evidence is clear: the most effective learning happens when people practice real skills in realistic situations, receive support from mentors and colleagues, and have opportunities to make mistakes in safe environments. Although this approach requires more effort than simply watching videos or reading materials, it produces lasting results that actually improve job performance.