Last week the UK Learning Power 100 landed, the first index of its kind to rank UK employers on how seriously they invest in workplace learning. Produced by Lyceum Education Group in collaboration with FT Longitude and published on 7 May 2026, it surveyed 2,000 UK employees about training quality, skills planning and confidence at work. The headline most outlets ran was the league-table itself. The number worth pausing on is buried a few pages in.
29% of UK employees are dissatisfied with their most recent training. Roughly three in ten. At an average UK L&D spend of £1,068 per employee per year (CIPD Learning at Work 2024), that is a meaningful chunk of the budget producing the worst possible result, which is people sitting through it and not changing afterwards.
What The 29% Is Actually Saying
Read those numbers as a Venn diagram. Most UK employees believe training, in the abstract, works. Roughly four in five say good training lifts confidence and capability. They are not anti-learning. What they are flagging is that the training they are actually given does not feel like that.
The Learning Power 100 frames this as an emerging-skills problem. 48% of employees rate themselves as merely adequate or worse on AI, cybersecurity and sustainability, the three areas every CEO deck now leads with. Satisfaction with training in those areas, the report notes, is significantly lower than satisfaction with training in established technical skills. The dissatisfaction is concentrated where the future is.
There is a quieter pattern underneath. The FT-backed research echoes a finding from CIPD's Good Work Index and TalentLMS 2026 L&D Benchmark Report: the biggest single barrier to learning is not the budget, the platform or the curriculum. It is time. The training that does land is the training that fits into the working day, that connects to a real conversation a real person is about to have, and that gets practised more than once.
Why The Average Programme Does Not Move The Number
The default response to a skills-confidence gap is to commission more content. A new module on prompt engineering. A new awareness session on phishing. A refreshed e-learning bundle on ESG. The content lands, the completion rate looks healthy, the next survey shows the confidence number more or less where it was.
The mismatch is structural. Roediger and Karpicke (2006), in Psychological Science, established that being tested on material lifts long-term retention by roughly 50% compared with re-reading the same content. Most workplace training is the re-reading kind. A slide deck, a video, a quiz at the end. The half-life of that learning, measured in weeks rather than months, is one of the most robust findings in the cognitive science of training.
Add Kirkpatrick and Kirkpatrick's Four Levels of Training Evaluation (2016) and the picture sharpens. The 29% number is essentially a Level 1 reading: how the participant felt. Level 2 is whether they learned. Level 3 is whether their behaviour changed at work. Level 4 is whether the business outcome moved. Most organisations measure Level 1, occasionally Level 2, almost never Levels 3 and 4. The Learning Power 100 is a useful new Level 1 benchmark. It does not tell us whether the other 71%, the ones who said they were satisfied, actually behaved differently afterwards.
What Actually Closes The Gap
The behavioural-science answer is well established and unromantic. Active retrieval, spaced over time, in conditions close enough to the live moment that the body learns the new default. Baldwin and Ford (1988) called this transfer of training, and forty years of subsequent research has done little to dislodge their finding: the strongest predictor of whether training changes behaviour at work is how closely the practice conditions resemble the work itself.
In Sidestream's own academic behaviour-change work, building on research from UCL, Cambridge and Bocconi, participants who learned a communication skill through immersive role-play scored roughly 20% higher on observed behaviour than those who learned the same content through video or slide-show training. Self-rated confidence did not predict observed performance. A Dunning-Kruger pattern we designed out of subsequent studies by replacing self-reports with behavioural measurement.
For an L&D buyer looking at the 29% number, that has a practical implication. Surveying confidence is necessary but not sufficient. The CFO question is whether the behaviour changed. Reading our piece on measuring the ROI of behaviour change sets that out in more detail.
What We Do About It
Our manager workshops and immersive simulations are designed for the part of the gap a video module cannot reach. Small groups work through realistic scenarios with professional actors, the conversation is run live with feedback, and the same scenario is practised more than once. By the third rehearsal the new behaviour starts to feel automatic rather than effortful, which is the only point at which it travels back into Monday morning.
The 29% number is not a content problem. It is a rehearsal problem dressed up as a content problem.
The Learning Power 100 is a useful first attempt to publicly grade UK employers on how seriously they treat workforce learning. The next question, the one the report does not yet answer, is which of the top-100 employers can actually demonstrate behaviour change at Levels 3 and 4. That is the conversation worth having. Book a call to look at what a Level-3 measurement plan would look like for your organisation.
Book a free 30-minute diagnostic call → or read about our research-backed approach.