Sidestream takes the Department for Education's Employer Skills Survey 2024 figure of £1,700 per employee per year, down from £1,960 in 2022, and asks where it goes. Across the past decade the outcomes that budget is meant to move have not: Gallup puts UK engagement at 10%, McKinsey finds 70% of change programmes miss their goals, DDI finds only 11% of organisations rate their leadership bench as strong.
The Department for Education's Employer Skills Survey 2024 put UK employers' training spend at £1,700 per employee per year. For a 1,000-person firm, that is £1.7 million annually. For a 10,000-person organisation, it is £17 million. By any measure, this is a serious budget line, even after falling from £1,960 per employee in 2022 (in 2024 prices).
Yet across the past decade, the headline outcomes that L&D budgets are supposed to influence have not noticeably moved. Gallup's State of the Global Workplace still puts UK employee engagement at 10%, below the Europe average of 12%. McKinsey's transformation research still finds 70% of large change programmes miss their stated goals. DDI's Global Leadership Forecast still finds that only 11% of organisations rate their leadership bench as "strong", down from 18% in 2011.
So where is the £1,700 actually going?
The Three Layers Of A Typical Budget
From the engagements we run with mid-to-large UK organisations, a fairly consistent pattern shows up when budgets are dissected:
- Layer 1, Compliance & baseline (40–55%): mandatory training (data protection, health & safety, code of conduct), compliance refreshers, the LMS itself. This layer must exist. It is not designed to change anyone's behaviour beyond the legal minimum.
- Layer 2, Generic capability (25–40%): leadership courses, communication training, team-building events, off-sites. Often well-rated in the moment. Rarely measured beyond Kirkpatrick Level 1 (reaction) or Level 2 (knowledge).
- Layer 3, Targeted behaviour change (5–15%): anything where someone has actually defined the behaviour to change, baselined it, designed an intervention to shift it, and measured the shift after the fact. This is the layer that moves the business numbers.
The disproportion is the issue. The layer most strongly correlated with outcome change typically gets the smallest share of budget, often under 10%, and is the first thing cut when costs come under pressure.
Why The Disproportion Is So Sticky
Two structural reasons.
The first is measurement drift. Compliance and generic capability training are easy to measure on completion and satisfaction. Behaviour change is harder. So organisations report, to boards, to regulators, to themselves, on what they can measure, and that is overwhelmingly Levels 1–2 of the Kirkpatrick framework. The numbers look good. The dashboards stay green. The actual behaviour at Level 3 stays largely the same. A 2026 benchmark puts the figure at only 37% of companies measuring training by business impact.
The second is delivery convenience. A 90-minute e-learning module can be deployed to 5,000 staff overnight. A live, scenario-based behavioural intervention with realistic practice and feedback can reach a maximum of perhaps 200 people over two days. Per-head cost is higher; per-head impact is also dramatically higher, but procurement processes built around scale tend to favour the cheaper delivery channel even when the evidence on retention and behaviour transfer (Roediger & Karpicke 2006; our own work building on UCL, Cambridge and Bocconi) clearly points the other way.
What A Better Allocation Looks Like
The shift we tend to recommend is unromantic: protect Layer 1, shrink Layer 2 by 20–30%, redirect that delta to Layer 3. The compliance layer must still exist. The generic capability layer is the one that quietly tolerates the most waste, the third leadership course on similar content for the same audience, the recurring off-site that nobody can remember a behavioural outcome from. Making that call well is a capability of the HR function itself, which is the subject of our HR training consultancy guide.
The freed budget pays for fewer, deeper interventions in the layer that actually moves the dial: a live behaviour-change programme on a defined target behaviour, baselined, delivered, and measured at 90 days. Done well, ten of these across an organisation move metrics that no number of e-learning modules ever will.
The Question Worth Asking This Week
If you sit on the L&D budget for your organisation, the most useful question to ask this quarter is not "have we delivered enough training?" It is "for our largest annual spend line within L&D, what is the named behaviour we are trying to shift, and how would we know if it had moved?"
If you cannot answer that question for the line, the spend is paying for delivery infrastructure, not for change. That is sometimes appropriate. It just should not be the default.
Book a free 30-minute diagnostic call → or read how to measure the ROI of behaviour change training.
In summary: where UK L&D budgets go
UK organisations spend an average of £1,700 per employee per year on training (DfE Employer Skills Survey, 2024), yet the outcomes that money should shift have stayed flat for a decade. The reason is allocation: most of the budget pays for delivery infrastructure, while the layer that actually moves business numbers, targeted behaviour change, gets the smallest share and is cut first.
- Average UK training spend is £1,700 per employee per year (DfE Employer Skills Survey, 2024), a serious budget line by any measure.
- Budgets split roughly into compliance and baseline (40 to 55%), generic capability (25 to 40%) and targeted behaviour change (5 to 15%).
- The behaviour-change layer is most correlated with outcomes but typically under 10% of spend, and the first thing cut under pressure.
- The recommended shift is to protect compliance, shrink generic capability by 20 to 30%, and redirect that to fewer, deeper interventions measured at 90 days.
Sources: Department for Education, Employer Skills Survey 2024 · Gallup, State of the Global Workplace 2026: United Kingdom · DDI, Global Leadership Forecast 2021: Global Leadership Trends · Kirkpatrick Partners, The Kirkpatrick Model