To embed leadership development, attach it to live work, make each person's manager the main developer, hold people to account for applying what they learned, and change the parts of the system that still reward the old behaviour. I keep seeing the same pattern: content is rarely the problem. People leave a good programme changed, return to an unchanged job on Monday, and the old pull has them back by Friday.
That gap between the event and the work is expensive, and the evidence says so. McKinsey reported that US companies spend almost $14 billion a year on leadership development, and in a later survey only 11 percent of more than 500 executives strongly agreed that their programmes achieve and sustain the desired results (McKinsey, 2017). That is a design problem. The development was bolted on instead of built in.
Most organisations spend their money and attention in the wrong order. They choose the provider carefully, argue about the agenda, and then invest almost nothing in what happens afterwards.
Why leadership development does not stick, and the embedding gap behind it
I call the distance between where development happens and where the work happens the embedding gap. Close it and behaviour changes. Leave it open and even good training evaporates against a daily reality it never touched. Geerts's 2024 review of leadership development research, published in Behavioral Sciences, puts a number on the decay as reported by earlier work: a 2003 study of 167 financial service managers, cited in the review, found that 64 per cent defaulted to their original work styles after development programmes. The review also notes that regression to a previous state can result from a lack of supervisor support or a culture that inhibits experiment.
- Development is separated from the work: The programme runs offsite on its own calendar, disconnected from the projects and pressures people face daily, so the new behaviour has no live context to attach to. McKinsey's 2014 article names this as decoupling reflection from real work.
- The manager was never involved: In my experience a person's manager shapes their behaviour more than any facilitator does. Yet most development leaves the manager out: they neither attend, reinforce, nor know what was taught. The CIPD's 2023 evidence review finds that recognition, feedback, mentorship and support from the organisation and the supervisor or peers motivate people to transfer what they learned to their day-to-day work.
- Nobody is accountable for applying it: People are held to results, not to whether they applied what they learned, so the learning becomes optional, and optional things lose to urgent things.
- The system around the person did not change: If promotion criteria, incentives and decision rights still favour the old way, the structure wins because it is present every day and the training was present for two. McKinsey's 2017 survey reported that successful programmes were roughly five to six times more likely to involve senior leaders as sponsors, mentors and coaches, and to adapt HR systems to reinforce the new leadership model.
How to embed leadership development so it holds
Embedding is not more training. Often it is less training and more design. These are the moves I make when the brief is to make development stick and not merely happen, and every one of them is about what surrounds the classroom: the manager, the work, the accountability and the structure.
Four moves that embed development
- Anchor it to real work: Attach development to a live remit, a hard project, a struggling team or a real decision. People grow by doing difficult things with support. The work is the curriculum.
- Make managers the primary developers: Equip and expect every manager to develop their people in one-to-ones, feedback and delegation. The manager is the most useful development tool in the company and usually the most neglected.
- Build accountability for application: Hold people to outcomes and to practising the capability you paid for. What is expected and reviewed gets done. What is merely offered gets shelved.
- Change the surrounding system: Align promotion, incentives and decision rights with the behaviour you are developing, so the structure pulls the same way as the training.
Here is how the four moves look in the flow of work. Use the table as a design checklist: for each moment, name the mechanism, the owner and the evidence that it is happening.
| Moment in the flow of work | Embedding mechanism | Owner | Evidence it is working |
|---|---|---|---|
| The weekly leadership meeting | A standing slot where one leader brings a live decision and the group coaches it | CEO or chair of the meeting | Leaders use the same questions without a facilitator in the room |
| One-to-ones | The manager reviews one agreed behaviour from the programme each month | Each manager | The leader can name what their manager is developing in them |
| Project and planning reviews | The capability being built is a named criterion in the review | Project sponsor | Reviews cite the behaviour, not only the result |
| Promotion and talent reviews | Applying the capability is a factor in who moves up | CHRO with the CEO | Promotion cases refer to observed behaviour |
| Six to nine months on | A second round of measurement and a refresh for the cohort | CHRO | Behaviour holds or improves after the programme ends |
We attached development to the leaders' real work: planning, reviews and one-to-ones. I coached the CEO to model the behaviours in the executive meeting first, and the next layer then adopted them. After 24 months the behaviour was part of how the team ran and had outlasted the programme. In another change the programme finished and attention moved on. I advised building reinforcement in from the start, so leaders reviewed the behaviour regularly and rewarded it, and the change became normal. The lesson from both: reinforcement is planned before the first session, not added after the results disappoint.
Start small. In the first 30 days, pick one cohort and one behaviour, and wire it into one existing meeting and one set of one-to-ones. Do not build a new forum for the purpose. A new forum is an event with a calendar invitation, and it fails for the same reasons. Ask the sponsor to attend the first two sessions of the weekly slot and to say, in their own words, why the behaviour matters to the business. Senior attendance is the cheapest reinforcement signal you can send, and McKinsey's data points the same way: senior leaders acting as sponsors, mentors and coaches mark out the programmes that work.
Then protect the time. Geerts's review notes that participants in programmes that underperform frequently cite a lack of time to reflect, discuss and apply what they learned. If the practice slot is the first thing cut when the quarter gets busy, you have your answer about how seriously the organisation takes the behaviour. Say so, plainly, to the sponsor.
The manager is the multiplier you are ignoring
If I had to pick one lever, it would not be the programme design, the facilitator or the framework. It would be the manager of the person being developed. A supportive manager who reinforces the new behaviour in daily work will make an average programme stick. An absent or dismissive one will make the best programme evaporate. Yet managers are the group most consistently left out of the development they are meant to reinforce.
So when I redesign development to embed it, I start there. I bring managers into the process, teach them how to coach the specific capabilities being built, and make developing their people an explicit, reviewed part of their own role. This is the logic behind integrating leadership development into everyday activities: stop treating development as a place you go and treat it as a way you work.
There is a compounding effect most organisations never make possible. When managers develop their people well, those people become better managers, who develop their own people well. The capability replicates down through the organisation because you built the mechanism that reproduces it instead of the event that delivers it once. A programme develops the people who attend it. An embedded system develops the people who will never attend anything, because their managers carry it to them.
How to measure whether development has stuck
An embedding effort without measurement is another good intention. Geerts's review recommends collecting data from multiple sources at the outset, throughout, at the end, and six to nine months after an intervention, and says end-of-programme data alone can give a false negative because outcomes take time. It also notes that programme evaluations typically only assess participant satisfaction. SHRM's 2025 guide to measuring the return on leadership development gives the arithmetic as total tangible benefits minus total programme cost, divided by programme cost, times 100, and quotes a leadership advisory consultant recommending a wait of at least nine months before checking whether participants apply what they learned.
- Agree two or three observable behaviours with the sponsor before the first session.
- Take a baseline: a 360 or stakeholder interviews, and the performance data you already hold.
- Check at the end of the programme, then again at six to nine months, with the same raters.
- Ask the manager one question each month: what are you developing in this person, and what did you see?
- Share the result, including a poor one. A weak first number is a design input, not a verdict.
For the full chain a CHRO can defend to a CFO, read how CHROs prove leadership development changes business performance and why most companies cannot measure leadership development ROI. For what counts as evidence in the first place, see what is evidence-based leadership development.
Where embedding connects to leadership capability architecture
Embedding is not a technique you apply to a programme. It follows from building development into your leadership capability architecture. The Embed step is where capability gets built into the leadership workflow. It sits between Design, where I diagnose constraints and align systems to strategy, and Scale, where leaders develop leaders. When development is a designed part of how the organisation runs, connected to the work, owned by managers and reinforced by the structure, it stops being fragile.
It is also why so many programmes fail to change business performance: they invested in the event and neglected the system that would have carried it. I set out the causes in what causes leadership development programmes to fail, and the diagnostic version, five tests of your own programme, is in why leadership development programmes fail. I wrote The Leadership Shift because I kept seeing organisations invest in leadership development that did not stick. The idea I want readers to take is that capability has to be built into the system leaders work in, not delivered alongside it.
Ask what happens on the Monday after
If you are about to approve another year of leadership development, ask a different question from which provider or which framework. Ask what happens on the Monday after. Ask who reinforces the new behaviour, where people practise it, and what in your structure will either hold it or undo it. Answer those and the development compounds. Ignore them and you buy, once more, an event people enjoyed and then forgot.
If you want a place to start, the leadership development that sticks guide sets out the sequence, and the Architecture Accelerator is where I build it into a leadership team's operating rhythm. CapabilityAI is my AI leadership companion, designed around development in the flow of real work.
A programme changes who attends it. An embedded system changes how the company produces leaders, whether or not a programme is running.