In my view a leadership development program fails when the organisation treats leadership as something people attend instead of something the business runs on. I watched a CFO shut down a brilliant product innovation proposal in a board meeting last year, despite the fact that her leadership development program had explicitly trained her to champion strategic risk-taking. Not because the numbers didn't work: they did. Not because the strategy was flawed: it wasn't. She killed it because her leadership team couldn't articulate why it mattered beyond the spreadsheet. The test is simple: six months after the last session, can you name a decision that is now made differently because of it? If you cannot, the program has not worked, whatever the feedback scores say.
I treat this as an architecture problem, not a quality problem. Working with hundreds of senior leaders has shown me this: the gap between a director who stalls at that level and one who moves into the C-suite rarely comes down to technical skill. Most programs are well designed and well delivered. What they lack is anything around them that lets the learning change how leaders decide, delegate and handle conflict on a Tuesday. I have set out the four structural causes in what causes leadership development programs to fail. This piece does a different job: it helps you diagnose which failure you have, and shows what to build instead.
Why do leadership development programs fail?
They fail because the learning happens in one place and the work happens in another, and nothing joins them. Three independent sources describe the same gap from different angles, and I use each for its mechanism, not as a benchmark for your company.
McKinsey's 2014 article by Pierre Gurdjian, Thomas Halbeisen and Kevin Lane, drawn from conversations with hundreds of chief executives, names four common mistakes: overlooking context, decoupling reflection from real work, underestimating mind-sets, and failing to measure results. On the second it says that even after very basic training sessions, adults typically retain about 10 percent of what they hear in classroom lectures, versus nearly two-thirds when they learn by doing. On the fourth it says too many evaluations begin and end with participant feedback, which tempts trainers to deliver a syllabus that pleases more than it challenges.
Harvard Business Review's 2016 article by Michael Beer, Magnus Finnström and Derek Schrader opens with the scale of the problem. It reports that American companies spend $160 billion on employee training and education, and close to $356 billion globally in 2015, without getting a good return, and that for the most part the learning does not lead to better organisational performance, because people soon revert to their old ways. The figures cover training generally, not leadership programs alone, and the rest of the article sits behind a paywall, so I rely only on that opening.
A 2024 paper by Jaason Geerts in Behavioral Sciences adds the research view. It estimates that organisations invest around USD 60 billion a year in leadership development, says the workplace application of learning is typically low, and notes that many programs underperform or fail. Its framework draws on four literature reviews and sets out 65 strategies that apply at the foundation of a program and before, during, at the end of and some time after it. The two spending figures measure different things (all training in 2015, and leadership development a year), so do not add them up. What they share is the diagnosis: the failure is mostly in what happens around the learning.
Why do feedback scores hide a failing program?
Feedback scores measure the experience of attending. They tell you whether the room was well run, the facilitator credible and the lunch adequate. They cannot tell you whether a leader now behaves differently, because that happens weeks later, in a meeting the facilitator never sees. McKinsey makes the same point from the supplier side: when evaluation ends with participant opinion, trainers learn to give people what pleases them. A program can therefore score well, be renewed on the strength of the scores, and change nothing. I would stop using satisfaction as the headline measure and relegate it to what it is, a check on logistics.
There is a second effect. Because the scores are good, nobody has a reason to look at the surroundings: the bosses, the rewards, the decision rights. The program gets credit for effort and none of the blame for the missing result. That is how a failing program survives for years.
How do you tell what is wrong with your leadership development program?
Symptoms are easy to see and easy to misread, so I pair each with the cause I would check first, a test you can run this week and the first fix. The pairings are my own judgement drawing on the sources above and on my work with leadership teams, not a validated diagnostic. Test them against your own evidence.
| What you see | What it usually means | A test you can run this week | First fix |
|---|---|---|---|
| Strong feedback scores, no visible change in how leaders work | Evaluation stops at the participant's reaction | Ask three bosses of past participants what the person does differently | Agree the behaviours you expect to change, and who will observe them, before the next cohort |
| Leaders can explain the model and still ignore it in meetings | Learning is decoupled from real work | Check whether any participant applied it to a live business decision during the program | Tie each participant to a real project with a business owner |
| The content feels generic to people in different roles | The program ignores context, with a long list of competencies | Count the capabilities the program develops | Cut to the few that the strategy actually depends on. McKinsey suggests two to three |
| Delegation is taught and controlling behaviour continues | The mind-set underneath the behaviour was never addressed | Ask a participant why they still take the decision back | Add coaching on the assumption behind the habit, not more instruction on the skill |
| Participants return to bosses who have not changed | No sponsor and no reinforcement in how the business runs | List what promotions and bonuses last rewarded | Brief and equip the bosses, and change one reward or decision right alongside the program |
| The same program is rerun every year for a new cohort | Nobody diagnosed who needs what | Ask what capability gap each seat was filled to close | Run a diagnostic first and fill seats by gap |
| HR cannot show the board anything beyond attendance | No baseline was taken | Look for any measurement from before the program | Take a baseline now, then re-measure at six to nine months |
| The executive team is exempt | The top contradicts what the program teaches | Ask whether the executives have done what they ask of others | Start with the top team's own operating habits |
Read the table as a sorting exercise. If most of your answers land in the first four rows (evaluation, application, context and mind-set), the program's own design and follow-through need work, and a redesign will help. If they land in the last four, the program is not the issue. The surrounding structure is, and a better program will just fail with better production values.
What do you build instead of another leadership development program?
Build the system the program is supposed to feed. In my work that sequence is the 5D Leadership Architecture: Discover, Desire, Define, Deliver, Drive. Most programs start at Deliver. They book the workshop before anyone has diagnosed the constraint or tied the learning to the strategy. The training is often fine. The sequence is the problem, which is why the energy has gone by the following Monday. The framework page is at the 5D framework.
- Discover: Diagnose before you design. Establish which capabilities are thin, where decisions pile up on too few people, and what the strategy demands. A 360 and a structured diagnostic give you a baseline to compare against later.
- Desire: Anchor the program to where the business is going. Name the two or three capabilities that matter for the next eighteen months, instead of a general picture of good leadership.
- Define: Write down the decision rights, rhythms and pathways the learning will run through. A leader who learns to delegate needs a published map of what they can delegate.
- Deliver: Attach development to live work: real projects, coaching on the assumption behind a habit, and recurring rituals. The classroom is a part, not the whole.
- Drive: Make leaders responsible for developing other leaders, and measure whether capability is compounding, for example whether your best people have stopped being single points of failure.
The practical difference shows up in what you buy and how long you stay involved. The comparison below sets the two models side by side. It is my own working distinction.
| Dimension | A program as an event | A program inside an architecture |
|---|---|---|
| Starting point | A course or workshop | A diagnosis of the capability gap |
| Unit of change | The individual participant | The participant and the system around them |
| Link to the business | Generic content on good leadership | Two or three capabilities the strategy depends on |
| Who else is involved | Participants and facilitators | Bosses, peers and the executive team as sponsors |
| Where the learning lives | Slides and a binder | Decision rights, rhythms and live projects |
| How you judge it | Satisfaction at the end | Behaviour change at six to nine months, plus a business measure |
| After the program | Energy fades | The next cycle starts from what the measures show |
Here is an illustrative scenario, invented for this article and not a client. Suppose a CHRO inherits a program that has run for three years with strong scores. She asks three questions: which decision is now made differently, who sponsors each participant, and what was measured before the first cohort. The answers are none, nobody and nothing. She does not cancel the program. She pauses recruitment for the next cohort, runs a 360 and a short diagnostic on the current top team, picks two capabilities the strategy depends on, and rewrites the brief so that each participant leaves with a live project, a named boss who has been briefed, and a behaviour that someone else will observe. The content barely changes. What changes is everything around it.
How do you measure whether a leadership development program worked?
Measure before, during and after, and measure behaviour as well as reaction. Geerts recommends collecting data from multiple sources at the outset, throughout, at the conclusion and six to nine months after the intervention. McKinsey describes one approach: a 360-degree feedback exercise at the beginning of a program, followed by another after six to 12 months, so that you assess the extent of behavioural change instead of relying on participant opinion.
Choose measures you can observe. A shift in how often a leader's decisions get reopened. The share of a leader's decisions made by their team without escalation. A manager's own team reporting a changed pattern. Then connect one of them to a business outcome the sponsor cares about. I have not found a measure that works without a baseline, which is why the first fix in the table is so often the same. If you are a CHRO building this case for the board, how CHROs prove leadership development is changing business performance and why most companies cannot measure leadership development ROI go deeper, and how to measure leadership capability across an organisation covers the capability baseline.
What should a CEO or CHRO ask before renewing a leadership development program?
Put these questions to whoever runs the program, in writing, before the next budget is released. The answers tell you whether you are renewing a system or a habit.
- Which decision, in which role, is now made differently because of the last cohort, and who saw it?
- What capability gap was each seat filled to close, and who diagnosed it?
- What changes in how the business rewards or decides, alongside the learning?
- What did we measure before the program started, and what will we measure at six to nine months?
- Have the executives done what they are asking participants to do?
- What will we stop doing if the evidence shows no change?
If the honest answers are thin, the fix is not a better supplier. It is a short diagnostic that tells you where capability is missing and what the program should be for. The wider reasons capability does not keep pace with growth are in why leadership capability does not scale with company growth, the systems a scaling company needs are in what leadership systems fast-scaling companies need, and the pipeline view is in how to build a leadership pipeline in a growth-stage company. When quality is uneven across the business, what CEOs and CHROs should do is the place to start.
When the question is whether to buy coaching or build the system around it, executive coaching or leadership systems sets out the trade. The free Leadership Capability Diagnostic is a sensible first step, the leadership assessments give you a baseline across a team, and CapabilityAI is the companion for the learning that follows each session. For teams that want the architecture designed with them, that is the work inside the Architecture Accelerator.
A leadership development program is a delivery mechanism, not the system
The best program I could design would still fail inside a business that rewards the old behaviour, hides its decision rights and never measured where it started. Training transfers information. Development changes what a leader does under pressure when nobody is coaching them in the moment, and only a system makes that likely. Stop asking whether the program was good. Ask what the business now does differently, and build the structure that makes the answer yes.