What a decade of research changed our minds about

Some things we believed at the start turned out to be wrong

When you spend a decade studying how leaders actually change, and then a further stretch of years testing the findings against real engagements, you accumulate a quiet record of the things you once believed and no longer do. Not dramatic reversals. Mostly corrections. A position held with confidence early on, softened by enough contrary evidence that holding it became untenable.

This is a field note rather than an argument. I want to set out, plainly, the assumptions about leadership development that the research and the practice forced us to revise or abandon. Some we held ourselves. Some we inherited from the wider field and only later learnt to mistrust. None of this is a claim to have arrived. It is a record of changing our minds, which is the only honest thing a research tradition can offer.

We believed insight produces change. It mostly does not.

The first assumption to go was the most seductive one, because it flatters everyone involved. The premise is that if a leader can be brought to a clear, honest insight about themselves, the behaviour will follow. Show them the pattern. Help them name it. The naming will do the work.

It rarely does. A decade of watching leaders arrive at genuine, often moving, insight about their own patterns taught us that insight is the start of the work, not the result of it. A leader can see, with real clarity, that they avoid conflict because they need to be liked, and continue avoiding conflict for years afterward. The seeing and the changing are different events, separated by a great deal of effort, and the gap between them is where most development quietly fails.

What we revised this to is closer to: insight opens a door that then has to be walked through, repeatedly, under conditions that make walking through it hard. The insight is necessary. It is just not the change. We wrote about the mechanism behind that gap in what the research says about lasting leadership change, and the short version is that behaviour holds when it becomes part of identity, not when it becomes visible to the conscious mind.

Consider a finance director at a manufacturing group who, in an early session, articulated his controlling tendencies with unusual precision. He described exactly how he took decisions back from his team, why he did it, and what it cost them. It was one of the more lucid pieces of self-observation I had heard. And for the next several months, nothing changed. He saw the pattern with total clarity and ran it anyway, because seeing a pattern and being able to stand somewhere else when the pressure arrives are two separate capabilities. We had assumed the first delivered the second. It does not.

We believed more content helps. Often it gets in the way.

The second assumption was quieter and more institutional. It is the belief that a richer programme, more models, more frameworks, more sessions, more reading, produces a better developed leader. It is the assumption that built an entire industry, and we were not immune to it.

What the practice taught us is that content has sharply diminishing returns, and at some point turns negative. A leader carrying twelve frameworks does not lead twelve times better than a leader carrying one they have genuinely integrated. More often the abundance becomes a way of avoiding the harder, narrower work. There is always another model to learn, which is more comfortable than sitting with the one uncomfortable thing the existing models have already surfaced.

We revised our view to something almost austere. The constraint in most leadership development is not a shortage of content. It is a shortage of integration time, honest feedback, and the willingness to stay with a single difficult truth long enough for it to change something. Adding content to a leader who has not integrated what they already have is not generosity. It is avoidance, dressed as rigour. Our 4D method is deliberately thin on content for exactly this reason: it sequences depth over coverage.

We believed skills transfer on their own. They transfer only when they are owned.

The third correction is the one with the most evidence behind it, and the one we now treat as close to settled. The assumption is that a skill taught in a development setting will transfer to the job, provided it was taught well. Train the coaching conversation, and the leader will coach. Train the difficult feedback, and the feedback will happen.

Transfer turned out to be far more conditional than the field assumed. A skill demonstrated competently in a workshop, where the leader feels safe and observed and supported, frequently does not survive contact with a hard Tuesday afternoon. The skill was real. It just was not the leader's, in the sense that mattered. It sat alongside who they were rather than inside it, and under load the self falls back on what is inside, not on what is alongside.

A regional sales lead at a logistics business illustrates the shape of this. She left a development programme genuinely able to run a coaching conversation. Her practice sessions were strong. Back in the role, during a soft quarter, the coaching stopped completely. She reverted to directive instruction within a fortnight, not because she had forgotten how to coach but because coaching had never become how she understood her own leadership. It was a technique she deployed when conditions allowed, and conditions stopped allowing it. We used to call that a transfer failure. We now call it an integration failure, which points at a different and harder remedy.

This is the distinction that eventually organised our whole approach, and it is why we argue that most programmes are training the wrong half. The half that holds under pressure is not the half most syllabuses are built around.

We believed character is fixed. The research says it is not, but it is slow.

This is the correction I am most reluctant to state simply, because it is so easily misread in either direction. Early in the work, the working assumption, mine included, leaned toward a kind of fatalism about character. You could develop a leader's skills and judgement, the thinking went, but who they fundamentally were, their relationship to authority, their tolerance for discomfort, their default under threat, was largely set.

The research did not support that. What it supported, instead, was a more careful and less convenient position: character is not fixed, but it changes on a timescale and through a mechanism that most development is not built to accommodate. It does not respond to a two-day offsite. It responds to sustained, repeated engagement with one's own patterns, under conditions of genuine safety, over months and sometimes years. It is slow, it is non-linear, and it cannot be rushed without producing defensiveness instead of growth.

So we abandoned the fatalism, and we also abandoned the opposite error, the cheerful claim that anyone can become anything with the right programme. Both are wrong. Character moves. It moves slowly, it moves in response to specific conditions, and the rate cannot be set by the sponsor's budget cycle. We had to revise not whether character develops but how patient an honest account of it has to be.

A chief operating officer in a professional services firm sat across a roughly two-year arc with us. Early on, his default under pressure was to armour up: to manage perception, to control the narrative, to keep his team at a careful distance. None of that shifted in the first six months. It shifted in the second year, gradually, and it held. Had we measured at month four and concluded that his character was fixed, we would have been wrong by exactly the margin the slow timescale demands. The change was real. It was just not on the schedule the field tends to assume.

We believed we could measure what mattered early. We were measuring the wrong moment.

A quieter assumption, but a consequential one, concerned timing. We believed, as most of the field does, that you could measure development shortly after an intervention and learn something true about whether it had worked. Move the 360, run the post-programme survey, read the result.

The trouble is that the measurable moment and the meaningful moment are not the same moment. Skill-additive change and identity-integrated change look identical in the weeks after a programme. They diverge later, specifically under pressure and reduced observation, which is precisely when no one is running a survey. So the field measures at the moment of maximum ambiguity and treats the reading as a verdict. We did this too, for longer than I would like.

What we revised this to is uncomfortable for anyone who has to justify a budget: the honest evaluation window for whether development held is six to 18 months out, observed under real conditions, not 30 days out on a feedback form. The early measurement is not useless, but it answers a smaller question than the one everyone wants answered. It tells you the skill was acquired. It does not tell you the leader changed.

What we did not change our minds about

It would be dishonest to make this only a catalogue of reversals, because a few convictions survived the decade intact and, if anything, hardened. The core one is the inside-out principle: that lasting behaviour is the downstream expression of who a leader is, and that development working only on the outside produces change that lasts until the next hard quarter. Everything above is, in a sense, a series of corrections in service of that one durable finding.

We also did not change our minds that both paths matter. The skills, the decisions, the observable behaviours of the role are not optional, and we have never argued that inner work alone is sufficient. Self-knowledge without the capability to act on it is, as we have put it elsewhere, expensive self-knowledge. The conviction that held is that the two have to move together, which is the whole reason our model is built around the pairing rather than around either half.

And we did not change our minds about credit. The instruments we use to surface this material are not ours, and a decade has only deepened the view that saying so plainly matters. The Five Lens Development Platform is Ennea International's. The future-readiness assessment is Tomorrows Compass's, and we are a licensed measurement partner, not its author. Our own contribution is the DUAL model and the 4D method, and the discipline of being exact about which is which is part of the rigour, not separate from it.

How a decade of corrections shows up in the work

The accumulated effect of all this is a practice that looks, from the outside, almost too modest. We resist adding content. We are sceptical of early measurement. We treat insight as a beginning rather than an outcome. We pace the inside-out work to the leader rather than the calendar, and we are honest that character moves slowly. None of this makes for an exciting brochure.

But it is what a decade of evidence actually licenses us to claim, and no more. The discipline of the work is partly a discipline of subtraction: removing the assumptions that flattered us, the content that reassured us, and the measurements that confirmed what we wanted to believe. What is left is narrower and, we think, truer.

Three questions this leaves you with

If you are responsible for developing leaders, the corrections above translate into a small set of questions worth asking before the next programme begins.

Are we mistaking insight for change? If the design assumes that helping leaders see their patterns will shift them, it is built on the assumption we abandoned first.

Are we adding content to avoid depth? If the answer to disappointing results has been a richer curriculum rather than deeper integration, the diagnosis may be backwards.

Are we measuring the right moment? If success is judged at 30 days, the verdict is being read at the moment of maximum ambiguity.

If those questions land, our assessments are a sober starting point for seeing where a leader actually stands, and the services and use cases pages describe how this calibrated, deliberately unhurried work takes shape in practice. None of it is dramatic. A decade of research has mostly taught us to be suspicious of development that is.

A closing note on changing one's mind

The reason to publish a piece like this is that a field tradition earns trust not by being right at the start but by being honest about what it revised. We got several things wrong, corrected them against evidence, and kept the few convictions that survived. That is what a research-led practice is supposed to look like, and it is the standard we would want any organisation to hold us to.

If you are weighing whether to invest in leadership capability and you would rather work with people who will tell you what they have abandoned as well as what they believe, start a conversation with us.

The leaders described here are representative composites drawn from patterns we observe in practice, not identifiable individuals.