We’ve spent decades trying to teach people how to be better leaders.
We’ve built leadership programmes, run workshops, created competency frameworks and developed increasingly sophisticated ways of assessing people. And yet I think we’ve missed something fairly fundamental.
We’re still not very good at measuring whether leadership development actually works.
The numbers are telling.
90% of leaders say soft skills are more important than technical skills in 2026. Yet fewer than half of organisations offer leadership training at all levels of seniority.
At the same time, 64% of young professionals want to progress into management or leadership roles.
So there’s clearly an appetite for leadership development. People want it. Organisations know it matters.
But here’s the statistic that really interests me:
62% of businesses measure leadership effectiveness using employee surveys.
I find that difficult to square with the importance we’re placing on leadership skills. If communication, coaching, active listening and strategic vision are genuinely critical leadership skills, why are we mostly measuring them by asking people what they think?
Imagine if marketing measured the success of a campaign purely by whether someone had opened an email. We’d never accept that. Yet, in leadership development, we still do something remarkably similar.
The traditional approach to leadership development is pretty predictable.
Someone attends a workshop. They complete some learning. Perhaps they do an assessment afterwards. Their manager gives them some feedback. Then, six or twelve months later, we ask everyone how things are going.
The problem is that none of these vanity metrics tell us very much about what actually happens in between.
Course completion measures activity, not impact. A survey measures perception, not necessarily performance. And managers, however good they are, only see a fraction of what their people do. They can’t be everywhere at once, and if a senior leader is overseeing the work of 10 different managers, they will only see a tiny sliver of each manager’s actual performance and workplace behaviours.
Instead of happening in neat, time-boxed assessments or observation windows, leadership happens in conversations. It happens in the one-to-one where someone has to give difficult feedback. It happens when a manager listens to an employee who is struggling. It happens when a leader has to communicate a new strategy and get people behind it.
Those are the moments where leadership skills become visible. But traditionally, HR doesn't get to see them, meaning this lack of visibility creates a huge learning measurement problem.
There’s another reason I think this matters.
We often talk about leadership development as though people become leaders at a particular moment, whether they’ve been promoted, completed a leadership course or received a new job title.
But leadership actually starts happening long before any of those things. ‘When’ someone officially becomes a leader is pretty arbitrary. Anyone who’s found themselves leading a project, mentoring a colleague or managing an informal project team can quickly find themselves adopting the role of an accidental leader before they’ve been officially recognised as one. By the time they appear on an HR spreadsheet as a manager, they've probably been demonstrating leadership skills for months.
Traditional measurement struggles with this because it relies on snapshots or artificial assessments. But as we know, leadership happens continuously – even if someone doesn’t yet know they’re performing in a leadership capacity.
I think AI changes the equation. That’s not because AI is going to replace managers. It won't.
And it’s not because we suddenly need another piece of software telling leaders how to lead.
The interesting thing about AI is much simpler: it can observe patterns in work that humans simply don't have the time or opportunity to observe. Those overstretched senior managers can’t have eyes on every manager every moment of every day, so they make their performance observations and decisions based on very limited information.
With AI skills intelligence, we can start measuring skills in the flow of work. A simple, routine work meeting suddenly becomes a source of real evidence of behaviours and the skills leaders demonstrate, without the need for human observations.
We can look at how someone communicates, how actively they listen, how they coach others, how clearly they articulate a strategy and other behaviours that are visible in normal conversations – and crucially, we can do that continuously, in a way that scales seamlessly whether you’re developing 10 leaders or 10,000.
Instead of asking a leader, ‘How good do you think you are at listening?’, we can start saying, ‘Here's what we're seeing in your conversations. Here's where you're strong, here's where there might be a growth opportunity and here's what you could work on, and here are the examples from your real leadership moments that demonstrate it.’
That's much more useful. It's also much more personalised. One leader might need to work on active listening. Another might need to become better at coaching. Someone else might be excellent at both but struggle to communicate strategic direction. So why would so many organisations send all three people on the same leadership course?
This is the bit I find most exciting. For years, the model has essentially been:
Annual workshop → annual survey → manager review → repeat.
It's an event-based approach to something that isn't an event.
AI gives us the opportunity to move towards:
Every meeting → continuous measurement → personalised coaching → continuous improvement.
That's a pretty fundamental change. It means we don't necessarily need to take a busy leader away from their work for a whole day to develop their communication skills. Instead, we can help them develop those skills while they're actually communicating.
That's what I mean by moving from leadership training to leadership development. While training is something you attend, development is something that happens all the time, often without even realising.
This shift is the driver behind Helix, our AI skills intelligence platform.
Helix joins online meetings, tracks and measures skills in the flow of work and gives leaders scores and personalised feedback. That feedback can then be used to support further development and AI coaching as part of a genuinely useful feedback loop:
Measure → understand → coach → improve → measure again.
That's how I think skills development should work. And because AI can do the measurement continuously, it gives us something we've never really had before: an objective view of how skills are developing over time.
To be clear, I’m not saying we should be replacing the human side of leadership development with AI. Humans will always have an important role to play, whether that’s coaching, behaviour modelling or quick, in-the-moment feedback from someone you respect (‘Nice work on the presentation!’ can go a long way).
But AI absolutely has a role to play in the leadership development of the future. Specifically, making it measurable at scale. After all, if we're serious about developing better leaders, we need to know whether people are actually performing better.
For years, we've mostly measured whether they turned up. I think it's time we measured what happened afterwards.