How to Fix the Reinforcement Gap in Leadership Development

What learning research taught Leadership Development 

Hermann Ebbinghaus’s forgetting curve, first measured in 1885 and confirmed by every cognitive science study since, established that without reinforcement we forget roughly half of new learning within an hour and around 90% within a week. The most thoughtful leadership program in the world is fighting this curve from the moment it ends. 

Modern business data confirms what the curve predicts. The discipline of training transfer, which measures how much workplace training actually changes on-the-job behaviour, has consistently found that only 10 to 20% of training translates into sustained behavioural change without active reinforcement. McKinsey’s ongoing work on capability building has reached the same conclusion from a different angle. The highest-performing organisations are not those that invest most in training events. They are those that have built the daily reinforcement structures around the events [1]. Without reinforcement, training has trouble to stick which impacts how much it can transform workplace culture. 

And this problem is only sharpening. Microsoft’s 2024 Work Trend Index found that knowledge workers face near-constant interruption through the working day, with managers carrying significantly higher cognitive load and digital debt than their reports [2]. Even after a strong learning event, managers return to an operating environment that erases most of what they learned within days. The question is no longer how to train. It is how to make training stick when the working environment is actively pulling against it.

Where the pressure shows up: people managers 

The pressure is most acute in the people manager. Today’s people managers are being asked to achieve more, with leaner teams, against tighter timelines and with greater personal accountability for engagement, retention and performance than the role has ever carried. The remit has expanded without a matching expansion in the support around it. Most of it due to better technological support and AI integration into everyday work. 

Recent commentary on the post-pandemic management problem, including Forbes’ coverage of “quiet managing” as the successor pattern to quiet quitting [3], has highlighted a particular failure mode. Under load, managers become conflict-averse. They avoid difficult conversations. They defer performance feedback. They quietly work around underperformance rather than address it. The root cause is the absence of the skills required to engage employees and address underperformance. 

Leadership development programs typically deliver these skills in a workshop or a coaching session. But they do not deliver them on a Tuesday afternoon just when the manager has ninety seconds to decide whether to address a behaviour with a team member or let it slide. The gap between the one yearly workshop where the skill is learned and the moment where it needs to be applied is too vast and this is where the leadership development might fail.

The high value of human coaching 

Human executive coaching remains one of the most powerful leadership development interventions available. The depth of a one-to-one conversation with a skilled coach, the safety to surface what is going on underneath the role, the developmental insight that reshapes a leader’s understanding of themselves: none of this can be reproduced by software. The strategic value of human coaching is irreplaceable and any organisation serious about leadership capability should be investing in it. 

Many skilled coaches typically begin with a diagnostic tool such as the Team Management Profile (TMP), a psychometric profile that maps how a person prefers to work across eight role preferences. A coach can take a leader well beneath the surface of their day-to-day behaviour and into the specific preferences shaping their decisions, their communication and the kind of contribution they prefer to make. The TMP, or another profile, gives the coach a precise vocabulary for the leader’s distinctive way of operating, and gives the leader language for patterns they may have spent years sensing without ever being able to name. 

Coaching might deepen awareness further with a 360-degree view. 360-degree leadership assessment captures how a leader is experienced by their direct reports, peers and managers. Those tools surface the gap between how the leader sees their own leadership and how the system around them experiences it. That gap is where emotional intelligence grows fastest. It is also where capability gaps surface with the backing of data. The personal growth plan that comes out of this work is grounded in evidence as well as the coach’s experience and thorough understanding of a leader. 

These are the sessions that produce the moments leaders remember years later. They carry strong realisation moments where they might discover that they work especially hard in one area because this is where they differ most from the norm, it might reframe their discourse “I had no idea I came across that way” or validate the strengths they have been operating with didn’t have a name for. This is the eye-opening, reframing work that human coaching does and that nothing else can match. 

But human coaching has a structural limitation. A leader who meets with their coach once a fortnight gets roughly one coaching hour out of every 80 working hours. The other 79 hours are where the behaviour change need to happen. And while every executive coaching conversation includes some version of “how to practise this.” Development work between sessions can still feel cryptic when developing new skills. And it is there that leaders might not get the reinforcement they need. 

Scaling internal reinforcement across organisation 

Most organisations already know that a single workshop, seminar or short program will not hold, so they build their own reinforcement around it. They often develop an internal capability, such as an L&D function or a group of internal coaches, who carry the shared language and the development models into the everyday. They might coach the first-line and middle managers and wider teams. Through this work a common vocabulary takes root and development becomes part of how the organisation works. 

That approach asks a great deal though and is not within every business’ reach. Building and sustaining an internal coaching capability takes people, time and budget that a large enterprise can absorb but a medium-sized business often cannot. A leaner HR team may be a handful of people holding recruitment, wellbeing, compliance and culture all at once, with little spare capacity to coach every layer of the organisation through their development journey. For them, the cascade model is more aspiration than option. 

And even where the capability does exist, it can only stretch so far. The most responsive and most skilled L&D team cannot be in every corridor and on every call. They are spread across many departments and competing priorities and they cannot answer every question in the moment a manager actually has.

Where AI coaching changes the picture 

AI coaching, used well, is not a replacement for human coaching. It is the reinforcement layer that can embed learning. The unique value of an AI coach is the property no human coach can offer at scale: presence. An AI coach is available in the ninety seconds before a difficult conversation, in the five minutes after a tense meeting, on the morning before a hard week. It can answer questions in the moment the question actually matters. 

But the quality range across AI coaching tools is wide. A general-purpose chatbot will produce general-purpose prompting, with associated risks. What modern technology now allows, and this is the part that changes the conversation, is for an AI coach to be trained on the same psychological models the human coach is using in a workshop. A coaching AI grounded in a validated, evidence-based framework of work and leadership behaviour produces something else entirely: prompts that line up with the developmental work the human facilitator has already begun, in the language the leader has learned to think in.  

Our own TeamOS, built on more than forty years of theory and data behind the TMP psychometric profile of work preferences, is one example of what it looks like when the scientific foundation is taken seriously. Organisations choosing an AI coaching tool should be asking what its psychological underpinning actually is, because the gap between a chatbot and a coaching AI grounded in real science is the gap between activity and development. 

For people managers facing the conflict aversion that produces quiet managing, this matters a lot. The development work of becoming a leader who addresses difficult behaviour is not done in a single session. It is done over a course of time spanning dozens of small moments where the manager either practises the new behaviour or defaults to the old one.  

An AI coach that offers the right micro support in those moments, grounded in the same diagnostic the human coach used to identify the gap, is doing the reinforcement work that no human coach, or L&D team can fit into a yearly seminar or even an extensive program with a fortnightly cadence.

The partnership between human and AI

The most effective leadership development programs are beginning to combine these two layers deliberately. The human coach provides depth, insight, accountability and the safety to do real developmental work. They highlight the pattern, they shape the program and they hold the leader accountable to the work. The AI coach provides repetition, immediacy and the daily prompting that turns insight into habit.  

For organisations attempting culture transformation at scale, where the goal is for thousands of small daily behaviours to shift in a consistent direction across hundreds of managers, the reinforcement layer is no longer optional. It is the mechanism by which the transformation actually happens. While before the solution was to have extensive internal training to facilitate session, reinforcing the language and models of development used, AI can alleviate some of that burden. 

Pairing human coaching with AI reinforcement, both anchored in the same scientific foundation, is how embedded development investment can be achieved without extensive internal coaching capability.

Without repetition, there is no learning

The Ebbinghaus curve has not changed in 140 years. What has changed is our awareness of it and our answers to the need for repetition of learning.

The best leadership development programs pair depth and reinforcement, deliberately and at scale, grounded in psychological models robust enough to support both layers. For the first time in the history of corporate learning, the reinforcement gap has an affordable and credible solution thanks to scientifically-grounded AI coaching which can be used in support to humans to achieve sustainable cultural transformation. 

References

[1] McKinsey & Company, State of Organizations work (2023-2024) https://www.mckinsey.com/capabilities/people-and-organizational-performance/our-insights 

[2] Microsoft , Work Trend Index 2024: AI at Work Is Here. Now Comes the Hard Part. https://www.microsoft.com/en-us/worklab/work-trend-index/ai-at-work-is-here-now-comes-the-hard-part 

[3] Forbes, Quiet Quitting Isn’t the Problem. Quiet Managing Is (2026) https://www.forbes.com/sites/robertamatuson/2026/02/17/quiet-quitting-isnt-the-problem-quiet-managing-is/ 

Ebbinghaus, H. (1885). On Memory. Original research on the forgetting curve, confirmed and refined consistently by modern cognitive science. 

Margerison, C. and McCann, D. (1990, revised editions). The Team Management Profile and the Margerison-McCann Team Management Wheel. Team Management Systems.