70% of Your Leadership Training Budget Is Wasted Within Months. Here’s the Proof.

Organizations invest billions in leadership training every year, and within weeks, most of it is gone. Research consistently shows that without reinforcement, professionals forget 70% or more of training content within a month of delivery. For leadership skills specifically, where the goal is not knowledge recall but durable behavioral change, the decay curve is even more unforgiving. Understanding how and why leadership skills decay, and which development architectures are built to prevent it, is one of the most consequential questions an L&D team can ask. The answer determines not just ROI, but whether any of the development investment actually shows up in how managers lead.

Ebbinghaus Was Right: and the Problem Is Worse Than Most Teams Think

The forgetting curve was first documented by psychologist Hermann Ebbinghaus in the 1880s. His research showed that memory decays rapidly in the absence of reinforcement: roughly half of new information is forgotten within an hour, and up to 70% within 24 hours. Without spaced repetition, the curve continues downward.

What is less understood is how this curve applies specifically to leadership development. Ebbinghaus studied syllable memorization — a closed, declarative task. Leadership skills are open, procedural, and context-dependent. They are harder to acquire and, critically, they decay through a different mechanism: not through simple forgetting, but through displacement by ingrained behavioral habits.

A manager who learns a new framework for structuring difficult feedback conversations does not simply forget it over time. They revert to the habitual communication patterns they have used for years — patterns that are deeply embedded in procedural memory. The new skill, never sufficiently practiced to displace the old habit, quietly disappears under the pressure of real work.

This means the forgetting curve for leadership skills is not primarily a memory problem. It is a habit-formation problem. And it requires a fundamentally different solution.

Six Training Approaches, Six Very Different Retention Curves

Research from meta-analyses by Lacerenza et al. (2017) on leadership training effectiveness, Arthur et al. (1998) on skill decay, and Tatel & Ackerman (2025) on open versus closed skill retention provides a clear basis for comparing training approaches on their long-term retention outcomes. The differences are dramatic.

Training Format	30 days	90 days	12 months
E-learning / passive content	~20%	10–15%	~10%
Single intensive workshop	~35%	~25%	~20%
Spaced workshops (no practice)	~45%	~30%	~25%
Workshops + practice → then stops	~60%	~45%	~30%
Workshops + ongoing spaced practice	~75%	~70%+	65–97%

Retention ranges informed by: Lacerenza et al. (2017), Arthur et al. (1998), Tatel & Ackerman (2025), and Research Institute of America (2022). Figures represent approximate retention of training impact, not knowledge recall alone.

The pattern is consistent: formats built around passive content delivery: e-learning modules, single workshops, even well-structured cohort programs without ongoing practice, experience steep decay within the first 90 days. Formats that build in ongoing, spaced practice maintain substantially higher retention over the same period and continue to compound over time.

The Research Institute of America’s findings are particularly striking for L&D practitioners: participants in traditional training sessions retained just 10 to 20 percent of information after three months. That is the baseline most corporate programs are working from — and most do not measure it.

Why Practice Changes the Curve: Procedural Memory Formation

The mechanism behind this divergence is procedural memory, the type of memory that governs skilled performance under pressure. When a leader practices giving difficult feedback repeatedly, in realistic scenarios with calibrated feedback, the skill shifts from declarative memory (knowing what to do) to procedural memory (being able to do it without deliberate effort).

Procedural memory is substantially more durable than declarative memory. It is what allows an experienced manager to navigate a tense performance conversation without consciously consulting a framework, because the skill has been internalized through repetition to the point where it is automatic.

This is why simulation-based practice produces a fundamentally different retention curve. It is not a better version of a workshop. It is a different category of learning: one that builds the kind of memory that survives the pressures of daily work. Research from Ericsson’s deliberate practice framework (2004, 2008) confirms that expert performance in complex, high-stakes domains is built through targeted, high-repetition practice with expert feedback, not through knowledge transfer.

70%+  of leadership training impact is lost within months without ongoing practice (Lacerenza et al., 2017)

The Three-Phase Model That Maintains Retention Long-Term

The research does not just diagnose the decay problem: it points to a practical implementation architecture. The data supports a phased approach that matches practice intensity to the stage of skill development, rather than front-loading all development effort into a single training event.

In the first phase: covering roughly the first six months, the goal is procedural memory formation. This requires combining initial training with biweekly simulation practice: realistic, scenario-based exercises where managers practice specific skills (feedback conversations, coaching moments, conflict navigation) and receive immediate, calibrated feedback. This intensity builds the procedural foundation that makes later maintenance effective.

In the second phase: from months seven through twelve, practice frequency reduces to monthly, but does not stop. The procedural base established in the first phase means less frequent practice still maintains high retention. The goal here is consolidation, deepening the skills already formed rather than building new ones from scratch.

From month thirteen onward: quarterly practice sessions prevent the gradual decay that follows even procedurally-embedded skills over time. The research suggests that once a strong procedural base is established, even relatively low-frequency maintenance keeps retention at levels that no amount of initial training alone achieves. The key insight is that maintenance never stops entirely. Programs that end at month six lose 49 percentage points of retention by month 24 compared to those that continue in maintenance mode.

What This Means for How L&D Teams Design Programs Today

The implications for program design are direct and practical. If the goal is behavioral change that persists beyond the immediate post-training period, and in virtually all organizational contexts, it is, then design choices must be evaluated against their long-term retention profile, not their immediate impact scores.

Three design principles follow from the research:

• Practice is not a supplement to training, it is the primary mechanism of skill formation. Content delivery creates awareness. Practice builds capability. Programs designed primarily around content delivery will decay regardless of how well the content is designed.
• Spacing matters more than intensity. A two-day workshop delivers high intensity over a short window — the worst profile for long-term retention. The same number of practice hours distributed over six months produces dramatically better outcomes. L&D teams should redesign around frequency, not event size.
• Maintenance is not optional. Programs with strong initial phases but no ongoing reinforcement will lose most of their impact within a year. The question is not whether to include a maintenance phase, but how to make it operationally sustainable at scale.

This last point is where most organizations get stuck. Designing a maintenance phase for hundreds or thousands of managers, monthly or quarterly practice sessions with expert feedback is operationally impossible through traditional coaching or facilitation models. The cost and scheduling burden alone makes it prohibitive.

How AI Coaching Makes the Retention Architecture Operationally Viable

The practice-based retention model described above is not a new idea. L&D researchers have understood the mechanics of skill decay and procedural memory for decades. What has changed is the operational feasibility of implementing a high-frequency practice architecture at scale and that change is being driven by AI.

Platforms like Coachello embed practice directly into the tools managers use daily: Slack and Microsoft Teams, making scenario-based practice available on demand, without scheduling friction, coaching cost, or geographic constraints. A manager preparing for a high-stakes conversation can rehearse it with realistic AI-powered Roleplays, receive structured feedback on their communication approach, and iterate, in the flow of their working day, not during a set-aside training event.

This is what makes the continuous and maintenance practice models operationally viable for global organizations. The intensive phase is supported through a combination of workshops and regular AI coaching scenarios. The building and maintenance phases are sustained through lower-frequency interactions that require no additional infrastructure, facilitation resource, or budget cycle.

“I was struggling with giving clear feedback to my team. After a few sessions, I learned how to provide constructive feedback in a way that’s clear and supportive. My team now responds more positively, and performance has improved.”
— Manager, Welcome to the Jungle | +43% motivation | +97% actions implemented

The result is a development architecture that mirrors what the retention research prescribes: high-frequency initial practice, progressive spacing, and ongoing maintenance, without the cost or logistical barriers that have historically made this model impractical below the executive level.

What to Ask Your Current Training Provider

If you are evaluating your existing leadership development programs against the retention evidence, five questions will reveal whether the program design is likely to produce durable outcomes:

• What is the practice-to-content ratio? If the majority of program time is spent on content delivery rather than active practice, the retention curve will be steep.
• What is the feedback mechanism? Peer feedback and self-reflection have limited effectiveness for skill calibration. Expert or AI-generated feedback produces significantly better outcomes.
• What happens after month one? Programs with no structured reinforcement beyond the initial training event will lose the majority of their impact within a quarter.
• How are skills assessed behaviorally? Completion metrics and satisfaction scores do not measure skill retention. Behavioral observation, scenario performance, and manager-reported application do.
• Is there a maintenance phase? If the program has no answer to this question, it is not designed for long-term retention.

Programs that cannot answer these questions clearly are, in structural terms, optimized for event delivery rather than skill development. The forgetting curve will do the rest.

The Bottom Line

Leadership skill decay is not a mystery: it is a predictable consequence of development architectures built for content delivery rather than capability formation. The research is unambiguous: e-learning alone retains roughly 10 percent of its impact at two years. Workshops alone are not much better. The only approaches that maintain meaningful retention are those built around ongoing, spaced, simulation-based practice.

The good news is that the solution is equally clear. Organizations that redesign their leadership development architecture around the three-phase practice model, intensive formation, progressive building, long-term maintenance, consistently outperform those relying on episodic training events, on every metric that matters: behavioral change, team performance, retention, and promotion readiness.

AI coaching platforms like Coachello are making that architecture accessible at scale: turning what the research has prescribed for decades into something operationally real for every manager in an organization, not just the executive few. The forgetting curve is not a law of nature. It is a design failure. And it is one that can be fixed.

Want to see how AI coaching builds the practice architecture that keeps leadership skills from decaying?

Book a demo