Measuring Leadership Development Impact: How to Prove ROI Without Falling for “Happy Sheets”

Why leadership development ROI is so hard to prove

Most organisations are not short on leadership programmes. They are short on defensible evidence that those programmes changed behaviour, improved performance, and delivered value.
The research problem is consistent: evaluation often stops at reaction scores, attendance, or completion rates, then gets presented as “impact”.
That might be convenient, but it does not withstand a serious question from a CFO, a board, or a governance committee.

The core point made in Jaason M. Geerts’ 2024 paper on maximising leadership development impact and ROI is that ROI is not a single metric you add at the end.
It is an outcome of a whole system of choices: what you target, who you select, how you design learning, how you support transfer, and how you evaluate results in the real workplace.
If any of those system components are weak, ROI becomes unmeasurable or unconvincing.
(Geerts (2024), “Maximizing the Impact and ROI of Leadership Development”)

The evaluation trap: measuring what is easiest, not what matters

Reaction data (“participants liked it”) is not meaningless, but it is not impact.
Learning checks (“they understood the model”) are still not impact.
Even self-report confidence (“I feel more leaderly”) can be misleading.
Impact begins when leaders behave differently at work, and that change is linked to outcomes that stakeholders actually care about: team performance, retention, quality, safety, customer satisfaction, or delivery speed.

A useful way to structure this is to treat leadership development as a programme that must be evaluated like any other intervention.
In formal terms, this sits inside the discipline of programme evaluation, where the goal is to assess effectiveness and value-for-money using a clear theory of change, credible evidence, and realistic comparisons.

In practice, many L&D functions default to familiar evaluation models.
The Kirkpatrick Model is the best-known, and for good reason: it forces you to move beyond reaction and learning toward behaviour and results.
Some organisations extend this with an explicit ROI layer using the Phillips ROI Model, which attempts to monetise results relative to programme costs.

The issue is not the frameworks themselves.
The issue is that many organisations use the language of “Level 3” and “Level 4” while collecting evidence that still sits at Level 1.
Geerts’ paper is helpful because it treats ROI as a systems problem and offers a structured set of strategies across the full lifecycle, not a “measure it later” afterthought.
(Geerts (2024))

A board-ready logic chain: from programme design to ROI

If you want leadership development ROI that is credible, you need an explicit logic chain that a skeptical stakeholder can follow.
Here is a clean version you can reuse in governance conversations:

1. Define the business problem you are solving (not “develop leaders”, but “reduce avoidable attrition in frontline teams” or “improve delivery reliability”).
2. Define the target behaviours that plausibly drive that outcome (coaching cadence, feedback quality, decision hygiene, escalation discipline).
3. Establish baseline evidence for those behaviours and outcomes (before the programme starts).
4. Design learning + transfer mechanisms to change behaviour in real contexts (not just workshops).
5. Measure behaviour change with credible data sources (not only self-report).
6. Link behaviour to business outcomes with realistic analytic logic (trend, comparison groups where feasible, contribution not overclaiming).
7. Convert outcomes to ROI where appropriate, including costs, time, and assumptions that can be challenged and refined.

This is the difference between “we ran a programme” and “we ran a measurable intervention”.
It also helps you avoid the classic failure mode where evaluation is bolted on at the end, after key measurement opportunities have already been missed.
(Geerts (2024))

The Geerts lifecycle: what to do before, during, and after

Geerts synthesises evidence into a structured “optimising system” of strategies across stages of leadership development.
You do not need all strategies to improve ROI.
You do need to address the major leakage points: poor targeting, weak transfer, and weak evaluation.
(Geerts (2024))

1) Before: target the right leaders and the right outcomes

ROI improves when programmes are focused on leaders whose behaviour change is likely to matter.
That means clarifying selection logic and readiness, not simply filling cohorts.
Where possible, segment the intervention:
high-impact roles, known capability gaps, and environments where managers have enough discretion to apply the learning.

A practical move: define two to four “critical leadership behaviours” that will be measured for every participant.
Make those behaviours observable, coachable, and linked to business outcomes.
Then capture baseline data for those behaviours before the first session.

2) During: build transfer into the design, not into the slide deck

One of the most consistent themes across learning science is that application beats exposure.
Programmes that depend primarily on content delivery tend to create short-lived insight, not durable behaviour change.
A higher-ROI design usually includes:

• Real-work assignments that must be completed in the participant’s team context
• Coaching support (internal or external) aligned to the target behaviours
• Manager-of-manager reinforcement, so accountability does not vanish after the workshop
• Peer practice with structured feedback, not just discussion

This is also where measurement design matters.
If you want to measure behaviour change, you need to specify what “good” looks like and create repeatable collection points.
For example: a short monthly behaviour checklist completed by the participant’s manager, combined with a self-reflection prompt, and a small set of objective signals (attrition, quality defects, delivery metrics) where relevant.
(Geerts (2024))

3) After: evaluate the change trajectory, not a one-off snapshot

Sustainable behaviour change is rarely instantaneous.
It is a trajectory.
Evaluation that ends at programme completion will tend to under-detect impact (or overstate it with optimism bias).
A stronger approach is to define follow-up points: 30, 90, and 180 days, with a consistent measurement cadence.
This turns evaluation into an operational habit, not a ceremonial report.
(Geerts (2024))

How to measure behaviour change without overcomplicating it

The best evaluation systems are often simple, repeatable, and credible.
Here is a practical measurement stack that tends to work in real organisations:

Level A: Behaviour observation (fast and repeatable)

• Manager ratings on 3 to 5 target behaviours (monthly, 3 minutes)
• Peer feedback on one behaviour where peers can actually observe it (quarterly)
• Self-report reflection used as qualitative context, not as proof

Level B: Operational performance indicators (role-relevant)

• Team attrition, absence, and retention stability
• Quality, error rates, safety incidents, customer outcomes
• Delivery reliability, cycle time, and throughput where applicable

Level C: Programme contribution logic (honest attribution)

Many teams overclaim by treating correlation as causation.
A better stance is contribution:
“Given these baseline trends, these exposure levels, and these changes in behaviour, what is the most plausible contribution of the programme?”
This mindset aligns with programme evaluation best practice and avoids credibility damage when stakeholders challenge assumptions.
(Programme evaluation)

If you need a recognised evaluation language for internal governance, you can still use
Kirkpatrick’s four levels
as the headline structure, and then add ROI calculation logic selectively using the
Phillips ROI Model
for programmes where monetisation is feasible and worth the extra effort.

The ROI calculation: what to include (and what to avoid)

ROI is attractive because it produces a single number.
ROI is also dangerous because it invites false precision.
A credible ROI story does three things:
it includes the full cost picture, it documents assumptions clearly, and it avoids monetising outcomes that cannot be credibly monetised.

Include these cost components

• Programme design and vendor costs
• Facilitation and coaching costs
• Participant time cost (often the biggest hidden cost)
• Operational overhead for measurement and follow-up

Be cautious with these benefits

• Self-reported productivity gains without operational evidence
• “Culture improvement” claims without defined behavioural proxies
• Financial conversion of soft outcomes without a defensible chain

If you are going to monetise benefits, do it in a way that remains credible under challenge.
Use ranges, scenario assumptions, and transparent sensitivity checks rather than one magic number.
This is aligned with the intent of ROI methodologies that extend beyond satisfaction metrics.
(Phillips ROI Model overview)

What this means for HR and L&D leaders: a practical checklist

If you want your next leadership programme to be “ROI-defensible”, you can apply the following checklist before launching:

• Outcome clarity: Can we name the business outcomes and the target behaviours in one sentence?
• Baseline: Do we have baseline behaviour data and baseline outcome data?
• Transfer: Do participants have real-work assignments, coaching, and managerial reinforcement?
• Measurement cadence: Are we collecting behaviour data at 30/90/180 days?
• Stakeholder credibility: Would a skeptical stakeholder accept the evidence sources?
• Contribution logic: Are we claiming plausible contribution rather than certainty?

These steps map directly onto the systems-oriented stance in Geerts’ paper: ROI improves when evaluation is designed as part of the programme system, not appended at the end.
(Geerts (2024))

How we help: measurement that stands up to scrutiny

If you want leadership development that is measurable, governance-ready, and defensible, you need three capabilities in the same room:
programme design, measurement design, and real-world implementation pragmatism.
This is where independent psychometric and evaluation expertise can prevent expensive “activity without impact”.

Next steps for Leadership development 

If you are building that capability system across corporate and education audiences, you can anchor the skills language and measurement design in a single “skills authority engine” and then route applications to the right environment.
For example, you might define core skills and evidence standards on mosaic.fit, route corporate implementation and governance thinking through your professional practice on Rob Williams Assessment, and translate measurement principles into practical training pathways for schools via SchoolEntranceTests.com.

FAQs

What is the biggest mistake organisations make when evaluating leadership development?

Treating satisfaction as impact. Reaction data can be useful, but it does not demonstrate workplace behaviour change or business results.
Using recognised structures like Kirkpatrick helps, but only if you collect real evidence at the behaviour and results levels.

How soon should we measure behaviour change after a leadership programme?

Measure immediately for early signals, then re-measure at follow-up points (for example 30, 90, and 180 days).
This approach better captures trajectories of change, which is consistent with evidence-led programme evaluation practice.
(Geerts (2024))

Do we need to calculate ROI for every leadership programme?

Not always. ROI calculation can be costly and may not be appropriate for every intervention.
Many organisations use ROI selectively for high-cost, high-visibility programmes and use lighter-touch evidence approaches elsewhere.
The Phillips ROI Model is one recognised approach when monetisation is feasible.

What’s a “credible” data source for behaviour change?

Credible sources are those that reflect real workplace observation or operational outcomes: manager ratings on specific behaviours, peer feedback where peers can observe the behaviour, and objective indicators such as retention stability or quality metrics.
Self-report can add context, but should not be treated as proof on its own.

Related reading 

• RWA: AI literacy readiness
• RWA: AI assessment design
• SET: AI literacy skills training