Evaluating an Executive Coaching Program: Metrics, ROI and Limits

Sooner or later, someone asks whether the coaching budget is paying off. It is a fair question. Executive coaching is a significant investment, and HR and people teams need to show that it is being used well. The trouble is that the easiest numbers to produce, such as satisfaction scores and enthusiastic quotes, say little about impact, while the most impressive numbers, such as precise return-on-investment multiples, are often built on shaky assumptions.

This guide sets out a practical, honest way to evaluate executive coaching. It borrows from two established training-evaluation frameworks, draws on what researchers say about measurement problems in coaching studies, and shows how to protect confidentiality while still gathering useful evidence.

Decide what success means before the first session

Evaluation starts at contracting, not at the end. If nobody agrees on what should change, there is nothing meaningful to measure later. In the opening agreement among leader, sponsor and coach, try to answer three questions:

1. What specifically should be different if coaching works? Name behaviors, not traits. "Delegates the quarterly forecast to her team" beats "becomes more strategic."
2. Who would notice? Identify the people best placed to observe the change, such as a manager, direct reports or peers.
3. What business result might follow, and how directly is it connected? Be realistic about how much one leader's behavior drives the number.

Record a baseline for each where you can, even if it is just the manager's and the leader's own description of where things stand today.

Use a layered framework

The Kirkpatrick Model, created by Donald Kirkpatrick in the 1950s for evaluating training, offers four levels that translate reasonably well to coaching:

• Reaction: how favorable, engaging and relevant the leader found the experience.
• Learning: what knowledge, skills, attitudes, confidence and commitment the leader gained.
• Behavior: whether the leader actually applies the critical behaviors on the job.
• Results: whether the targeted organizational outcomes occur.

For coaching, the level that deserves the most attention is usually behavior. Reaction data tells you whether leaders liked their coach, which matters for fit but not for impact. Results data is valuable but hard to link to one person's coaching. Behavior change, observed by others over time, is often the most credible middle ground.

If you want ROI, do it properly

The ROI Methodology from the ROI Institute adds a fifth level, return on investment, which compares monetary benefits with costs. It defines ROI as net program benefits divided by program costs, expressed as a percentage, and lists steps that include isolating the improvement attributable to the program, converting it to monetary value, calculating fully loaded costs and identifying intangible benefits that are not converted to money. It also describes taking a conservative approach.

Each step involves judgment. Isolating the effect of coaching from a new market, a reorganization or a strong team is difficult, and the techniques the institute lists, such as control groups, trend-line analysis and estimates, vary in rigor. Converting a leader's better meetings into dollars usually relies on estimates.

That does not make ROI useless, but it does mean an ROI figure should travel with its assumptions. If you present one, show how you isolated the effect, who made the estimates and what you left out. Peer-reviewed meta-analyses of coaching, such as those by Jones, Woods and Guillaume and by Nicolau and colleagues, report effect sizes rather than a validated general ROI multiple, so be wary of borrowing headline ROI figures from elsewhere.

Gather evidence from more than one source

The research literature offers a cautionary tale here. In their 2016 systematic review in PLOS ONE, Simmy Grover and Adrian Furnham found that coaching studies rely heavily on self-report and warned that self-report measures are not appropriate for assessing objective organizational outcomes. They also noted that only a handful of studies followed people up months after coaching ended.

You can avoid the same traps in your own program:

• Combine the leader's self-assessment with observations from people who work with them.
• Ask the same specific questions at the start and some months after coaching ends, not only at the last session.
• Where it fits, use existing data such as engagement survey results for the leader's team, retention of direct reports or progress on agreed projects, while acknowledging that many factors affect these.
• Look across a cohort of coached leaders rather than drawing conclusions from one case.

Jones, Woods and Guillaume's 2016 meta-analysis found smaller positive effects in studies that used multisource feedback. Whatever the explanation, it is a reminder that results depend on how you measure. When it comes to reporting, modest and credible beats impressive and fragile.

Protect confidentiality while you measure

Evaluation can quietly undermine the trust that makes coaching work. If leaders believe their sessions feed a performance file, they will stop being candid.

The ICF Code of Ethics asks coaches to have a clear agreement about what information is exchanged and how among all parties. Build evaluation into that agreement from the start. Agree who collects stakeholder feedback, who sees individual responses and what goes into aggregate reports. One safeguard is to report program-level findings across several leaders, while individual results are shared only with the leader and, if agreed, their manager.

For example, imagine a company coaching twelve senior managers. HR reports to the executive team on the share of goals that managers and their bosses rated as progressed, along with themes from anonymous stakeholder surveys. No one outside each three-way relationship sees what an individual leader worked on.

Report honestly

How you describe results matters as much as the numbers. Some guidelines:

• Say "associated with" or "followed by" rather than "caused" unless you used a design that supports causal claims, such as a comparison group.
• Report the share of leaders who made progress, not only the best stories.
• Separate satisfaction from impact.
• State limitations plainly: small numbers, self-report, other changes happening at the same time.
• Avoid presenting estimated dollar values as if they were measured.

Honest reporting builds the credibility you will need when the budget comes up for renewal. It also aligns with the ICF code's expectation that statements about the potential value of coaching be true and accurate.

For a closer look at the research itself, see our guide Executive Coaching Research: What Meta-Analyses Show and Don't.

Key takeaways

• Agree on specific, observable goals and a baseline at the start of each engagement, not after it ends.
• Frameworks such as the Kirkpatrick Model's four levels help separate reaction and learning from behavior change and results.
• ROI can be calculated, but every figure depends on assumptions about isolation and monetary value that should be disclosed.
• Combine self-report with observations from others and follow up after coaching ends.
• Protect confidentiality by agreeing on data flows up front and reporting at the program level.
• Describe results in careful language that does not claim more than your evidence supports.

Sources

• Kirkpatrick Partners, The Kirkpatrick Model
• ROI Institute, ROI Methodology
• Grover and Furnham, Coaching as a Developmental Intervention in Organisations: A Systematic Review of Its Effectiveness and the Mechanisms Underlying It, PLOS ONE (2016)
• Jones, Woods and Guillaume, The effectiveness of workplace coaching: A meta-analysis of learning and performance outcomes from coaching, Journal of Occupational and Organizational Psychology 89(2) (2016)
• Nicolau, Candel, Constantin and Kleingeld, The effects of executive coaching on behaviors, attitudes, and personal characteristics: a meta-analysis of randomized control trial studies, Frontiers in Psychology (June 2, 2023)
• International Coaching Federation, ICF Code of Ethics (effective April 1, 2025)