Research on Coaching ROI: What the Studies Really Show
Nour
Founder, Coachful
Research on coaching ROI includes substantial returns reported in particular programs. It does not establish a reliable financial return that every buyer should expect. The sample, benefit estimate, cost definition and meaning of “ROI” can change the headline considerably.
There is also a separate research question: whether coaching affects an outcome such as goal attainment or a leadership rating. Evidence of an outcome effect does not itself establish a monetary benefit, a cash saving or a profitable investment.
This guide reads the original reports behind three familiar headlines, considers what newer outcome research adds and completes a fictional funding appraisal. The purpose is to make a better decision without turning someone else’s estimate into your guarantee.
Separate four questions before comparing numbers
Was the service valued? A client may report that coaching was useful. That is relevant information about their experience, but it does not show a financial return.
Did an outcome change? A manager may use a clearer decision process or receive a different rating. Establish the measure, observer and period. An improvement still needs interpretation.
What did coaching contribute? Staffing, a new system, market conditions and other development can affect the same result. The contribution question concerns what would likely have happened without the intervention, not merely what happened afterward.
Was the benefit worth its cost? A financial evaluation needs a defensible monetary benefit and a stated cost basis. A worthwhile developmental outcome can remain unpriced when monetization is unsupported.
The UK government’s Magenta Book distinguishes process, impact and value-for-money evaluation. It also recognizes important benefits and costs that cannot be monetized. This is evaluation guidance, not evidence that coaching produces a particular return.
Read the three familiar reports in context
| Primary report | Reported headline | Financial evidence to keep beside it |
|---|---|---|
| Manchester Review, 2001 | Average described as 5.7 times investment. | 43 monetary estimates within a 100-executive sample. |
| Anderson / MetrixGlobal, November 2001 briefing | 529% excluding retention; 788% including it. | 30 respondents among 43 targeted participants in one organizational case. |
| ICF Global Coaching Client Study, 2009 | Median company return reported as 700%. | 189 respondents overall supplied both gains and costs; company figures concern the relevant subset. |
These are different reports, populations and calculations. Do not average the three headlines or describe them as repeated measurements of a single industry-wide return. Publication dates also matter: a document retrieved today is not a new study merely because a repository recently indexed it.
Manchester: an adjusted retrospective estimate
McGovern and colleagues’ original Manchester report covered 100 executives from 56 organizations who completed coaching between 1996 and 2000. Forty-three supplied monetary estimates. The authors capped high estimates and adjusted them using participants’ attribution to coaching and confidence in their judgments.
The report describes the average as 5.7 times investment and gives a worked net-return formula that subtracts program cost. Keep the source’s wording and formula together instead of silently treating the multiple as a standardized gross benefit-cost ratio.
The authors state that a controlled trial was not possible and that few participants had kept records of the affected business areas. Their explicit estimation procedure is informative, but retrospective confidence is not independent causal evidence. This is a historical sample estimate, not a forecast for a new buyer.
MetrixGlobal: one case with a consequential retention choice
Anderson’s November 2001 executive briefing describes coaching within a wider leadership-development effort that included assessments, mentoring, a workshop and strategic projects. Thirty of 43 targeted participants returned the questionnaire, followed by telephone inquiry about impact. Costs covered all 43 participants.
That briefing reports 529% excluding employee-retention benefits and 788% including them. The retention assumption changes the headline materially. A version retained in the ICF research portal reports the 529% result excluding retention; it should not be cited as if every circulated version contains identical wording.
Read the figures as claims from a particular organizational evaluation. The combined development context and participant estimates do not establish a randomized coaching-only effect. Neither figure is a default return for all executive coaching.
ICF: preserve both the subset and the return definition
The 2009 ICF executive summary says that 189 respondents, 9% overall, supplied both gains or losses and coaching expenditure. It explicitly cautions about the small ROI sample. The company-return figures concern respondents able to provide company financial amounts, not every organization in the survey.
The summary reports a median company return of 700% and says that 100% means earning the investment back. Its scale therefore differs from net ROI, where break-even is 0%. A median also differs from a mean: it identifies the middle of the relevant ordered observations, not the arithmetic average of all companies’ returns.
The often-repeated 86% recouping figure belongs to those able to supply company figures. Missing financial observations cannot simply be counted as either successful or unsuccessful investments. Their absence limits what the reported subset can tell a prospective buyer.
See why missing financial answers matter
Consider a separate fictional survey of five organizations, each spending £10,000. Only two provide a benefit estimate: £2,000 and £100,000. The middle of those two reported benefits is £51,000. That describes the responding pair; it does not establish the middle benefit for all five organizations.
If the three missing organizations each had zero benefit, the five benefits would be £0, £0, £0, £2,000 and £100,000. The population median in that invented completion would be zero. If the three missing benefits were each £100,000, the five would be £2,000 followed by four £100,000 figures, and the median would be £100,000.
Both completions fit the same two observed answers. Neither should be assumed true. The example shows why missing figures prevent a direct leap from a financial subset to an expected return for everyone. It is not a reconstruction of any historical study’s missing data.
Ask who could calculate a monetary result and why. A company with a clearly priced commercial project may find that easier than a client working on an unpriced relationship goal. Reporting coverage helps you understand which question the observed number can answer.
What newer outcome research adds
Cannon-Bowers and colleagues’ 2023 workplace-coaching meta-analysis found a moderate positive overall outcome effect after removing an outlier. The authors also noted that none of their included studies assessed outcomes at the organizational-results level, limiting cost-benefit conclusions. Read the primary meta-analysis for its inclusion criteria and limitations.
A separate 2023 meta-analysis by de Haan and Nilsson included 37 randomized studies, representing 39 samples and 2,528 participants. Its published abstract reports a moderate overall effect and indications of publication bias, with larger effects for self-reported than observed outcomes.
Those findings support a more substantive discussion of coaching outcomes than a satisfaction testimonial alone. They do not supply a pooled cash return. A standardized effect size expresses a difference relative to statistical variability; it is not a percentage increase in revenue or a monetary multiplier.
For broader interpretation, see research on coaching effectiveness. The question here is what can legitimately be carried from outcome evidence into a financial decision.
Why a randomized result can still leave ROI open
A suitable randomized design can strengthen a comparison of measured outcomes. The Magenta Book’s methods annex explains how random assignment supports impact estimation and describes practical limits, including implementation and effects on the comparison group. That is different from retrospectively asking participants what share of improvement they attribute to coaching.
But a strong design answers the question it actually measures. Imagine an entirely fictional trial in which a work-behavior rating averages six in the coaching group and five in the comparison group. If the relevant standard deviation were two, the simple unadjusted standardized difference would be one divided by two, or 0.5. This teaching calculation is not a reproduction of a published meta-analysis or its bias-adjusted estimator.
It would be incorrect to call that “50% ROI.” The units are rating-scale variability, not pounds. You still would not know whether the behavior changed expenditure, generated additional margin or affected an outcome worth leaving unpriced. You would also need the relevant intervention costs.
Read research limitations in that spirit. An outcome study can contribute useful evidence while leaving the financial question unresolved. Conversely, an impressive monetized estimate can have weak causal support. Design strength and financial completeness are separate dimensions, rather than interchangeable labels of research quality.
Recognize the missing steps between effect and money
A study might report a change in a self-rated leadership measure. To convert that into a financial claim, you would need a credible connection to the business outcome, a supported attribution and an appropriate monetary valuation. Each step can introduce uncertainty.
Likewise, time released can be useful capacity without being a cash saving. If an employee uses a freed hour for another task, the payroll may remain unchanged. Avoided expenditure needs a credible alternative: which purchase or payment would otherwise have occurred, and why?
Retention is particularly sensitive to that alternative. A person staying does not prove they would have left without coaching. Compensation, management, personal circumstances and labor-market options can influence the decision. A replacement-cost estimate also needs a local basis, not an impressive salary multiplier borrowed from an unrelated setting.
Keep overlapping benefits separate. If a time-saving estimate already values the hours used to produce extra output, adding the full output value again may count the same resource twice. Ask the evaluator and finance owner to explain the basis rather than automatically adding every positive number.
State the formula before using a percentage
For this article, net ROI means monetary benefit minus cost, divided by cost, multiplied by 100. The benefit-cost ratio means benefit divided by cost. These definitions are explicit; they are not a claim that every coaching report has used the words identically.
With a cost of £10,000 and a supported benefit of £20,000, the benefit-cost ratio is 2:1 and net ROI is 100%. At £10,000 benefit, the ratio is 1:1 and net ROI is 0%. A break-even investment has recovered its cost but has no net surplus.
A gross return of seven times cost would mean a 7:1 ratio and 600% net ROI under these definitions. A 700% net ROI would instead require benefits of eight times cost. That distinction alone can change how a headline is understood.
These are arithmetic illustrations. They do not establish that the proposed benefit exists or is attributable to coaching. A perfectly calculated ratio can still rest on an unsupported input.
Begin the fictional funding appraisal
Lena leads people development at a fictional company. She proposes an eight-week service for eight newly appointed managers, running from 16 November 2026 to 10 January 2027. Her director Kiran holds the relevant budget authority. Finance owner Iqbal asks whether the research justifies promising a seven-times return. Coach Tara will deliver the service if approved, but is not the company’s independent financial evaluator.
The practical objective is for the managers to prepare clearer delegation agreements and reviews. The company can choose to fund that development without requiring a proven positive cash return, but it needs a complete cost and an honest evaluation plan.
All people, budgets, decisions and documents below are invented. No historical study supplies the case’s financial inputs, and no real procurement approval or customer outcome is claimed.
Lena initially puts “research shows seven-times ROI” in a draft proposal. Iqbal asks for the original source, sample, return definition and relevance. Their first completed action is a claim review, before committing the money.
Complete the source and applicability check
Lena identifies the 2009 ICF summary as the source of her seven-times wording. She records the financial subset and break-even convention, then removes the statement that every buyer should expect that return.
She also reads the other historical reports and the newer outcome studies. She cannot combine them into a new average because they address different questions. The evidence supports considering a carefully specified coaching service; it does not establish this company’s future financial result.
Her revised proposal statement reads:
“Published coaching research includes positive outcome findings and historical reports of financial returns in particular samples. The financial reports do not establish an expected return for our proposed eight-manager service. We propose a bounded developmental investment with explicit cost, delivery and review conditions. No cash-return commitment is included.”
Iqbal accepts that wording for the internal appraisal. The original draft is retained with a correction note, so the change is traceable. An updated document does not make the earlier overstatement disappear from the decision history.
Count the complete appraisal cost
The proposal fee is £16,000 for the eight-manager service, including Tara’s agreed delivery, preparation and administration. Internal time and evaluation work are additional economic costs. They are not counted again inside Tara’s fee.
| Cost item | Basis | Amount |
|---|---|---|
| Coaching service | Agreed complete provider scope. | £16,000. |
| Manager participation | 8 managers × 6 hours × £50 per hour. | £2,400. |
| Internal coordination | 8 hours × £50 per hour. | £400. |
| Evaluation work | 10 hours × £60 per hour. | £600. |
| Total appraisal cost | Provider fee plus all stated internal allowances. | £19,400. |
For each manager, the six-hour allowance covers four one-hour sessions plus two hours for preparation and review. It is an allocated economic-time cost, not six hours of extra payroll automatically paid because coaching occurred.
All amounts use the same basis excluding any recoverable indirect tax; Iqbal confirms the appropriate accounting treatment for this fictional organization. The provider cash fee and total economic cost are different figures. Neither is described as an actual payment already collected.
The approved scope has no additional tool purchase or travel requirement. If either becomes necessary, it must enter the revised cost before a later return calculation. Keeping the denominator at £16,000 while adding benefits from everyone’s time would overstate the proposed economics.
Reject unsupported benefits before calculating scenarios
Lena’s initial worksheet lists three possible benefits: £30,000 for retaining a manager, £10,000 of time released and £5,000 of avoided external support. These sum to £45,000, but they are possibilities, not verified benefits.
Iqbal asks for the counterfactual and evidence for each. There is no supported estimate that a manager would leave without coaching, so the £30,000 is removed from the appraisal’s claimed financial benefits. It can remain a question for later evaluation, not a saving already earned.
The £10,000 represents modeled staff capacity. No expenditure reduction has been identified. It is therefore labeled a prospective capacity scenario, with no cash-saving claim. The £5,000 contract has not been canceled or shown to be avoidable because of coaching, so it remains an unverified possibility.
The result is not “zero benefit.” It is that a defensible monetized coaching benefit has not yet been established. No actual net ROI can be reported. Assigning zero to an unknown input and calling the resulting negative percentage a measured outcome would also mislead.
This is the point at which calculation discipline matters most. Removing unsupported numbers can make the appraisal less impressive while making the funding decision more trustworthy.
Calculate a sensitivity table without calling it a forecast
Lena and Iqbal use three hypothetical monetary-benefit inputs to understand the £19,400 cost. None is assigned a probability or presented as an estimate supported by the research.
| Assumed benefit | Net benefit after £19,400 cost | Benefit-cost ratio | Net ROI |
|---|---|---|---|
| £0. | −£19,400. | 0:1. | −100%. |
| £10,000. | −£9,400. | About 0.515:1. | About −48.5%. |
| £20,000. | £600. | About 1.031:1. | About 3.1%. |
The break-even monetary benefit is £19,400 on this basis. The table shows what would follow if an input were defensible; it does not make it defensible. In particular, the £10,000 row must not be relabeled “expected cash saving” merely because that number also appeared in a capacity illustration.
A useful next question is what evidence could support a benefit within the chosen period. That may require a comparison, an independently checked cost change or a better account of other influences. It is not answered by selecting the most attractive historical multiple.
Issue the actual funding decision and evaluation brief
The company’s fictional decision rules allow a developmental pilot costing no more than £20,000, provided participation is agreed, the service scope is clear and reporting does not promise an unestablished financial result. The £19,400 appraisal leaves £600 within that ceiling.
Before Kiran decides, Lena shares the proposed service scope and reporting terms with each manager. All eight voluntarily confirm that they wish to participate on those terms if funding is approved. Lena records those confirmations; participation agreement is therefore established before the funding approval.
Iqbal confirms the cost basis. Kiran approves the bounded eight-manager service for the defined developmental purpose, using the revised wording. This completes the funding decision; it does not mean the future service has been delivered or has paid for itself.
The final evaluation brief records:
“Evaluate delivery and managers’ use of clearer delegation agreements over the agreed service period. Before delivery, record the starting practices and relevant context. At review, examine work examples, participant accounts and authorized stakeholder observations, including contrary or unintended effects.
Keep coaching reflections outside sponsor reporting. Finance retains the cost register and decides whether any later monetary valuation has a defensible basis. Concurrent management training and operational changes will be documented. No separate coaching cash ROI will be reported unless the evidence supports both attribution and monetization.
The funding approval is for this service only. A future extension requires a fresh review of actual delivery, costs and value.”
Lena sends the approved brief through the company’s internal evaluation channel, and the delivery owner acknowledges it before onboarding begins. The seven-times promise is absent from the final proposal. The decision is complete even though the evaluation and any future benefits remain to be observed.
Organize coaching evidence in Coachful without inventing ROI
Coachful is our #1 recommendation for organizing the goals, preparation, sessions and chosen actions that support an evaluation. It does not independently certify attribution or convert a completed task into an organizational financial return.
Tara can prepare an original worksheet about a manager’s intended delegation practice, deliberately deliver it and discuss the selected response in that person’s coaching. She explicitly adds agreed next actions to the client plan instead of assuming a form submission creates the right task automatically.
In this fictional service, Tara is the sole authorized coaching staff member. Each manager uses their individual private Coachful conversation for preparation arrangements, and Tara replies there within two working days, Monday to Friday, Europe/London, with 25 December and 1 January closed as disclosed in the agreement. The sponsor is not added to those conversations or automatically given access to private reflections.
Only the permitted summary leaves the coaching record through the agreed reporting process. Iqbal’s financial register and company evaluation data remain in their authorized company systems. Shared operational data, private writing and sponsor reporting are not interchangeable audiences.
The ICF Code of Ethics provides relevant confidentiality and record-handling expectations for ICF professionals. An employer paying for coaching does not by itself settle what private material may be disclosed.
The images show demo prompts and client preparation, not a validated ROI dashboard. Explore Coachful’s features for the actual client workflow. For local measurement design beyond research interpretation, see executive coaching ROI measurement.
A contrasting decision: useful value remains unpriced
In another fictional organization, a small nonprofit funds a leader’s coaching to prepare a difficult role transition. Its board can afford the defined service and wants a clearer handover and explicit working boundaries. It has no defensible estimate of avoided turnover or additional revenue.
The board records the actual cost, agrees proportionate review and approves the service for that purpose. It does not multiply the leader’s salary by a generic replacement factor to manufacture a large benefit.
At review, the handover document is complete while one relationship question remains unresolved. The board retains both observations. It judges the practical work useful and ends the agreed service, without reporting a cash ROI or claiming the coach caused every improvement.
This is a completed value judgment with a financial boundary. It is different from asserting that financial evaluation is never useful or that a positive testimonial justifies any price.
Use a short research-claim review before publication
- Locate the primary document. Confirm author, date and version, not only a recent page quoting it.
- Find the financial denominator. Identify who supplied figures and who was missing.
- Read the formula. Check gross versus net, break-even convention and mean versus median.
- Examine inputs. Separate recorded expenditure, modeled capacity, estimated avoided cost and unpriced value.
- Check attribution. Ask what the design can say about coaching rather than the whole development effort.
- Write the permitted claim. Keep the sample and limitation beside the number, then decide whether it belongs in the intended communication.
If the full method is unavailable, say what you could inspect. Do not present an abstract, repost or executive briefing as access to undisclosed raw data. If the evidence cannot support the intended claim, change the claim or gather better evidence.
Frequently asked questions
What is the typical financial ROI of coaching?
The reports discussed here do not establish one reliable expected return for all buyers. Different populations, costs, estimates and definitions prevent that conclusion.
Can I use 788% as a guarantee?
No. It belongs to a particular historical case and includes a consequential retention assumption. It is not a promise about a new engagement.
Does a positive meta-analysis establish financial return?
No. Outcome effects require additional evidence about organizational consequences, attribution, monetary value and costs before supporting a financial ROI.
Does an unknown benefit mean a measured ROI of minus 100%?
No. Minus 100% follows from an assumed zero monetary benefit in a scenario. An unknown or unestablished benefit should be reported as such.
Can I fund coaching without monetizing every benefit?
Yes, where the decision maker has an appropriate purpose, affordable cost and proportionate review. Record that value judgment honestly rather than supplying a fabricated return.
