Retention impact of executive coaching

Retention is a major—often undercounted—source of ROI from executive coaching. This article explains how to measure retention effects credibly and convert them into pounds without overstating the case. For the full framework, see our pillar guide: How to measure ROI of executive coaching programmes.

1) Why coaching affects retention (evidence)

Meta‑analyses consistently find that coaching improves goal attainment, self‑efficacy and resilience—capacities that reduce burnout risk and strengthen performance. See Frontiers in Psychology (2023 meta‑analysis) and Theeboom et al. (2013). Manager quality is also pivotal for engagement, and engagement links to staying power: Gallup’s research indicates managers account for roughly 70% of the variance in team engagement; see Forbes (2025 summary of Gallup findings).

2) What to measure (definitions and windows)

• Primary outcome: voluntary manager attrition over a defined window (e.g., 6–12 months).
• Secondary outcomes: absence days per FTE, internal mobility, manager engagement pulse scores.
• Cohorts: coached managers vs matched non‑coached managers (similar tenure, function, region, baseline performance).
• Windows: 8–12 week baseline; 12–16 week coaching; 6–12 month follow‑up for retention outcomes.
  
  

3) Design for credibility (pre/post + matched cohort)

• Compute pre → post changes and compare cohorts; add a difference‑in‑differences cut where possible.
• Track mechanism indicators (self‑efficacy/resilience short scales; cadence and quality of 1:1s).
• Document confounders (restructures, comp changes) and run sensitivity checks.
  
  

4) Valuing avoided attrition (conservative rules)

Use a conservative, contextualised approach. Widely cited research from the UK (Oxford Economics for Unum) estimates the cost of replacing an employee at around £30,000, driven largely by lost output during ramp‑up; see

Oxford Economics/Unum “Cost of Brain Drain” summary and the full report (PDF). To keep estimates current, pair this with your own recent recruitment fees and ramp times.

Valuation formula (illustrative):

• Avoided attrition value ≈ (# fewer leavers) × (Recruiting fees + Onboarding/training + Lost productivity during ramp).
• If role‑specific data is available, value per role × role mix is preferable to a flat figure.
  
  

5) Worked example

Coached cohort (n=20 managers) records 1 voluntary leaver over 12 months; matched cohort (n=20) records 4 leavers. Estimated avoided leavers = 3.

If your current data shows £8,000 in fees + £2,500 onboarding + ~£20,000 lost productivity, avoided attrition value ≈ 3 × £30,500 = £91,500 (directional). Present as a range (±10–20%).

6) Link retention to the wider ROI picture

• Higher continuity → stronger customer relationships and forecast reliability (fewer hand‑offs).
• Lower absence and burnout risk → steadier execution capacity.
• Internal mobility gains → reduced external hiring costs.
  
  

7) Governance and safeguards

• Aggregate sensitive people data; follow your DPA/ISO processes.
• Be explicit about purpose: better leadership and sustainable performance.
• Avoid over‑claiming: clearly separate retention value from revenue or time‑saved benefits to prevent double counting.
  
  

Bottom Line

Q: How do we value the retention impact of executive coaching?

A: Use avoided attrition value per leaver (fees + onboarding + lost productivity) multiplied by fewer leavers; prefer role‑specific inputs where possible.

Q: How should we measure retention impact credibly?

A: Run a pre/post with a matched cohort over 6–12 months, track voluntary attrition and mechanism indicators, and add a difference‑in‑differences view.

Q: What assumptions should we agree with Finance?

A: Recruiting fees, ramp time, lost productivity method, and which leavers count (e.g., voluntary only). Keep assumptions conservative and visible.

Q: What caveats should we disclose?

A: Programme scope, any restructures/market shocks, and the age of external benchmarks (e.g., Oxford Economics 2014) alongside your current internal data.