Coaching, Development, and Performance Conversations in Practice

This page is the practice side of coaching and development — the how-to of the one-on-one conversations, development design, and performance diagnosis that managers actually run. It moves from conversation craft (asking versus telling, the GROW structure, career conversations), to development design (plans, stretch, the 70-20-10 rule), to building coaching capability in managers, and finally to the harder case: diagnosing underperformance before acting on it.

Most of this material is directional rather than settled. Coaching and development sit in a literature where the practitioner frameworks are far better known than the controlled evidence behind them, so each section below states plainly how strong the evidence is — and flags where a popular figure (the “70-20-10” ratio, “7x coaching ROI”) is a heuristic or a vendor claim rather than a measured fact. The underlying science — self-determination theory, goal-setting, the coaching mechanism — lives on the Motivation and Coaching at Work research page; this page is what a manager does with it.

Is asking really better than telling — does an inquiry approach develop people more?

The principle that drawing people out through questions develops them more than directing them is long-standing and supported by autonomy-support theory and controlled studies of question type, but rigorous head-to-head asking-versus-telling causal evidence is scarce and the answer is context-dependent. Directive instruction is often correct and faster for novices, for safety-critical or emergency situations, and where the employee lacks the knowledge to generate options.

The best-known framing is Edgar Schein’s Humble Inquiry (2013). Schein argues that Western management defaults to a “culture of Tell” that “shuts people down,” and defines Humble Inquiry as “the fine art of drawing someone out, of asking questions to which you do not know the answer.” Schein offers no controlled evidence; it is a practitioner argument from decades of consulting.

The underlying mechanism — that autonomy-supportive, question-led interaction promotes motivation and learning — is best evidenced through self-determination theory. At the level of practice, the closest controlled evidence is on question type: Grant and O’Connor (2010) and replications show solution-focused questions outperform problem-focused ones on affect, self-efficacy, and goal approach. Correlational field work — Ellinger, Ellinger and Keller (2003) and later managerial-coaching studies — links managers’ coaching behaviour (listening, asking, facilitating rather than directing) to higher employee role clarity, satisfaction, commitment, and performance, but these are cross-sectional self-report designs that cannot show that asking causes development.

The honest reading is a contingency stance, not an ideology. The evidence supports inquiry as a default for developing capable people, not as a universal rule. In a small firm, where working managers are often promoted experts who default to telling because it is fast, the defensible position is “ask first when developing, tell when the situation demands it.”

Sources: Grant and O’Connor, The differential effects of solution-focused and problem-focused coaching questions, Industrial and Commercial Training 42(2), pp. 102–111 (2010); Edgar H. Schein, Humble Inquiry (2013); Ellinger, Ellinger and Keller, Supervisory coaching behavior, employee satisfaction, and warehouse employee performance, Human Resource Development Quarterly 14(4) (2003).

Confidence: industry-consensus, with the caveat that telling is sometimes correct (novices, emergencies).

How do you structure a coaching conversation — and does the GROW model actually work?

GROW (Goal, Reality, Options, Will) is the most widely taught structure for a coaching conversation, but it is a practitioner framework, not a tested theory, and there is little or no controlled evidence that the GROW sequence itself produces better outcomes than any other structure. What is better evidenced are GROW’s component parts: goal-setting and solution-focused, autonomy-supportive questioning.

GROW was popularized by Sir John Whitmore in Coaching for Performance (1992) and was developed in the 1980s from Timothy Gallwey’s “Inner Game” work. Even Anthony Grant — a leading coaching researcher — notes (2011) that session-structure frameworks like GROW are taught almost universally yet have had “little or no debate in the literature about… the relative advantages or disadvantage of different specific session structure frameworks.” Grant calls GROW “a simple map to help guide the coaching session,” a methodology “to be used, not [an ideology] to be rigidly adhered to.”

The evidenced payoff sits inside the structure, not in the letters. The “G” maps onto goal-setting, and the questioning style maps onto solution-focused, autonomy-supportive inquiry. Grant and O’Connor (2010), a randomized pilot (N=39), found solution-focused coaching questions increased positive affect, self-efficacy, and goal approach where problem-focused questions did not — a result replicated cross-culturally. These are small, mostly student-sample lab studies measuring within-session affect, not long-run development.

For a 20-to-200-person employer with no formal coaching infrastructure, a simple shared structure like GROW has real practical value as a common language for managers — provided it is presented as a flexible scaffold, not a guaranteed mechanism.

Sources: Anthony M. Grant, Is it time to REGROW the GROW model?, The Coaching Psychologist 7(2), pp. 118–126 (2011); John Whitmore, Coaching for Performance (1992, origin of the GROW model); Grant and O’Connor, The differential effects of solution-focused and problem-focused coaching questions, Industrial and Commercial Training 42(2) (2010).

Confidence: directional — GROW is a practitioner framework; the components are better-evidenced than the acronym.

How do you have a career conversation that keeps a high-potential person engaged?

The recommended practice has four practitioner moves: surface what the person actually wants rather than what you assume; be honest about real opportunities rather than over-promising promotions; give real exposure and stretch in the meantime; and ask before they are halfway out the door. These rest on thin controlled evidence and borrow their strength from the development and retention literatures.

The “stay interview” — asking a valued current employee what would make them stay, rather than waiting for the exit interview — comes from Beverly Kaye and Sharon Jordan-Evans’ Love ‘Em or Lose ‘Em. It is a book and practitioner source: the stay-interview practice itself has not been validated in controlled trials. There is no body of randomized or quasi-experimental work showing stay interviews or career conversations causally reduce turnover. The closest rigorous evidence is on the mirror practice — a Cochrane review (Webster and Flint, 2014) concluded that evidence on whether exit interviews reduce turnover “is currently not available.” So the practice is best justified indirectly: it operationalizes the perceived-growth and autonomy/competence mechanisms that are supported in the retention and self-determination literatures, and the “stretch in the meantime” element rests on DeRue and Wellman’s developmental-challenge findings.

The honesty element has real substance. Over-promising a promotion you cannot deliver predictably backfires; the defensible practice is candour about what is and isn’t realistically available, then development you can deliver now. For a small firm with a flat structure and few promotion slots, honest career conversations are both more necessary and more awkward. Large-enterprise career-pathing infrastructure does not transfer; the conversation does — through lateral growth, scope expansion, and visible stretch, paired with candour.

Whether developing people in this way actually reduces departures is treated on the Employee Development and Business Outcomes page.

Sources: Webster and Flint, Exit interviews to reduce turnover amongst healthcare professionals, Cochrane Database of Systematic Reviews (2014); Beverly Kaye and Sharon Jordan-Evans, Love ‘Em or Lose ‘Em: Getting Good People to Stay (5th ed., 2014, origin of the stay interview); DeRue and Wellman, Developing leaders via experience, Journal of Applied Psychology 94(4) (2009).

Confidence: directional.

Do development plans, stretch assignments, and the 70-20-10 rule actually develop people?

70-20-10 is a practitioner heuristic, not an established empirical finding, and the numbers should never be presented as measured proportions. What is better evidenced is that challenging job experiences drive development — but with diminishing returns, and stretch without support and feedback tips into overload and harm.

The “70-20-10” rule — that development comes 70% from challenging experience, 20% from others, 10% from formal training — derives from survey work by McCall, Lombardo and Morrison (The Lessons of Experience, 1988) and was popularized later. The figures come from retrospective self-reports by already-successful executives, with no control group, no longitudinal data, and suspiciously round numbers. Clardy (2018) concludes the evidential basis for the 70% claim is weak and that policies relying on it “are likely to be misleading.”

The better-evidenced core is that challenging job experiences drive development. DeRue and Wellman (2009), studying 225 on-the-job experiences across 60 managers, found developmental challenge predicts leadership-skill development — but with diminishing returns: beyond a point, more challenge yields less learning, and “access to feedback can offset the diminishing returns associated with high levels of developmental challenge.” The practical implication is precise: stretch helps, but stretch without support and feedback tips into overload and harm.

For a small employer, this is both an advantage and a trap. Genuine stretch is cheap to offer — broader scope, new responsibilities, cross-functional exposure — without formal programs. But DeRue and Wellman’s finding means a stretch assignment dropped on someone without feedback or coaching support is as likely to burn them out as develop them. And the 70-20-10 ratio should never be used to justify cutting training (“only 10% matters”) — a misuse the evidence does not license. The feedback that makes stretch safe is treated in depth on the Feedback and Performance research page.

Sources: DeRue and Wellman, Developing leaders via experience: the role of developmental challenge, learning orientation, and feedback availability, Journal of Applied Psychology 94(4), pp. 859–875 (2009); McCall, Lombardo and Morrison, The Lessons of Experience (1988, origin of the 70-20-10 heuristic); Clardy, 70-20-10 and the Dominance of Informal Learning: A Fact in Search of Evidence, Human Resource Development Review (2018).

Confidence: directional — 70-20-10 is a heuristic, not an established ratio.

Does training managers to coach (and adopting a coaching culture) change behaviour and outcomes?

Coaching is a learnable managerial capability: brief mindset training can shift coaching behaviour, and field studies show skills training transfers to observer-rated behaviour. But the evidence is small-sample and often self-report, and coaching-culture ROI figures are vendor claims, not controlled findings.

The cleanest causal evidence is Heslin, Latham and VandeWalle (2005): a 90-minute self-persuasion workshop shifted “entity theorist” managers — who believe ability is fixed — toward a growth (“incremental”) mindset, an effect sustained over six weeks, and the shifted managers became more willing to recognize improved employee performance. The companion 2006 paper showed managers’ implicit person theories predict the coaching employees report receiving, and that inducing an incremental mindset increased the quantity and quality of their performance-improvement suggestions. Manager mindset is therefore both a precondition for coaching and itself trainable. (Caveat: these outcomes used appraisal and vignette tasks and student “managers,” not longitudinal field coaching.)

For field transfer, Peláez Zuberbuhler and colleagues (2020) ran a controlled, non-randomized trial: 41 managers received a coaching-skills program, and supervisor-rated coaching skills rose with very large effect sizes (employee-rated skills with smaller ones). Treat this as promising, not conclusive — small, single-firm, non-randomized, with likely rater-expectancy bias. Grant (2010) adds that adopting coaching behaviours takes time — roughly six months before perceived benefits outweigh costs.

Crucially, the “coaching culture ROI” figures from the coaching industry are vendor self-report claims, not controlled findings. The “700% / 7x ROI” figure comes from a 2009 ICF client survey, not an experiment, and must never be cited as fact.

For a 20-to-200-person employer, the encouraging news is that the cheapest, most evidence-backed lever — a short mindset or skills workshop plus follow-up — is within reach and does not require a formal “coaching culture” program. The realistic expectation is modest behaviour change that takes months to embed, not a 7x return.

Sources: Heslin, Latham and VandeWalle, The effect of implicit person theory on performance appraisals, Journal of Applied Psychology 90(5), pp. 842–856 (2005); Heslin, VandeWalle and Latham, Keen to help? Managers’ implicit person theories and their subsequent employee coaching, Personnel Psychology 59(4) (2006); Peláez Zuberbuhler, Salanova and Martínez, Coaching-Based Leadership Intervention Program: A Controlled Trial Study, Frontiers in Psychology 10:3066 (2020); Jones, Woods and Guillaume, The effectiveness of workplace coaching: A meta-analysis, Journal of Occupational and Organizational Psychology 89(2) (2016); ICF and Human Capital Institute, Building a Coaching Culture / ICF Global Coaching Client Study (2014, vendor ROI claims).

Confidence: industry-consensus that coaching is learnable, with the explicit caveat that ROI figures are vendor claims.

How do you diagnose underperformance — can't versus won't?

Persistent underperformance is a diagnosis problem before it is an action problem. First rule out a Human Rights Code-protected cause that triggers the duty to accommodate; then diagnose whether the gap is incapacity (a “can’t” — skill, clarity, tools, or fit, which routes to performance management) or a choice (a “won’t” — conduct, which routes to discipline). Acting before the cause is diagnosed is how employers end up running a coaching process on a discipline problem (toothless), a discipline process on an incapacity problem (unfair and legally risky), or any performance process at all on what is actually a disability that should have been accommodated (potentially discriminatory).

Step 0 — Accommodation overlay first. Before labelling anyone a poor performer, ask whether the shortfall could be a manifestation of a disability, mental-health condition, addiction, or another Code-protected ground (for example, family status). In Ontario, the Human Rights Code duty to accommodate to the point of undue hardship is triggered by known needs, and the duty has both a procedural part — the employer must actually investigate and consider accommodation — and a substantive part, the reasonableness of what is offered. The duty can be engaged even without a formal request where an employee is clearly unwell; the employer is expected to inquire, though not to diagnose. Skipping this step does not just risk a human-rights complaint; it makes any later performance-based termination far harder to defend. If a protected ground is in play, the accommodation analysis runs before and alongside any performance process — the substantive duty itself is set out in the duty-to-accommodate note.

Step 1 — Is it worth fixing? Mager and Pipe’s first question is brutally practical: describe the gap in observable, behavioural terms (“submits reports 2+ days late, with rework required on about half”), then ask what would happen if it were left alone. If the consequence is mere aggravation, deeper analysis may not be worth the cost. If it materially hurts the business or the team, proceed.

Step 2 — The fork: can’t versus won’t. This is the heart of Mager and Pipe’s performance analysis. The diagnostic question is: could the person do it if their life depended on it?

• If no — it is a “can’t” (incapacity). Now sub-diagnose, because the fix differs:
  • Skill or knowledge gap — they were never trained, or have lost a rarely-used skill — addressed with training, job aids, on-the-job coaching.
  • Unclear expectations — they do not actually know what “good” looks like — addressed by clarifying standards and goals. This is the single cheapest, most common, and most overlooked cause; Pulakos and O’Leary put “communicating clear work expectations” at the centre.
  • Missing tools, time, information, or feedback — the system is set up against them — fix the environment, not the person.
  • Genuine fit problem — the role’s essential demands exceed what this person can do even with support — this is still “can’t,” and it routes to honest performance management and, potentially, a managing-out decision (see the coaching-versus-managing-out note).
  • All of these route to performance management and coaching — not discipline.
• If yes — they could do it, but do not — it is a “won’t” (a choice). Mager and Pipe’s “won’t” branch asks: is desired performance punishing? Is non-performance rewarding? Does performing well matter to them at all? Where the cause is a deliberate choice — refusal, rule-breaking, dishonesty — the matter is in conduct or misconduct territory, which routes to the discipline track: progressive discipline and, at the extreme, just cause.

Why the fork matters. Discipline answers culpable misconduct (“won’t”) with escalating consequences. Performance management answers non-culpable incapacity (“can’t”) with clarity, training, tools, and feedback. Using the wrong track is both ineffective and, in Ontario, legally hazardous: courts and tribunals expect employers to have correctly identified the problem and given a genuine, supported opportunity to improve before treating poor performance as a basis for dismissal. Whether a formal performance-improvement plan is the right vehicle for that opportunity is examined separately in the do performance-improvement plans work note.

For a small or mid-sized employer without a dedicated HR function to catch a misdiagnosis, the protective habit is to build the three questions into how managers think — Is a protected ground in play? Can’t or won’t? If can’t, which kind? — and to write the answer down at the time, in behavioural terms. That contemporaneous note protects both the employee, from being disciplined for something they could not control, and the business, from a wrongful-dismissal or human-rights claim. When the diagnosis points to a conversation that must actually be had, the difficult-conversations guide covers running the meeting itself.

Sources: Mager and Pipe, Analyzing Performance Problems: Or, You Really Oughta Wanna (3rd ed., 1997); Ontario Human Rights Commission, Policy on ableism and discrimination based on disability — the duty to accommodate (2016); Pulakos and O’Leary, Why is performance management broken?, Industrial and Organizational Psychology 4(2), pp. 146–164 (2011).

Confidence: industry-consensus.

---

This page is general information, not legal advice. The performance-diagnosis section touches Ontario employment and human-rights law — in particular the duty to accommodate, progressive discipline, and just cause — and those edges are summarised here only to show where a coaching question becomes a legal one. Where a protected ground may be in play, or where poor performance is heading toward discipline or dismissal, obtain advice before acting.