Leadership Academy — Session 5: Applying COIN in the Real World, and Coaching vs. Mentoring vs. Managing

COIN changes shape depending on who you're talking to and why. Plus a new distinction — managing, coaching, and mentoring are three different jobs, and mixing them up costs you.

Posted Aug 7, 2026

By Rahul Chandna

19 min read

Part of the Leadership Academy series — notes from Exaze’s Leadership Academy, Batch 1.

COIN isn’t one script — it changes with the relationship

Session 4 introduced COIN (Context, Observation, Impact, Next step) as the structure for a difficult conversation. This session spent most of its time on the part that doesn’t show up in a clean template: COIN behaves differently depending on who’s on the other side of it, and pretending it’s always the same conversation is where people get stuck.

Downward, performance issue — the straightforward case. A team member has missed a committed date more than once. Context: “I need to raise something with you which has been happening a number of times, and I want it sorted out now.” Observation: “You agreed on Monday to send this by Wednesday — this is the third time that hasn’t happened.” Impact: “If I don’t get the plan in time, I can’t give you feedback before it goes to the customer, and that delays the whole submission.” Next step: “Going forward, either you meet the commitment, or you tell me a couple of days in advance if you’re going to miss it. Don’t go silent and come back with a surprise. What feels workable for you?” Straightforward, because there’s formal authority behind the “next step” — you can actually ask for a commitment.

Peer to peer — no authority to lean on. A peer, not a report, is dismissive of a junior colleague’s ideas in meetings — “you’ve tried that, it doesn’t work,” cutting people off before they finish. The junior colleague starts going quiet in meetings as a result. The peer doesn’t experience himself as dismissive; from his side, he’s being efficient. Context, Observation, and Impact work the same way here as they do downward. Next step doesn’t. You have no standing to tell a peer what to do differently, so it shifts from an instruction to a request: “Can we make sure that next time someone gives a suggestion, we at least tell them why we’re not taking it? It costs us a minute or two, but not doing it costs us this.” The rule for peers: suggest or request. Never plead, and don’t issue next steps you have no authority to enforce.

Disagreement over a position — a different shape of COIN entirely. This one isn’t about behavior, so the standard order doesn’t fit. Someone on your team asks you to skip the final round of a hiring process — a coding assessment — because the candidate might walk if they wait any longer. You believe skipping it is genuinely risky; you’ve been burned by exactly this shortcut before. Here the modified sequence is: first find the other person’s position (ask, don’t state — “why are you saying that?”), acknowledge it (“I understand you’re worried about losing this candidate — that’s a real risk”), share your own position (“last time we skipped the technical test, the person didn’t last a month, and I don’t want that again”), find the shared interest (“we both want this role filled — the question is how we protect against the risk without adding it”), and propose something that closes the gap (“what if we do a 15-minute rapid assessment instead of the full 60-minute one?”). The goal is explicitly to persuade, not to overrule.

Expectation mismatch — stop litigating the past. You ask for a finished document; you get a draft, because the other person wanted to show progress, not hand over a finished piece. Both sides are technically right from where they’re standing — neither is lying, and there’s no clean answer to “who was correct.” The instruction here is blunt: when it’s an expectation mismatch, don’t relitigate what happened. There’s no winner in that conversation. Acknowledge the gap once, then move entirely to what you both need going forward — “let bygones be bygones, here’s what I need next time” — rather than getting pulled into re-establishing who said what.

One nuance on delivery raised in the session, worth noting critically rather than taking as settled fact: the instructor suggested that when the other person in an expectation-mismatch conversation is a woman, it can land better to show the gap rather than state it directly, attributing this to psychology research without citing anything specific. That’s a generalization I’d treat with real skepticism rather than a rule — but the underlying, gender-neutral version of the point is worth keeping: how directly someone wants to be told they’re wrong varies by person, and reading that correctly matters more than following a script.

Something practical: why you talk too fast in hard conversations, and what to do about it

A participant asked why they kept blurting out their whole point too quickly once a conversation turned uncomfortable. The answer: people speed up either because they’re excited, or because they want to get out of a situation that feels bad — and an uncomfortable conversation is exactly that situation. The average person speaks at roughly 120–140 words per minute; the practical fix offered was to consciously slow to about 80% of your normal pace, which gives the other person room to actually absorb what you’re saying rather than just process that you’re talking. The example given was Indian Prime Minister Atal Bihari Vajpayee’s deliberate use of pause in speeches — not silence for its own sake, but a beat that makes the listener think, or gives the speaker room to.

Prepare — leaders do, managers often don’t

A sharp, small distinction: managers typically walk into a hard conversation with only their own talking points ready — a list of “you didn’t do this, you didn’t do that” — and expect to hear the other side, decide, and leave. Leaders prepare before the conversation, more thoroughly than the person who called the meeting: what’s the context, what’s the specific observation, what’s the impact, what outcome are you actually looking for. The session’s planning template (worked through below) is built around a hard constraint: plan the conversation in under seven minutes. Long enough to think it through, short enough that “I’ll plan it properly later” doesn’t become an excuse to avoid having it at all.

Which direction does COIN travel?

Performance conversations are always downward — you don’t have a “performance conversation” with your boss. Disagreement and expectation mismatch can go either direction: the hiring example above was a report pushing back on a request from above; expectation mismatch can just as easily be raised by a team member toward their leader as the other way around.

The framework: Managing, Coaching, and Mentoring are three different jobs

The second half of the session moved from difficult conversations to a distinction the whole academy has been circling without naming directly: managing, coaching, and mentoring are not the same skill, and using the wrong one in the wrong moment either wastes time or damages the relationship.

Mode	What you’re doing	Key question you ask	Use it when	Overuse risk
Managing	Giving direction, holding accountability, upholding a decision.	“Is this getting done right, on time, correctly?”	The task has a correct answer, a decision needs upholding, or the situation is urgent.	You become a micromanager, and the person never builds real ownership.
Coaching	Holding back your advice and asking questions so the person finds their own answer.	“What do you think? What have you tried? What’s stopping you?”	The person already has the capability but it’s blocked, or they’re unsure of a capability they actually have.	People stop coming to you for a straight answer when they genuinely need one; it’s also not a substitute for performance management.
Mentoring	Sharing your own experience, knowledge, and path to help someone navigate theirs.	“What are you trying to do, and how can I help you get there faster?”	Someone is new to the team, navigating a role or career transition, or wants to understand how things actually work versus the theory.	The mentee becomes dependent on your specific path rather than building their own judgment.

The instinct to jump straight to advice is natural — it feels like help, and often it is help. Coaching asks you to sit with the discomfort of not giving the advice you already have, and ask a question instead. The session’s own scenario: a team member says they freeze up the moment a particular stakeholder walks into the room. The instinctive response is a suggestion (“try this,” “here’s what I do”). The coaching response starts by acknowledging the feeling (“that sounds frustrating”), then asks what, not why — “what do you feel when you say you freeze?” — without ever pulling in what the other person (the stakeholder) might be doing wrong. The point isn’t to analyze the stakeholder. It’s to get the actual emotion out into the open, because a person thinking out loud with someone genuinely listening usually arrives at a better answer than any advice you could have handed them.

A rough decision guide, straight from the session:

What the person says	What that tells you	What to do
“I don’t know how to do this.”	Genuine knowledge gap.	Manage (give clear direction) or mentor (share how you’d approach it) — don’t coach someone who has nothing to draw on yet.
“I know how, but I’m not doing it / I don’t want to.”	Something’s blocking them — confidence, habit, a relationship.	Coach. Ask what’s in the way before anything else.
“I’m doing it, but it’s not working.”	They need help diagnosing, not a lecture.	Coach first to help them think it through, then advise or manage to close the specific gap.
“I want to understand the bigger picture — how do I grow in this direction?”	They’re not looking for a fix, they’re looking for a path.	Mentor. Coaching them here will feel like you’re withholding.
There’s an actual performance problem.	This needs a decision and a consequence, not exploration.	Manage first. Coaching and mentoring can enhance performance — they cannot substitute for managing it.

The instructor’s own example, unfiltered, because it’s about the person compiling these notes: “I know Rahul. He is an abuser for mentoring. He just mentors people, keeps on mentoring. Then one day he suddenly becomes a manager and says, ‘how many of you have tried?’ Get rid of him. Not knowingly — because we ourselves didn’t know when we should have switched our mode.” Worth sitting with rather than laughing off: the pattern isn’t “which mode am I bad at” so much as “which mode do I default to regardless of what’s actually needed.”

What if the person hasn’t come to you at all? Sometimes you spot potential or a gap without being asked. The move is to decide privately first — does this need managing, coaching, or mentoring? — then open the actual conversation using COIN: state what you’ve observed, what you believe they could improve, and float the suggestion (“maybe we have a session on this — what do you think?”), then hand the decision back to them. Their response tells you which of the three modes the follow-up conversation should actually be.

A good leader isn’t fixed in one mode. The session was explicit that a single conversation can move through all three — the test isn’t “am I a coach or a manager or a mentor,” it’s “what does this specific person need from me right now.” Not everyone is naturally strong in all three modes either, and that’s fine as long as you know where your own gaps are rather than defaulting to your comfortable mode regardless of fit.

In the real world: Google’s decade-long Project Oxygen study (launched 2008, analyzing over 10,000 data points from performance reviews and interviews) found that the single strongest predictor of a great manager wasn’t technical skill — it was being a good coach, closely followed by empowering the team without micromanaging. Teams under managers who defaulted to coaching over directing saw better results, higher engagement, and lower attrition. It’s the coaching/managing distinction from this session, validated at large scale rather than as a training-room assertion. (Google re:Work · CNBC)

The framework: GROW, properly introduced

GROW — Goal, Reality, Options, Will — is the structure for a coaching conversation itself, so it doesn’t drift.

Stage	What it actually means	Note from the session
Goal	What the person wants to get out of this specific conversation — not their career, not their five-year plan.	People will often say something vague like “I want to feel better about this.” Push for specific: “what would feeling better actually look like?”
Reality	The current situation, as the coachee sees it — not as you see it.	This is their read of the situation, not a fact-check against your version of events.
Options	Possibilities, explored without judging them yet.	“What would you do if you knew you wouldn’t fail?” is the kind of question this stage is for.
Will	Moving from insight to an actual commitment.	“What will you do, specifically? When? How committed are you?”

One clarifying distinction the session drew: a coaching session’s goal is scoped to that conversation, not to the person’s whole trajectory. If someone’s long-term goal is “get selected for a national team,” no single coaching session solves that — the session’s goal might just be “leave today’s conversation with one thing to practice.” Confusing the two is a common way coaching conversations lose focus.

In the real world: Performance Consultants — the firm behind GROW’s creator, Sir John Whitmore — ran a public fireside chat in July 2026 with a live, unscripted GROW coaching demonstration. CEO Tiffany Gaskell played a manager checking in with a team member whose pushback in a meeting had left them feeling sidelined. Rather than opening with a verdict, she asked what the person had noticed, stayed with their defensiveness instead of smoothing it over, and — notably — skipped straight from Reality to Will once trust was established, rather than mechanically running the full sequence. The takeaway echoed this session almost exactly: a framework supports the judgement call, it doesn’t replace it. (Performance Consultants)

From the pre-reading: a bigger bank of GROW questions, and a step the session didn’t mention

The academy said active listening and “powerful questions” within GROW would be covered properly next session — the pre-reading gets a head start on both.

More questions per stage, from a leadership trainer’s own GROW breakdown, worth keeping alongside the ones already in the table above:

Goal: “What would make this a great quarter for you?” · “How will you know you’ve achieved your goal?” · “Describe what you’ll be doing when you can do this.”
Reality: “What have you already tried?” · “Who do you see doing this well, and why?” · “If you scored yourself out of 10 against this goal right now, what would you score?” · “What worries you about this? What excites you?”
Options: “What’s the upside and downside of doing this?” · “Who could help you?” · “What don’t you know?” (naming the specific skill or knowledge gap) · “Anything else?” — asked repeatedly, since the best option is rarely the first one offered.
Will: “How confident are you about doing this, on a scale?” · “What support will you need from me?” · “How will you keep yourself going?”

Two techniques worth adding to the structure itself: a “Check” step at the end of every stage — briefly summarising back what you heard before moving on — which the source describes as acting like a bridge between stages and makes the other person feel actually heard, not just processed. And at Reality specifically, a skill-vs-will diagnostic: is the gap that they genuinely don’t know how yet (skill), or that they know how but lack confidence (will) — because those need different responses from you. If you have your own idea during Options, the rule is to ask permission before offering it (“could I share a thought?”) and then ask what they think of it, rather than just stating it — preserving their ownership of the answer. One line worth remembering on why this matters: a manager’s “100% idea,” handed to someone else, often gets executed at 60–70%. That same person’s own “70% idea” — because they own it — often gets executed at 90–100%. (Helen Bryant — Leadership Accelerator)

On active listening, ahead of next session’s deeper dive: the core distinction is hearing (passive) versus listening (actively demonstrated — eye contact, repeating back, follow-up questions). One presenter’s own confession is a useful gut-check: he used to answer emails while people talked to him as a manager, and reflected afterward that “the person’s experience wasn’t as positive as it might have been, because I wasn’t fully present” — worth asking yourself the same question after your own 1:1s. A separate technique worth borrowing before any coaching or mentoring relationship even starts: “contracting” — explicitly agreeing the logistics and boundaries up front (when someone can and can’t come to you) — specifically to prevent the dependency that the mentoring table above already flags as mentoring’s overuse risk. And a sharp one-line distinction between coaching and mentoring, cleaner than most: a golf coach doesn’t need to be a better golfer than their player to add real value by helping them see their own game differently — mentoring nearly always requires the mentor to have walked the path; coaching doesn’t. (IMechE webinar, presented by Phil Slater)

A fuller worked GROW conversation, if the table above still feels abstract: a coaching demonstration video runs an entire real-sounding session with a professional who freezes up speaking in Zoom meetings and wants a promotion. At Reality, the coach asks what she’s done successfully in the past (in-person meetings) to isolate that the block is confidence, not skill. At Options, repeatedly asking “anything else?” surfaces a genuinely varied list: preparing the night before, using Zoom’s raise-hand feature to create a natural opening, “dressing for work” to feel psychologically ready, asking her boss to introduce her, and copying a colleague’s habit of briefly echoing what others say as a lower-risk way into the conversation. At Will, after she commits to an earlier bedtime, the coach asks “what could stop you from making that happen?” (the honest answer: Netflix) and then “what can you do to make sure that doesn’t happen?” — proactively troubleshooting the commitment before it can quietly fail, then locking it to specific dates rather than “soon.” (The Coaching Academy)

The planning template, applied

The Difficult Conversation Planning Template shared this week operationalizes everything above into a seven-minute worksheet: who the conversation is with, what type it is (performance / disagreement / expectation mismatch), the four COIN fields, anticipated pushback and how you’ll handle it, and — the two fields that separate planning from procrastinating — a specific date and time, and what you personally need to feel prepared.

Worked example (illustrative, combining the template with this session’s material): Who: a team member who’s missed two deadlines without flagging risk in advance. Type: Performance. Context: “I want to talk about the last two deliverables — I don’t think this is a motivation problem, so I want to understand what’s actually going on.” Observation: “Both times, I found out about the delay only when I chased you, not before.” Impact: “It meant I had no time to reset expectations with the client, so I absorbed both delays without warning them.” Next step: “From the next deliverable onward, I want a flag the moment you think a date’s at risk, even if you’re not sure yet.” Anticipated pushback: “They’ll say they didn’t want to raise a false alarm.” Handling it: acknowledge the instinct, then reframe — a false alarm costs nothing, a surprise costs trust (the same principle as Session 3’s SCOPE framework — escalate on threshold, not certainty). When: Thursday, 2pm, before sprint planning. What would help me feel prepared: re-reading the actual messages from both incidents so the Observation stays specific, not remembered-and-softened.

Assignment

Look back at real conversations with the people who report to you and ask, honestly, whether you were in the right mode each time — not whether you handled it well in the abstract, but whether you were managing when the moment called for coaching, or mentoring when someone actually needed a direct answer. Separately: plan and actually have one real difficult conversation using COIN and the seven-minute template — not hypothetically, with a date attached.

Closing note

A last, deliberately contrarian point to end on: the industry has spent the last several years glorifying “leaders” and quietly running down “managers,” as if management were the lesser skill. The session pushed back on that directly — both roles are aimed at outcomes; the difference is that leaders focus on the people who produce the outcome, managers focus on the outcome those people produce. Neither is inherently right or wrong. The next session is two weeks out, and will go deeper into GROW — the specific questions that make each stage work, and what active listening actually requires versus what it feels like to do it.

Previous: Session 4 — Communication Modes and Difficult Conversations · Back to the series index

Sources:

Self Improvement

This post is licensed under CC BY 4.0 by the author.