Leadership Academy — Session 4: Communication Modes and Difficult Conversations

Calibrating a message to its audience, a four-line status update format, and COIN — a structure for the conversation you've been putting off.

Posted Jul 31, 2026

By Rahul Chandna

8 min read

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

Picking up expectation management

Reactive expectation management — explaining a problem after it’s already happened — carries a real trust cost. Proactive management, flagging a risk before it fully lands and saying what’s being done about it, builds trust instead. SCOPE, set properly at the outset and maintained diligently, is still the foundation everything else in this post sits on top of.

The framework: four communication modes

Calibrating a message to its audience isn’t a soft skill — getting it wrong signals you don’t actually understand who you’re talking to.

Audience	How to communicate
Managing Up (your manager, leadership)	Lead with the headline and its implication. Confident, clear tone. Drop hedge words like “I think” or “I believe” — they read as uncertainty. State plainly whether something is on track or delayed, and why, without burying the problem.
Managing Across (peers, partner teams)	Collaborative framing — here’s where we are, here’s what we need from you. “We”/”I” language, peer to peer, not directive.
Managing Outward (clients, candidates)	Clear, professional, and warm — sound human. Never leave someone in a silence loop; even “no update yet, but we’re on it” beats saying nothing.
Managing Down (your own team)	Context-rich. Teams are usually told the what and the how, and left to guess the why — that’s what breaks alignment. Be honest when there’s been an escalation or a complaint rather than going quiet; the team can’t prioritise well without the full picture.

Example (from the session): A partner account manager (from Zensar) sent one dense seven-slide update to a mixed audience of business stakeholders, architects, and product owners — everyone got the same detail, meant for no one in particular. The fix isn’t a better slide, it’s splitting the send: a short summary “to” the decision-makers, full detail available for whoever wants to dig in, rather than one format forced on every audience.

In the real world: Amazon’s six-page narrative memo, introduced by Jeff Bezos, is a deliberate answer to exactly this calibration problem for managing up. PowerPoint is banned for major decisions; meetings open with 15–30 minutes of silent reading of a structured memo before anyone speaks, so every stakeholder — regardless of how much context they walked in with — engages with the same level of detail. Bezos’s stated reason: bullet points let you hide weak logic, while full sentences force the writer to expose the gaps. (CNBC)

The four-line status update

Readable in one to two minutes: status (on track / at risk / blocked / complete); what’s happening this week, framed as outcomes rather than a list of activities; what’s next and when the following update lands; and any dependency or ask specific to that particular recipient, not a generic list they have no power to act on.

One practical habit that follows from this: don’t send one dense update to a mixed audience of decision-makers and interested bystanders. Send the short version to the people who decide, and a tailored version to everyone else — that’s not duplicated effort, it’s aligning expectations properly for each group.

Three kinds of difficult conversations

Treating every hard conversation the same way is itself a mistake — the three types need different handling.

Type	How to approach it
Performance or behaviour issue	Be specific and evidence-based. Separate the behaviour from the person — “I get agitated” is a behaviour; “he’s always like that” is a personality label, and labels don’t change behaviour. Focus on impact, not judgement.
Disagreement	Usually caused by different information, not different values. Separate position from interest, and understand the other side before advocating your own — it de-escalates instead of triggering a circular argument. “Let’s agree to disagree” is often treated as a good outcome when it’s really an unresolved, slightly bitter one.
Expectation mismatch	Rarely one person’s fault cleanly — it’s usually shared. Acknowledge the gap without assigning blame, then focus forward on preventing it recurring (this is where SCOPE comes back in).

The framework: COIN

A structure for the conversation itself — deliverable in ninety seconds to three minutes before opening the floor to the other person.

Letter	In the conversation
C — Context	Why this conversation, why now — framed around the issue, not the person. Avoid loaded openers like “we need to talk” or “don’t take this personally” — they create anxiety and make it feel personal regardless of the disclaimer.
O — Observation	Specific, provable behaviour only. Rule of thumb: if it could be shown on video, it’s an observation; if not, it’s your interpretation. “You seem lost these days” is interpretation. “You didn’t participate in the last three sprint meetings” is observation.
I — Impact	The effect on the project, team, or client — never framed as a threat or consequence to the individual (“we may have to let you go” doesn’t drive improvement).
N — Next step	Be specific about what changes going forward. Don’t offload the decision back onto them with something vague like “what do you think should happen now?” — that isn’t leadership, it’s avoidance. Then pause and actually invite their response.

COIN and SBI aren’t the same tool: SBI is for giving feedback. COIN is for the conversation you’ve been putting off.

Example (built around the session’s real Observation line — the surrounding conversation is illustrative): Context — “I want to talk about how you’ve been showing up in sprint meetings the last few weeks, because I think something’s off and I’d rather ask than guess.” Observation — “You didn’t participate in the last three sprint meetings” (real, from the session — contrast with the non-observation “you seem lost these days”). Impact — “The team’s missing your input on estimates, and two calls this week went to me that should’ve gone to you.” Next step — “I’d like you back leading that part of standup from next week — what’s getting in the way of that right now?” Then pause, and actually listen to the answer.

In the real world: Kim Scott’s Radical Candor framework — “care personally, challenge directly” — has a specific origin story that’s basically a COIN conversation. After she presented strong results to Google’s founders, her boss Sheryl Sandberg pulled her aside and told her plainly that she said “um” too often and it made her sound less credible, then walked her through media training to actually fix it. Scott has called it the kindest thing a boss ever did for her — direct, specific, clearly on her side. (First Round Review)

Netflix operationalizes the same instinct at company scale with its “Keeper Test” — would you fight to keep this person if they told you they were leaving? — and a stated feedback rule (4A: Aim to Assist, Actionable, Appreciate, Accept or Discard) that frames withholding critical feedback as a form of disloyalty to the team. (Harvard Business School Working Knowledge) Worth holding against Ray Dalio’s Bridgewater, which pushed “radical transparency” further — recording meetings and logging candid critiques of colleagues — and got real results, but also real, well-documented criticism from former employees about the culture tipping into intimidation. Directness has a range, and it’s possible to go past the point where it’s still serving the person on the other end of it.

On delay

Delaying a difficult conversation doesn’t remove the problem — it compounds it, and it silently sets a new, wrong expectation that the behaviour is acceptable, because you never said anything for three months. There’s a real difference between justified delay — gathering facts before a high-stakes conversation, while still communicating something in the interim so silence doesn’t create its own anxiety — and delay driven by fear of conflict or of being wrong, which needs to be named honestly as fear rather than dressed up as diligence.

“A person with a sharp knife roaming around needs to be flagged immediately, regardless of uncertainty about impact. A person with a plastic knife can be handled more casually.”

Once you’re addressing a real problem, the reasons behind it — traffic, an outage, a bad day — are largely beside the point of solving the downstream issue it caused. Getting pulled into sympathetic “why” mode can come at the expense of actually fixing the thing. Worth holding as a general instinct, not an absolute rule.

A recurring theme, stated plainly

“You’ll get the frameworks right if you introspect properly and honestly. Ninety-nine percent chance you’re living in a lie if you believe you’re already doing everything right.”

There’s a planned gap of a couple of weeks in the programme next, with assignments to cover it. More sessions to come.

Previous: Session 3 — Psychological Safety, Stakeholders, and Managing Expectations · Back to the series index

Sources:

Self Improvement

leadership leadership-academy communication

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