I've been asked to give someone difficult feedback and I've been putting it off.

What to do when you've been putting off difficult feedback

Los Gatos, mid two-thousands. Inside Netflix, the managers have started noticing something the company's HR system was not designed to surface.

The expensive part of avoiding the difficult conversation isn't the conversation itself. It is the slow accumulation of people who are technically adequate but who aren't lifting the team — people who stay because no one wants to fire them, and who eventually leave on someone else's schedule rather than their own.

Patty McCord and Reed Hastings, working through the underlying machinery, name the discipline they want managers to run. The Keeper Test. A prompt managers are required to ask themselves on a routine cadence about everyone who reports to them. If this person told me tomorrow they were leaving for another company, how hard would I fight to keep them?

If the honest answer is not very hard, the prescribed action is to let them go immediately. Generous severance. No formal performance improvement plan. No theatre of warning letters and rehearsed conversations everyone in the room can already see the end of.

The system Netflix builds around this prompt is uncomfortable to read about. Plenty of people have, in the years since, recoiled from the coldness of the formulation. The underlying point, though, isn't the coldness. It's the recognition that the manager's reluctance to have the difficult conversation is almost always more expensive for the team than the difficult conversation would be. The cost just sits in a different ledger.

When you've been told to give someone hard feedback and you've been putting it off, the question Netflix's discipline forces is whether the conversation you're avoiding is the one the team is already paying for.

So let's go to the office and work through it.

Name what the avoided conversation is already costing the team

"I've been asked to give someone difficult feedback and I've been putting it off."

The feeling is anxious.

You know what needs to be said. You have known for weeks. Possibly months. And every week, the moment when it would have been easier passes, and the conversation gets a little harder than it would have been the week before.

Two choices. They look like the same conversation. They need different first moves.

When they will defend their status

Choice one: the person you're about to talk to is going to defend their status. Their identity is tied up in being seen as good at this. Direct feedback will trigger the defence before the message lands, and you'll spend the next hour managing their composure rather than addressing the work.

If that's the read, don't deliver the feedback as feedback. Ask them for advice. Describe a hypothetical team member — someone who, by sheer coincidence, sounds exactly like them — and ask what they'd do. Not as a trick. As a route in.

Edgar Schein, working on what he called humble inquiry, found that people are almost incapable of hearing criticism of themselves and startlingly clear-eyed about the same behaviour described in the third person. Whatever advice they give you about the hypothetical person, they have given to themselves. The defensive apparatus has been sidestepped, and the message has landed.

When they will defend their reasoning

Choice two: the person you're about to talk to is going to defend their reasoning. They will argue the logic of every point. The feedback will turn into a debate you cannot win, because the framing is adversarial and the format is set up to favour whoever talks last.

If that's the read, deliver it on a walk. Not across a desk. Side by side, outside, walking.

Jonathan Haidt's rider and elephant model is doing the work here. Rational argument rides on top of intuitive response, and the intuitive response to a face-to-face conversation across a table is that something is being done to you. The intuitive response to a side-by-side walk is that something is being worked through with you. Same words, different posture. The argument doesn't go away. It just stops being the centre of gravity.

Status, or reasoning. Same difficult conversation. Two different first moves.

How to make hard conversations ordinary

Three tools. The discipline is to make the difficult feedback routine rather than rare — so that no single conversation is carrying a year of unsaid things.

Distribute the conversation across the year

The first is

Performance Management — but in a specific shape.

Performance Management as a corporate practice traces back to General Electric in the nineteen-eighties under Jack Welch and the imported framework of stack ranking; the contemporary practice that survived the twenty-tens repudiation of forced ranking is closer to Esther Derby's facilitation lineage and the agile retrospectives tradition than to the Welch-era discipline.

The reason Performance Management matters here is that the terror of the annual review is mostly a function of cadence. The annual review is terrifying because it is the first time anything is being said. A year of small things have built up unsaid, and the conversation has to clear the year in an hour. Of course it goes badly.

The unique insight is that the discipline is the rhythm, not the ceremony. Weekly or fortnightly fifteen-minute conversations — what's working, what's not, what would help — distribute the conversation across the year. No single meeting carries a year of unsaid things, because nothing has been left unsaid for more than a fortnight.

What you get is feedback that sounds like a conversation rather than a verdict. Difficult feedback inside a routine cadence is a routine event. Difficult feedback inside an annual review is a crisis.

So. How to run it.

Cadence. Every direct report, fifteen minutes, weekly or fortnightly. Same slot. Protected. The point isn't the duration — it's the regularity. A fifteen-minute conversation that always happens beats a sixty-minute one that mostly doesn't.

Questions. What worked this fortnight? What didn't? What would help? Same three. Every time. The repetition is the discipline; the answers change.

Surface. The whole point is that nothing is allowed to grow large by being unsaid. If the answer to what didn't work is I'm not sure how to say this, you say it.

Note. Three lines after each session. Not for the file — for the next session. So the conversation builds.

When the formal review arrives, it is the documentation of conversations that have already happened. No surprises. No accumulated terror. The review becomes administrative rather than confrontational, which is the only condition under which formal reviews ever do useful work.

Diagnose whether it's capability, opportunity or motivation

The second is

Com-B Model.

Com-B comes from Susan Michie at University College London, codified in the twenty-eleven paper The Behaviour Change Wheel: A New Method for Characterising and Designing Behaviour Change Interventions. Its lineage runs through public-health behaviour change research, where the discipline of asking which kind of problem is this? before prescribing an intervention is older than most management literature.

The reason com-B matters before a difficult feedback conversation is that most feedback fails because the manager has misdiagnosed the problem. The team member can't do the thing — but is being trained as if they don't want to. Or they don't want to — but are being supervised as if they can't. Wrong diagnosis, wrong intervention.

The unique insight is the three-letter split. Capability, Opportunity, Motivation, leading to Behaviour. Capability — they don't have the skill. Opportunity — the environment is preventing them. Motivation — they don't want to. The three are not interchangeable, and the intervention that fixes one cannot fix the other two.

What you get when you run the diagnostic before the conversation is feedback that lands on the actual problem. Training a motivation problem doesn't fix it. Supervising a capability problem doesn't fix it either. The diagnostic tells you which kind of conversation you're about to have.

So. How to use it.

Capability. Ask: can they do this? If a more senior or more experienced person did the same task, would they get the same result? If yes, you're looking at a capability gap. The intervention is teaching, pairing, or coaching — not feedback about effort.

Opportunity. Ask: is the environment preventing them? Are they being interrupted, blocked, denied resources, given conflicting priorities? If the answer is yes, you're looking at an opportunity gap. The intervention is removing the obstacle — not feedback about output.

Motivation. Ask: do they want to? If they have the skill and the environment is clear, the gap is motivation. The intervention is the conversation about what's actually going on for them — not training, not extra resources, but the harder conversation about whether the work is what they want to be doing.

The discipline is to do the diagnostic before the conversation, not during it. Walking in with the wrong diagnosis is how feedback conversations turn into the kind of meeting both sides remember as the time the manager didn't get it.

Let the feedback be about the work, not the person

The third is

Kano Model.

Noriaki Kano developed the Kano Model in the late nineteen-seventies at Tokyo University of Science, formalising it in a nineteen eighty-four paper that gave the framework its diagnostic shape. Originally a customer-satisfaction tool from quality-management practice, it has since become a product-prioritisation framework whose deeper utility is as a diagnostic for the question which features matter to the people you're building for, and which ones don't.

The reason Kano matters here is that some difficult feedback conversations aren't actually about the person. They're about prioritisation. The person is over-investing in features users don't care about, and the conversation that needs to happen is about the work, not about them. Pretending the work conversation is a personality conversation is what makes the feedback land badly.

The unique insight is the three-band split. Must-haves, Performance attributes, Delighters. Must-haves — features users notice when absent and don't notice when present. Performance attributes — features whose quality scales linearly with satisfaction. Delighters — features users don't expect and respond to disproportionately.

What you get is a way to diagnose the over-investment without making it personal. The conversation becomes this work was sitting in the Delighter band but the user is missing a Must-have rather than you've been spending time on the wrong things. Same content, different surface. The work has been put on the table, not the person.

So. How to run it.

Map. Across a single page or whiteboard, list what the person has been working on. One line per item.

Classify. Must-have, Performance, Delighter. Be honest. Most teams discover the bulk of recent work has been Delighter while the user has been complaining about a Must-have.

Surface. Here's where the work has been going. Here's where the user is feeling the pain. The gap is the conversation, and the gap is structural rather than personal.

Re-prioritise. What moves into the Must-have band next sprint? What gets deferred? The work conversation lands cleanly because the framework, not the manager, is doing the prioritising.

The Kano Model isn't a feedback tool. It's a tool that lets the feedback be about the work rather than about the person — which is, often, the only kind of feedback the person can hear.

A precedent: the texture of hard news can be practised

That's the toolkit. One more story before we close.

The Netflix Keeper Test we opened with describes the discipline at the cold formulation — bald, structural, designed to force the conversation by making it impossible to avoid. The story we close with describes the discipline at the human end — what it looks like when the difficult conversation is delivered without either side losing dignity.

Wellington, October two thousand and seventeen onwards. Jacinda Ardern is sworn in as Prime Minister of New Zealand at thirty-seven, leading a coalition government and arriving in office having been Labour leader for less than two months.

The discipline she practises in how she delivers hard news is specific, and unusual in political communication: direct, emotionally literate, without the softening language most political careers are built on. She names the difficult thing plainly. She acknowledges the difficulty of it. And then she expects the conversation to continue.

The pattern shows up most visibly in March two thousand and nineteen, after the Christchurch mosque shootings. Within the week, she refuses to name the attacker — explicitly, on the floor of Parliament. She meets the bereaved in person. She announces the law change. New Zealand's gun laws are amended within weeks of the attack.

It shows up daily during the pandemic — go hard, go early — without the political hedging the moment seemed to invite from peers in other countries.

It shows up inside her cabinet, where former colleagues have since described how Ardern would name the difficult thing plainly, acknowledge the difficulty of it, and then keep working. Not announce a position and exit. Not soften the position into ambiguity. Hold the position, and stay in the room while the room caught up.

She steps down in January two thousand and twenty-three — I no longer have enough in the tank — naming her own limit with the same discipline she had used to name everyone else's. The exit is the most documented application of the practice.

So.

The deeper point: practise how it's said, not just what

The Keeper Test works because the bald formulation forces the conversation — the test is whether the manager would fight to keep this person, and the answer being honest is what makes the discipline hold. Ardern's practice works for a different reason. It demonstrates that the difficult conversation can be delivered without either side losing dignity, which is often the thing the manager has actually been avoiding. Not the content. The texture.

When you've been putting off difficult feedback, the question to ask is whether what you're avoiding is the content or the texture of how it would have to be said. Ardern's example is that the texture is the part that can be practised — and that the practice is what the formal cadence is for.

So. Your Next Move from this playbook.

What feedback are you saving for the formal review that the person should have heard six weeks ago — and whose comfort are you protecting by waiting?