There’s a particular kind of dread that settles in the night before you know you have to have a difficult conversation. You rehearse it in the shower. You draft opening lines in your head while making coffee. You tell yourself it’ll be fine, and then you spend the rest of the morning hoping the other person calls in sick.

I’ve been having difficult conversations as a manager for a long time now, and I want to be honest about something: they don’t get easier. What changes is that you get better at having them. You learn to sit with the discomfort rather than rushing through it. You learn that the conversation you’re dreading is almost never as bad as the one you’ve been having with yourself about it.

Whether it’s telling someone their performance isn’t meeting expectations, delivering news about a reorganisation that’s going to affect people’s roles, or pushing back on your own boss about a decision you think is wrong, these conversations are unavoidable. And the cost of avoiding them is almost always higher than the cost of having them.

Why We Avoid Them

Let’s start with the obvious: difficult conversations are uncomfortable. But when I started thinking more carefully about why they’re uncomfortable, I found it went deeper than just “nobody likes conflict.”

Duretti Hirpa’s work on this resonated with me. There’s a genuine anxiety that comes from the power dynamics involved. When you’re the manager, you hold structural power whether you want to or not. The person sitting across from you knows that. They might be smiling and nodding while internally catastrophising about what this means for their career, their mortgage, their sense of professional identity. Forgetting that power dynamic, or pretending it doesn’t exist, is one of the most common mistakes I see managers make, myself included in my earlier years.

There’s also the identity piece. Stone, Patton, and Heen’s work on “learning conversations” really clicked for me here. Difficult conversations threaten people’s sense of who they are. Telling someone their work isn’t good enough isn’t just feedback about a deliverable, it can feel like an attack on their competence, their value, their belonging. And it works both ways. As the person delivering the message, your own identity is on the line too. Am I being fair? Am I a good manager? What if I’m wrong about this?

Understanding that these conversations operate on three levels, what happened, how people feel about it, and what it means for their identity, changed how I approach them entirely.

Curiosity Over Blame

The single most useful shift I’ve made over the years is replacing blame with curiosity. This sounds simple, almost trite, but in practice it’s genuinely transformative.

The blame frame sounds like: “You missed the deadline and it caused problems for the team.” It’s factual, it’s direct, and it immediately puts someone on the defensive. The conversation becomes about who’s at fault, and once you’re in that territory, nobody’s learning anything.

The curiosity frame sounds like: “The deadline slipped and I want to understand what happened from your perspective.” Same situation, completely different conversation. You’re inviting them to help you understand rather than defending themselves against an accusation.

Addy Osmani writes about this, replacing blame with curiosity, and I think it’s particularly important in engineering contexts where we’re trained to do root cause analysis on systems but somehow forget to apply the same thinking to human situations. When a service goes down, we don’t start the post-mortem by blaming the person who deployed the change. We ask what happened, what the contributing factors were, what the system allowed to happen. We should bring that same discipline to conversations about performance and behaviour.

This doesn’t mean avoiding directness. You still need to be clear about what the problem is and what needs to change. But starting from curiosity means you might actually learn something. Maybe the deadline slipped because the requirements changed three times and nobody told you. Maybe the person’s been struggling with something outside work that’s affecting their focus. You won’t know unless you ask, and you won’t get honest answers if they feel like they’re being prosecuted.

Kind vs Nice

One of the distinctions that stuck with me from Camille Fournier’s writing is the difference between being kind and being nice. Nice is comfortable. Nice avoids the hard truth because it might upset someone. Nice gives a glowing performance review to someone who’s struggling because you don’t want to ruin their day.

Kind is harder. Kind tells someone the truth because they deserve to know where they stand. Kind gives them the information they need to improve, even when that information is painful. Kind respects someone enough to be honest with them.

I spent the first few years of my management career being nice. I’d soften feedback until it was unrecognisable. I’d bury the difficult message in so many caveats and compliments that the person walked away thinking everything was fine. Then I’d be frustrated three months later when nothing had changed, which wasn’t fair on them at all. They hadn’t changed because I hadn’t actually told them anything needed to change.

Being kind means being direct. It means saying “your code reviews have been superficial lately and it’s affecting the team’s confidence in the review process” rather than “maybe we could all try to be a bit more thorough in reviews.” The first version is uncomfortable but actionable. The second is comfortable but useless.

The Learning Conversation Framework

The framework that’s influenced my approach most is the “learning conversations” model. The core idea is that every difficult conversation is actually three conversations happening simultaneously:

The “what happened” conversation, the facts, the events, the observable behaviours. This is where most managers want to live because it feels objective and safe. But facts alone don’t persuade people to change. Mary Lynn Manns’ research on paths to change reinforced this for me, data is necessary but not sufficient. People don’t change behaviour because of a spreadsheet. They change when you address their beliefs and feelings about the situation.

The feelings conversation, the emotions on both sides. This is where most managers want to flee. But acknowledging feelings isn’t the same as agreeing with them. “I can see this is frustrating” doesn’t mean “you’re right to be frustrated.” It means “I recognise you’re a human being having an emotional response, and I’m not going to pretend that isn’t happening.”

The identity conversation, what this situation means for how people see themselves. This is the one that catches people off guard. When someone gets defensive in a performance conversation, they’re often not arguing about the facts. They’re protecting their sense of themselves as competent, hardworking, valuable. If you can address that directly, “I want to be clear that this conversation is about a specific pattern I’ve observed, not about your overall value to the team”, you can sometimes defuse the defensiveness before it takes hold.

Performance: Ongoing Conversation, Not Annual Event

One thing I’ve learned the hard way is that performance conversations should never be a surprise. Ines Sombra’s framing of performance as an ongoing conversation rather than a periodic event changed how I think about this.

There’s a useful distinction between event-based and systemic underperformance. Someone missing a deadline because they underestimated a task is an event. Someone consistently missing deadlines because they can’t prioritise effectively is systemic. The first needs a quick conversation and maybe some support. The second needs a structured approach, clear expectations, regular check-ins, documented progress.

The “no surprises” rule is something I hold myself to rigidly now. If someone is put on a performance improvement plan and they’re genuinely shocked, I’ve failed as their manager. They should have heard the feedback weeks or months earlier. They should have had the opportunity to course-correct before it became formal.

This is where documentation matters, and I don’t mean documentation as a paper trail to protect yourself legally, though it does serve that purpose. I mean documentation as a tool for clarity. Writing down what was discussed, what was agreed, and what the expectations are going forward removes ambiguity. It protects both parties. And it forces you to be precise about what you’re actually asking for, which is harder than it sounds.

When the Conversation Is About Leaving

Sometimes the difficult conversation is about helping someone recognise that they’d be better off somewhere else. Fournier writes about “coaching someone out” and it’s one of the most nuanced things a manager can do.

This isn’t about being manipulative. It’s about being honest when someone’s skills, interests, or career goals don’t align with what the role or the organisation can offer. “I think you’d thrive in an environment that values X more than we’re able to right now” is a genuinely kind thing to say if it’s true.

The framework I find useful here comes from thinking about difficult relationships through the lens of Exit, Voice, Loyalty, and Optionality. Sometimes the right answer is voice, speaking up and trying to change the situation. Sometimes it’s loyalty, accepting the situation and committing to making it work. And sometimes it’s exit, recognising that the best path forward is a different path entirely. Part of your job as a manager is helping people see all of those options clearly, not just the ones that are convenient for you.

Delivering Bad News

Reorganisations, redundancies, project cancellations, these are conversations where you’re delivering news that’s going to affect people’s lives and you often had limited input into the decision yourself.

Fournier’s advice here is straightforward and I’ve found it holds up: talk to people individually wherever possible, don’t force yourself to deliver messages you fundamentally can’t support, and be honest about what you know and what you don’t.

The temptation is to spin. To make the reorganisation sound like an exciting opportunity. To frame the redundancies as “right-sizing.” People see through this instantly, and it destroys trust. Better to say “I know this is difficult news, and I want to be straight with you about what I know and what I don’t” than to pretend everything is wonderful.

One thing I’ve learned is to set context early. Don’t let people spiral through five minutes of anxiety while you build up to the point. “I need to talk to you about some changes to the team structure” in the first thirty seconds is far kinder than ten minutes of small talk followed by a bombshell.

Preparation Isn’t Overthinking

I used to think that rehearsing a difficult conversation was a sign of weakness, that a good manager should be able to handle these things naturally. I was wrong. Practising what you’re going to say, anticipating how the other person might respond, thinking through the key points you need to cover, that’s not overthinking, it’s preparation.

I’ll sometimes write down three things: the core message I need to deliver, the feeling I want the other person to leave with, and the one thing I absolutely cannot leave unsaid. It takes five minutes and it stops me from rambling, softening the message into oblivion, or forgetting the most important point because I got flustered.

The conversation will never go exactly as you planned. That’s fine. The preparation isn’t a script, it’s a compass. It keeps you oriented when the conversation takes unexpected turns, which it will.

The Conversation After the Conversation

The difficult conversation doesn’t end when the meeting does. What happens in the days and weeks afterwards matters just as much. Following up, checking in, demonstrating that the relationship hasn’t been damaged by the honesty, these are the things that build trust over time.

People remember how you made them feel far more than what you said. If you can have a difficult conversation and leave the other person feeling respected, heard, and clear about what happens next, even if the content was painful, you’ve done your job well.

I won’t pretend I get this right every time. I still occasionally avoid conversations longer than I should, still sometimes soften the message too much, still sometimes misjudge the emotional temperature of the room. But I’ve learned that the willingness to have the conversation at all, imperfectly, awkwardly, humanly, is what separates managers who build trust from those who don’t.

The difficult conversation you’re dreading right now? Have it. It won’t be as bad as you think. And even if it is, it’ll be better than the alternative of not having it at all.