Difficult conversations for new managers: the complete guide

Most management advice about hard conversations is written for people who have had a hundred of them. This is written for your first.

The conversations that go badly rarely go badly because the manager chose the wrong words. They go badly because the manager delayed until the problem was undeniable, buried the message under so much cushioning that it never landed, or made a promise in the last thirty seconds to escape the discomfort. The words matter far less than the structure.

What makes a conversation "difficult" in the first place?

A conversation is difficult when you expect the other person to be unhappy with what you say, and you have to say it anyway. That is the whole definition. It covers telling someone their work is not good enough, telling someone they did not get the promotion they expected, raising a behaviour that is affecting the team, and asking someone whether they are coping.

What those have in common is a gap between what the other person believes and what you are about to tell them. The size of that gap predicts the difficulty far better than the seriousness of the topic. Telling someone their work has slipped is straightforward if they already know. It is hard if they think they are doing well.

That has a practical consequence. Most of the work happens before the meeting, in narrowing the gap: regular feedback, clear expectations, no surprises. The conversation is difficult in proportion to how much you have avoided until now.

How do you open without softening the message?

Say the thing in the first thirty seconds. Not the third minute.

New managers almost always open with warm-up: how are you, how was the weekend, how is the project going. Then a bridge: "so, I wanted to chat about something." Then a hedge: "it's not a big deal, but." By the time the actual message arrives, the other person has spent two minutes working out that something is wrong and bracing for it. The warm-up did not soften anything. It extended the dread.

A working opening has three parts and takes about fifteen seconds:

1. Name the topic. "I want to talk about the client report that went out on Tuesday."
2. State your position. "It had errors in it that we've talked about before, and I need that to change."
3. Hand over. "Tell me what happened."

That structure is blunt on the page and lands as respectful in the room, because it treats the other person as someone who can handle information. Softening reads as kindness to the person delivering it and as evasion to the person receiving it.

What makes a reason specific enough to act on?

A reason is specific enough when the other person could use it to do something differently tomorrow.

"Your communication needs work" fails that test. So does "you're not quite at the level yet" and "there were some concerns." These are summaries of a judgement, not the evidence behind it. The other person cannot act on a summary, so they either dismiss it or invent a worse version of it.

Compare: "In the last three client calls, you answered questions about scope before checking with delivery, and twice we had to walk it back the next day." That is a behaviour, a frequency, and a consequence. It can be argued with, which is exactly what makes it useful.

Prepare two or three examples before the conversation. Not ten, which reads as a case being built against someone. Two or three that you are certain of, with dates.

What do you do with the silence after you say it?

Nothing. That is the skill.

The silence after a hard message is usually three to ten seconds. To the person who just spoke it feels much longer, and the instinct is to fill it, which is where the damage happens. Filling the silence is how "your work has slipped" becomes "your work has slipped, but honestly everyone's been stretched and it's probably not entirely your fault." The message survived about four seconds.

Count to ten in your head before you say anything else. The other person is processing, and what they say first tells you which conversation you are actually in: a disagreement about facts, a disagreement about significance, or an admission you did not expect.

If the reaction is defensiveness, that is a distinct conversation with its own moves — what to do when someone gets defensive covers the specific responses. If two people are involved and the problem is between them rather than with you, see how to mediate conflict between two team members.

How do you handle a reaction you were not expecting?

The three reactions that derail new managers are anger, tears, and silence. All three are easier once you accept that your job is not to stop them.

Anger is usually about fairness rather than fact. The response is to separate the two: "I can hear that you think this is unfair. I want to hear why. The facts I've described, though — do you disagree with any of them?" You are not conceding the judgement, but you are treating the objection as real.

Tears need one thing: a pause without a retreat. "Take a minute." Then wait. Do not apologise for the message, because apologising for it tells them the message was a mistake, and you will have to deliver it again later. Offer to continue in an hour if they want that.

Silence is the hardest because it invites you to keep talking. Ask directly: "I've said a lot. What's your reaction?" If nothing comes, name the option: "We can pick this up tomorrow if you'd rather think first." Then actually stop.

How should a difficult conversation end?

With a commitment, out loud, that both of you could write down the same way.

The most common failure in the last two minutes is drift. The manager, relieved that the hard part is over, says something warm and vague — "let's keep talking about it", "I'm here if you need anything" — and the conversation ends with no agreed action. Three weeks later nothing has changed and the manager is surprised.

End with three sentences:

• What is different from now on, in behaviour terms
• When you will look at it again, with a date
• What support you are offering, specifically

Then send it in writing the same day, short and neutral, so there are two records rather than two memories. Where the conversation was about performance and there is a formal process attached, the underperformance conversation sets out how that written follow-up should look.

Which conversations should you never have on the spot?

Anything with a consequence attached. Anything you have not checked. Anything you are angry about.

Reacting in the moment is right for small course-corrections — an interruption in a meeting, a missed handover. It is wrong when the conversation implies a decision, because you will make the decision under pressure and either overstate it or promise something you cannot deliver.

The tell is if you find yourself saying "we might need to look at whether this role is right for you" without having thought it through. That sentence, once said, cannot be taken back, and it changes what the person does that afternoon.

Give yourself an hour. An hour is enough to pick two examples and decide what you actually want to happen.

How do you get better at these without practising on your team?

Deliberately, and not in the room.

The problem with learning management conversations on the job is that the feedback loop is broken. You have the conversation, it goes badly, and you find out three months later when the person resigns. Nobody tells you which sentence lost them, because the person on the other side has every incentive not to.

The two things that reliably work are rehearsal and review. Say the opening out loud before the meeting — not in your head, out loud, because the sentence that reads fine silently often collapses when spoken. And afterwards write down what you actually said versus what you planned to say. The gap between the two is where your habit lives, and it is usually the same habit every time.

That is the reasoning behind why rehearsing a conversation works, which goes into what the research on deliberate practice does and does not support here.

Where should a new manager start?

With the conversation you are currently avoiding.

There is always one. It is usually small, and it has usually been small for several weeks, which is precisely what makes it a good place to start: low stakes, and you already know what you would say if you were braver about it. Have that one this week, using the structure here, and pay attention to which part you found hardest. Almost everyone finds the same part hardest — the silence — and knowing that about yourself is worth more than any script.

The one most new managers meet first is the promotion that did not happen, usually because it arrives on a fixed calendar rather than when you feel ready — how to tell someone they are not getting promoted is the playbook for it.

The specific conversations have their own shapes and their own traps. Each of the playbooks covers one: the promotion that did not happen, the performance that has slipped, the person who is burning out, the targets that were missed, the conflict between two people, the high performer who is difficult to work with. Read the one you need the week you need it, not before.