Managing a team · Managers
Praise that means something, feedback that lands, and the conversations most managers put off for a week longer than they should.
10 templates · free · no sign-up
Sending it as…
Name the behaviour and the effect. 'Great job' teaches nothing.
[Name] — the way you handled [specific situation] was excellent. Specifically: [what they did]. The result was [effect on the team, customer or project]. That's exactly the judgement I'd want on anything difficult. Thank you.
Observation, effect, request. No sandwich.
[Name], can I give you some feedback on [situation]? What I noticed: [specific, observable behaviour — not a character judgement]. The effect: [consequence]. What I'd like instead: [concrete alternative]. How does that land with you? Happy to hear if I've read it wrong.
Give them the agenda in advance so it's their meeting, not yours.
Hi [Name] — our one-to-one is [day] at [time]. It's your time, so bring whatever's on your mind. Things I'd like to cover if there's room: • [Topic] • [Topic] Anything you want added, drop it in here beforehand.
Ask about the blocker, not the delay.
Hi [Name] — how's [task] looking for [date]? If something's in the way, tell me now and I'll help clear it rather than find out on the day.
Before they burn out, not after they resign.
Hi [Name] — you've had a lot on lately: [list what you've noticed]. I'd rather we drop or delay something than have you carrying all of it. If you had to take one thing off the list this week, what would it be? No wrong answer here.
Send it the morning they start. First impressions compound.
Welcome, [Name]! Really glad you're here. Your first week, roughly: • Today: [what] • This week: [what] • By Friday: [realistic first win] [Buddy name] is your go-to for anything day-to-day. My door's open for everything else — and no question is too basic in your first month.
State the outcome and the constraints, then get out of the way.
[Name], I'd like you to take [task]. **What good looks like:** [outcome] **By:** [date] **Constraints:** [budget, tools, anyone to involve] **Decisions you can make alone:** [scope] — come to me only for [exceptions] How you get there is yours. Shout if you want a sounding board.
Explain the reasoning or they stop bringing you ideas.
[Name] — thanks for putting [idea] together, genuinely. We're not going to do it this quarter. The reason is [honest constraint: priorities, cost, timing], not the quality of the thinking. Keep it — if [condition changes] it becomes a much easier yes. And keep bringing me these.
Fix first, learn second, blame never.
[Name] — thanks for flagging this quickly, that's the right instinct. Right now: [immediate fix, who's doing what]. Once it's stable, let's work out what made it possible. I'm interested in the gap in the process, not in whose hands were on it.
Say the decision first. Burying it reads as evasion.
Subject: A change to [thing]
Team, From [date], [the change]. Why: [the honest reason, including the trade-off you accepted]. What it means day to day: [specifics]. What isn't changing: [reassurance, where true]. I know [the part people will dislike] is unwelcome. I'm around [when] to talk it through — and if there's something I've missed, I'd rather hear it than not. [Your name]
Nothing matches that
Try a plainer word, or clear the filters above.
Useful feedback has three parts: what you actually observed, what it caused, and what you would like instead. Skip any one and it fails. Observation alone sounds like nitpicking, effect alone sounds like blame, and a request without either sounds arbitrary.
The first part is an observation, not an interpretation, and this is where most feedback goes wrong. The report arrived after the deadline twice can be discussed, corrected or explained. You are disorganised can only be defended against, because it is a claim about who someone is rather than what they did. Keeping to observable behaviour is not a softening technique; it is what makes the conversation solvable, because behaviour can change and character is not on the table.
Burying criticism between two compliments is so widely used that people hear the praise as packaging and discount it. Worse, the actual message often does not register at all, and the manager leaves believing they delivered it.
Give praise when it is earned, on its own, and give corrective feedback on its own too. Both are taken more seriously when they are not being used to cushion each other. The reason the sandwich persists is that it makes the manager more comfortable, not the recipient. If the conversation feels difficult, that discomfort is the cost of doing it properly, and it is considerably smaller than the cost of someone discovering at a performance review that there had been a problem for eight months.
Great work tells someone you are pleased but not what to do again. Name the behaviour and its effect — the way they handled that particular call, and what it meant for the customer — and you have turned a nice moment into a repeatable instruction.
Specific praise is also much harder to dismiss as politeness, which is why it lands so much harder. People discount general compliments almost automatically, especially from managers, because they assume it is a management technique. A detailed observation cannot be discounted that way; it is evidence that you were paying attention. Public praise amplifies this, but check first, since some people find it excruciating and the gesture then costs more than it gives.
Corrective feedback works best within a day or two, while the specifics are still shared and correctable. Waiting a month means arguing about what happened before you can discuss what to change.
The exception is when either of you is angry. Feedback delivered while irritated tends to include the accumulated evidence of every previous instance, which turns a small correction into a case for the prosecution. A few hours is usually enough to separate the point from the frustration. Never correct someone publicly. Whatever the intention, an audience turns feedback into humiliation, and the recipient will remember the setting long after they have forgotten the substance.
The most common failure in one-to-ones is that they become status updates, which is a meeting the manager needs rather than one the report benefits from. Status can go in writing; the conversation should be for the things that do not fit in writing.
Send the agenda in advance and make clear it is their time, with your items as optional extras. Protect the slot, because cancelling repeatedly communicates a priority ranking more clearly than anything you say. Weekly or fortnightly works for most teams. And leave silence. The useful part of a one-to-one often arrives after a pause that the manager was tempted to fill, and learning to wait through it is most of the skill.
Handing over work badly produces either someone paralysed by unclear expectations or a manager who takes it back halfway through. Both are avoidable by being explicit about three things: what good looks like, the constraints, and which decisions they can make alone.
That third one is the one most managers skip, and it is the one that determines whether they are actually free to act. Say what they can decide without asking and what needs to come back to you, and then honour it. Checking in on progress is fine; reversing decisions you delegated is not, and doing it once will teach the person to ask permission for everything thereafter. The point of delegation is not to move the task, it is to move the judgement.
People rarely announce that they are drowning, particularly the conscientious ones, who tend to absorb more until something breaks. The signals are usually behavioural: work arriving later at night, a drop in quality on things they normally do well, going quiet in meetings, or an unusual number of small mistakes.
The intervention is a direct question with a concrete out. Naming what you have noticed and asking what they would drop if they had to gives permission that a general how are you getting on does not. Then actually remove something. A conversation about workload that ends with nothing changing is worse than no conversation, because it establishes that raising it does not help and they will not raise it next time.
The first response to an error sets whether people bring you the next one early or hide it until it is expensive. Thank them for flagging it, deal with the immediate problem, and leave the analysis until things are stable.
When you do the analysis, focus on what made the mistake possible rather than who made it. Almost every significant error has a process component: a missing check, an unclear handover, a permission that should not have existed. Fixing that prevents recurrence in a way that a conversation about carefulness does not. Managers who react sharply to mistakes get fewer reports of mistakes, not fewer mistakes, and the gap between those two things is where serious problems grow.
A performance conversation should never be the first time someone hears there is a problem. If it is, the failure is managerial, and the person is entitled to be shocked.
Ongoing feedback is what makes the formal conversation survivable. When it does come, be specific about the gap between what is happening and what is required, be clear about what needs to change and by when, and be honest about the consequence if it does not. Vagueness here is a kindness that is not kind: someone told they need to step up without specifics cannot act on it. Write down what was agreed and send it, because both parties will remember this conversation differently and the written version protects everyone.
How you decline ideas determines whether you keep getting them. A dismissive no, or worse a vague maybe that quietly dies, teaches people that thinking beyond their immediate work is wasted effort.
Explain the actual reason, which is usually priorities or timing rather than quality. Say what would have to be true for it to become a yes, because that turns a rejection into a condition they can watch for. And keep the idea somewhere real. Coming back six months later to say that the thing they proposed is now viable is one of the strongest signals a manager can send that ideas are genuinely wanted, and it costs nothing but a note.
The instinct to soften bad news by burying the decision is universal and counterproductive. People skim for the substance, find it in paragraph four, and conclude you were trying to hide it — which damages trust more than the change itself.
Lead with the decision. Then give the honest reason, including the trade-off you accepted, because a change presented as costless is not believed. Say what is not changing, where that is true, since most anxiety is about scope rather than substance. Then make yourself available. The announcement is not the end of the process; the conversations afterwards are where people actually make their peace with it, and a manager who disappears after sending the email is the one who gets the worst of it.
Managers who only give feedback create a strange asymmetry, and the quality of what comes back to them degrades quickly. Asking for feedback works only if the first few responses are handled well, because everyone is watching how the first honest answer is received.
Ask narrowly rather than generally. What should I do differently in our one-to-ones invites a real answer; do you have any feedback for me does not. When you get something uncomfortable, thank them, ask a clarifying question, and resist explaining why they are wrong even when you believe it. Then change something visible and say that you did. One acted-upon piece of feedback does more for the flow of information than a year of inviting it.
Anything significant should exist in writing, not because of process but because memory diverges. A short note after a feedback conversation — what was discussed, what was agreed, by when — takes three minutes and prevents the far longer conversation about what was actually said.
This matters most in the cases you least want to write up. Verbal-only performance conversations are the ones that later become disputed, and the version that stands is the one that was documented at the time. Keep it factual and free of characterisation, and send it to the person rather than only filing it. Feedback that someone cannot re-read is feedback they will remember selectively, usually as either harsher or vaguer than it was.
Upward feedback is the most avoided and often the most valuable, because senior people receive very little of it and the quality of their information degrades accordingly.
The technique is the same — observation, effect, request — but the framing matters more. Ask permission first, which gives them a moment to switch out of transmit mode. Anchor it in a shared goal rather than a personal preference, so it reads as being about the work. And be specific about what you would like instead, because a senior person hearing a vague complaint will usually agree pleasantly and change nothing. If they respond badly, that is information about them rather than a reason not to have tried, though it is reasonable to calibrate what you raise next time.
