28 October 2026 · 14 min read

How to give feedback: point it at the work, and ask for some back

Feedback helps on average. That sentence is why people give more of it, and it hides the part that matters.

In the research, feedback sometimes made performance worse. What seemed to separate the two outcomes was where the feedback sent people's attention: to the work, or to themselves.

The box Lorna fills in

Lorna leads a design team of six in a company of about eighty. For a month she has watched Bilal send client files out with last month's version number on the cover. Someone downstream catches it every time. Nobody has said a word to him.

She did try once. Last spring she told him he could be "a bit careless with detail", and he spent the next week polite, brief and a long way off. She has not tried since. The annual review is in three weeks, and the form has a box for exactly this.

So the box is where it ends up: Attention to detail - needs improvement. Bilal reads it standing at his desk. He does not read it as a note about version numbers. He reads it as a note about him, and by lunch he has a list of every time the brief changed under him.

Lorna and Bilal are invented. The research on how feedback lands is not.

Does feedback actually improve performance?

In a meta-analysis of 607 effect sizes from 131 papers, feedback interventions improved performance on average (d = 0.41), but 38% of the effects were negative; once one research group's 91 laboratory effects were set aside, 33% still were (Kluger and DeNisi, Psychological Bulletin, 1996).

Read that narrowly. A d of 0.41 is an effect size, about four-tenths of a standard deviation in performance. A "feedback intervention" here means an outside party giving someone information about how they did at a task, and most of the studies were experiments, not offices.

The authors found the benefit shrank as the feedback pulled attention toward the self and away from the task. Whether the feedback was positive or negative did not explain the pattern.

That last sentence is the one most advice ignores. Nearly all of it is about how gentle to be: soften it, balance it, warm it up. In this review, good news and bad news were not what told the two outcomes apart. What shrank the benefit was feedback that pulled attention onto the self.

Why do people get defensive about feedback?

A note about the person

Lorna's note could have been about two things: three covers, or Bilal. The box chose Bilal. "Needs improvement" rates a person, so attention goes where that review found the benefit shrinking, onto the self and off the task.

Why a low rating gets argued with

A second study fits the list Bilal wrote before lunch. Brett and Atwater looked at 360-degree feedback in a leadership-development setting. Less favourable ratings went with believing the feedback was less accurate and with negative reactions. Both went with finding it less useful, and people who found it less useful were seen by a facilitator as less focused on their development (Journal of Applied Psychology, 2001). The authors say this questions the widely held assumption that negative or discrepant feedback motivates change. These are associations from one kind of setting.

Three kinds of note

Douglas Stone and Sheila Heen, in their book Thanks for the Feedback, separate three kinds of note: appreciation (thank you, I noticed), coaching (here is a better way) and evaluation (here is where you stand). Notes often land badly because the giver meant one and the receiver heard another. Lorna meant coaching. The box delivered an evaluation.

Why asking is the missing half

The other half is asking. A meta-analysis of feedback-seeking found that organisational tenure, job tenure and age were negatively related to seeking feedback, and that frequent positive feedback, a high-quality relationship and a transformational leadership style were positively associated with it (Anseel, Beatty, Shen, Lievens and Sackett, Journal of Management, 2015). Those are associations, and the same review found the link between seeking feedback and performance small, so I will not claim that asking makes anyone better. My reading: asking looks cheaper in a relationship that is already good, which is a reason for the person with more power to go first.

Resistance to feedback: how a team goes quiet

Bilal's list is only half of it. The other half is Lorna's silence. She tried once, it cost her a chilly week, and she stopped. Run that across a team and the notes do not disappear. They move. Everyone in the design team could tell you about the version numbers, over lunch, and the one person who could use the note hears it a year late, softened into a box.

You can see this happening without a survey:

• The praise is public and the real notes are private. Reviews read well, and the useful comments travel in side conversations.
• People flinch at "can I give you some feedback?" You can watch the shoulders go up.
• The same point returns every year. The form says it again because nobody said it between reviews.
• New joiners stop commenting. They work out in weeks what a quiet first month is rewarded with.
• The lead hears last. Notes run down the line, so the person with the most influence has the least idea how their habits land.

Where it starts: the first hard note

Tunde is five months into his first job. On a Wednesday his manager tells him the client summary was too long. In three years at university nobody ever stood at his desk to say anything of the kind, and he says "okay", goes quiet and rewrites the whole thing overnight without asking which part. His manager reads the quiet as sulking and decides to be more careful next time. Neither of them will mention it again.

Business leaders tell a bleaker version of the same story. In a 2024 survey of 966 US business leaders, 75% said some or all of that year's graduate hires were unsatisfactory, 60% said some were fired, and struggling with feedback was a reason in 38% of cases (Intelligent.com, August 2024, updated 13 September 2024).

That comes from the people doing the hiring, not the people hired. My reading: "struggles with feedback" is a rating of a person, the same move the box made on Bilal. It says nothing about what the note was, whether it could be checked, or whether anybody had ever shown the new starter what to do with one. Tunde could have asked which part. Nobody had told him that was allowed.

Why annual reviews and the feedback sandwich miss

There are two easy answers.

The first is the formal route: the annual review, the 360. The CIPD found that 57% of employees with a line manager agreed that their manager provides useful feedback on their work, 21% neither agreed nor disagreed, and 20% disagreed (CIPD, Good Work Index 2026, July 2026). That is how employees rate their manager's feedback, not what they did with it. In the same survey, 15% of employees with a line manager never meet them to discuss performance, and nearly half of those (49%) found that adequate.

Lorna has a reason to wait. Acas's 2025 survey found that capability and performance was the commonest topic of conflict, named by 38% of the people who had experienced conflict at work, and 94% of the 252 managers who had a conflict with a direct report named it (Acas, research by the National Centre for Social Research, November 2025). That is the managers' account, not their reports'. My reading: when the commonest topic of conflict is how somebody is doing their job, putting the conversation off looks like a forecast of a fight.

A bigger instrument does not fix the aim either. Smither, London and Reilly pooled 24 longitudinal studies of 360-degree feedback and found that improvement in ratings over time was small, with corrected mean effect sizes of .15 for direct-report ratings, .05 for peers and .15 for supervisors, against -.04 for self-ratings (Personnel Psychology, 2005). For every rater source except supervisors, the 95% confidence interval included zero. Those are other people's ratings of change, not a measure of whether anyone accepted what they heard.

The second easy answer is the sandwich: praise, the hard part, more praise. My reading: it adjusts the one thing Kluger and DeNisi found did not explain the pattern, whether the feedback is positive or negative, and leaves alone the one that did, what the feedback points at. I am not citing a study of the sandwich itself.

How to give feedback that points at the work

None of this needs a product. These are applications of the findings above, not things any of the studies tested.

Make the subject of the note a thing

Before you speak, finish this sentence: "What happened to the work was..." If you end up writing "careless" or "defensive", you have written a rating of the person. "Three files went out with last month's version number on the cover" is a note about the work, and the person can check it against memory.

Ask for a note first, about one piece of your own work

Pick last week's brief or Tuesday's review and ask one person: "What is one thing that would have made that better?" Make your first sentence thanks and your second a question about where or when, and decide what to do with it later. Amy Edmondson's research on psychological safety, the shared belief that it is safe to speak up, finds that leaders who invite and welcome input make it safer for others to speak up too, and she advises leaders to admit they can be wrong.

Say which kind of note it is

"This is to help, not a rating" takes one breath, and it closes the gap Stone and Heen describe between what the giver meant and what the receiver heard. If it is a rating, say that, and say what it rests on.

Feedback examples: the same note, rated and described

The test is whether the other person can check the sentence against their own memory. Each pair below carries the same concern. The first version can only be defended against. The second can be looked into.

• Rated: "You're careless with detail." Described: "Three files went out with last month's version number on the cover, and each one was caught downstream."
• Rated: "You dominated that meeting." Described: "You spoke first on four of the five questions, and I held back my own answer."
• Rated: "You're not open to other people's ideas." Described: "Yesterday, when the deadline came up, you said the date was fixed before Sam had finished, and I stopped raising the problem."
• Rated: "Great job on the pitch." Described: "You cut the pricing slide to one number, and the client stopped asking about it."

The last pair is there on purpose. Praise fails the same test, and a specific thank-you tells someone what to do again.

Feedback questions managers ask

How do you give negative feedback to an employee?

Describe one moment, say what it cost, and stop. Do it in private, soon after it happened, and leave out adjectives about the person. "Three files went out with the old version number" can be checked, where "careless" can only be defended. Then ask what they saw, because they may know something you do not.

How often should managers give feedback?

Often enough that both people still remember the moment. By one measure that is rare: in Gallup's August 2025 reading, 22% of US employees said they had received meaningful feedback in the last week; the chart on the same page, whose series is labelled "% Strongly agree", gives 19% for August 2026 (Gallup, Global Indicator: Leadership and Management).

What do you do when someone gets defensive about feedback?

Stop making the case and ask one question. Someone defending themselves has heard a rating, and more evidence tends to sound like more of the rating. Ask what they saw or what you are missing, then leave it for a day. A calm return tomorrow will usually get further than pressing on today.

How do you give feedback to your manager?

Pick one scene, say what they did and what it did to your work, and leave their style out of it. Notes aimed at power are the costly ones, as senior doctors at Addenbrooke's described when asked about raising concerns. Say what helps as well as what hurts, so that the first note you give them is not a complaint.

How MeaningFuel helps a team reopen feedback

The daily insight: Feedback, Communication and Courage

The daily insight works on what a company chooses, and what it can choose are values and skills: here Feedback and Communication as skills, and Courage as a value. Each working day everyone gets one short insight on one of them. Resistance to feedback itself is a problem, and a problem is what a module is for.

In an ordinary week Lorna might meet a line on hearing something hard, and Bilal one on what makes a comment worth giving. Neither mentions version numbers. They keep the subject alive on days when nobody is being reviewed, so that the box is no longer the only time it comes up.

The module: Welcoming Feedback

Welcoming Feedback is the module for a team whose useful notes have gone quiet. It is a short course of daily lessons: ten short lessons across two working weeks, one each working day, for a team's named people. Whoever runs the account, or leads the team, books it for the team, it invites them, and those who join get one lesson each working day.

It starts with receiving, because its premise is that a team that cannot take a note cannot be given one. In the first week people practise a specific ask, thanking before explaining, and one follow-up question. In one lesson a colleague says a report "buries the point", and the move is to ask where they stopped reading, since a first note is usually a shortened, softened version of what the giver saw. The lead is asked to go first, because notes aimed at the person with power are the costly ones. If the first minute goes badly, a person can go back the next day and say so, which is the lesson Bilal never had after last spring.

Giving comes in the second week and keeps each note to one moment: the situation, what a camera would have recorded, and what it cost you. This is the situation-behaviour-impact method, a widely used feedback method from the Center for Creative Leadership. The aim is a note given within two days of the moment. The module draws on Amy Edmondson's research on psychological safety and on Douglas Stone and Sheila Heen's book.

What Lorna can see, and what she cannot

Insights shows how much of a team takes part and how the insights land, pooled, with nobody named. It cannot show whether a note was given or how it landed, because during the module she sees whether the team is doing it, and never anyone's answers or what they write. The sign she is looking for is the one the module ends on: whether "can I give you a note?" gets a flinch or a "please". She will only find that out in the room.

The module gives a team two weeks of rehearsal at asking for a note and receiving one well. Where the resistance centres on one relationship's history, that pair needs a repair conversation first, and the module says so. It promises no outcome.

Back to the box

The box on the form says what Lorna thinks of Bilal. A sentence about three version numbers says what happened to three files. He can argue with the first for a week and check the second in a minute. If she then asks what she does that makes his week harder, the notes stop running one way.

• Feedback
• Resistance to Feedback
• Team Leadership
• Communication
• Workplace Culture