Feedback training: why giving feedback is a skill, not a good intention

In almost every employee survey, the same outcome emerges. We don’t address issues with each other enough. We let things slide. People talk about each other rather than with each other. The response is predictable: a feedback training is organized. A half-day session, sometimes two. A model on the flipchart, a few exercises in pairs, a positive evaluation.

And six months later, the exact same outcome appears in the next employee survey.

That is not a failure of the trainer. It is a design problem. Giving feedback is treated as knowledge being transferred, whereas it is behavior that must be executed under tension. Those are two different things, and they require two different approaches.

What a feedback training usually does and does not deliver

Let’s be honest about the yield. A good feedback training delivers three things, all of which are valuable:

• A shared language. When everyone knows the same structure, a conversation no longer has to start from scratch.
• Awareness. People see how their own style comes across and recognize their avoidance patterns.
• Legitimization. The conversation about holding each other accountable is officially open, making it easier to hold.

What a training generally does not yield is the skill itself. Giving feedback is one of the most difficult communication skills there is, precisely because execution takes place under tension. The skill does not lie in knowing that you should mention concrete behavior. It lies in being able to do so on Wednesday afternoon, when the colleague in question has just had a tough week, you yourself are under time pressure, and the conversation takes a direction you hadn’t anticipated.

Anyone wanting to bridge that gap must accept that between understanding and capability lie a substantial number of repetitions.

Feedback is not a neutral intervention

Here lies the most ignored insight from research into feedback, and it should be the starting point of every feedback training.

In Psychological Bulletin (1996), Kluger and DeNisi published a meta-analysis covering 607 effect sizes and over 23,000 observations. On average, feedback improved performance with an effect size of 0.41. However, in more than a third of cases, performance declined following feedback. Not by chance, not due to measurement errors, and not because the feedback was negative.

Their explanation: feedback works by directing attention somewhere. If feedback directs attention toward the task, performance usually improves. If feedback directs attention toward the person, on who someone is and what this says about them, cognitive capacity shifts to self-protection instead of the work. The closer feedback gets to the self-image, the smaller the chance that behavior improves.

That explains why well-intentioned feedback so often backfires. “You are quite dominant in meetings” is a statement about someone’s personality. “You interrupted Anne three times before she finished her point” is a statement about behavior. The first variation triggers defensiveness. The second enables change.

In a training course, this distinction is usually explained. And subsequently, someone has to apply it in a situation where their heart rate rises. That is where it stalls.

Three design errors in the average feedback training

Error one: the model takes center stage instead of the situation. Feedback models are useful as scaffolding, but most difficult conversations do not fail at step one. They fail in the response of the other person, and that is where the model stops.

Error two: practice is done with easy cases. In the exercise, a cooperative colleague plays a slightly resistant employee. In practice, you face someone who bursts into tears, or who says: “Funny that you say this, because you do it yourself too.” That scenario is rarely practiced, and that is precisely the scenario on which people get stuck.

Error three: the training is an event rather than a routine. What you do once, you can do once. In Human Resource Management (2006), Saks and Belcourt showed that 62% of participants apply content immediately after a training, 44% after six months, and 34% after a year. Without repetition, two-thirds evaporates.

The underrated part: receiving feedback

Virtually every program spends eighty percent of its time on giving and twenty percent on receiving. In practice, that ratio is reversed.

A feedback culture does not emerge because people formulate better. It emerges because the receiver doesn’t explode, shut down, or take subtle revenge in the next meeting. As long as people experience that giving feedback carries a social cost, they will continue to avoid it. Completely rational.

What the receiver needs to learn is small and concrete: say nothing for a moment, ask further about what was observed, and only then respond. Three actions. All three difficult when you are affected. All three trainable.

Psychological safety is a prerequisite, not an outcome

Organizations often deploy feedback training to create psychological safety. That is the wrong order.

In Administrative Science Quarterly (1999), Amy Edmondson showed that psychological safety—defined as the shared belief that the team is safe for interpersonal risk-taking—is linked to learning behavior in teams, and that learning behavior forms the link to team performance. Asking questions, asking for help, reporting errors, and seeking feedback are all actions through which you make yourself vulnerable.

In other words: safety precedes the behavior you want to see. In a team where it is not safe, such a process leads to polite, superficial conversations. People go through the motions without taking any risk.

This does not mean you should wait until the culture is perfect. It does mean you need to build up practice somewhere other than in the team itself. Someone who has practiced three times with a conversation they find nerve-wracking enters that conversation differently. Not because they know the model better, but because the tension has subsided.

Practical example: a manufacturing company with self-organizing teams

A manufacturing company shifts to self-organizing teams. The team leader layer has largely disappeared. Where a manager previously addressed someone about arriving late or sloppy handovers, colleagues now have to do so among themselves.

A feedback training was conducted for all teams. The evaluations were good. Yet little changes, and the operational manager hears in the hallways exactly why: “I’m not going to address Kees about his work; I have to stand in the same shift with him again tomorrow.”

That is not a knowledge deficit. That is a risk assessment. And therefore, it is an issue of behavior change, not of knowledge transfer.

What helps in this situation is breaking down the problem. The team identifies three concrete situations that everyone recognizes: someone who doesn’t clean up their workplace, someone who structurally skips the safety check, someone who is distracted by their phone during the morning kickoff. For each situation, the desired behavior of the person taking action is defined. For example: state what you observed, ask how the other person views it, and make one agreement.

Then everyone practices those three conversations, according to the principle we described in AI role-plays: why practicing with an AI avatar does ensure retention. Not once in a classroom, but four to six times spread over two months, in AI role-plays where the conversation partner reacts differently every time. One time the colleague startles and apologizes. Another time he gets defensive. The third time he turns it around.

The value isn’t in perfect conversations. It is in ensuring that the first real conversation is no longer the first conversation.

Feedback training that actually sticks

Four interventions make the difference between a training evaluated with an eight out of ten and a training that is visible in behavior.

Work from situations, not from models. Collect in advance the ten conversations that people actually avoid. That list is the learning material, and it is immediately the sharpest agenda for leadership development you can have.

Make the difficult response the learning material. Don’t practice the opening; practice the response after. Tears, counterattacks, denial, silence. That is where the skill lies.

Space out the practice. Two short sessions per week for six weeks works better than a single day. The forgetting curve is not an opinion, and the only antidote is repetition.

Practice receiving just as often as giving. Explicitly put people in the receiver’s seat, with feedback that isn’t entirely justified. That is the realistic version.

Within PractAIce, those four principles are supported by scenarios that users build themselves from their own practice, by personas that react differently, and by an AI coach that provides feedback on demonstrated behavior afterward. If the conversation went wrong at one moment, you can go back to that exact moment and redo it as a short microtraining, rather than having to restart the entire conversation.

What you measure

Feedback training is rarely measured on anything meaningful. Satisfaction ratings say nothing about behavior. Three indicators are useful:

1. Frequency. How many corrective conversations have people held over the past month? Avoidance is the core of the problem, so counting helps.
2. Nature of the formulation. Are conversations focused on behavior or on the person? That can be read from practice conversations.
3. Aftermath. Does something escalate in the weeks following a conversation, or does it remain resolved?

Giving feedback is moreover not a standalone skill. It is linked to the broader set of soft skills present in every difficult conversation: listening, probing deeper, setting boundaries, and managing emotions. Read more about how to build an accountability culture without resorting to campaign jargon in Accountability culture: why giving feedback is behavior, not knowledge.

Frequently asked questions about feedback training

What is feedback training?

Feedback training is a learning process in which people learn to share observations about behavior in a way that allows the recipient to act on it, and to receive feedback without getting defensive. Effective variations are not built around a model, but around repeated practice with realistic situations.

Why does feedback training often fail?

Because it concerns behavior that must be executed under tension, whereas training mostly transfers knowledge. Without repetition in realistic situations, the effect fades within months.

What is the difference between giving feedback and addressing someone?

Addressing someone is a specific form of feedback in which you mark a boundary. There is an underlying norm and usually a request for change. That makes it more nerve-wracking and requires different conversational skills than complimenting or advising.

How often should you practice giving feedback?

Two short sessions per week for six to eight weeks, focused on one type of conversation, produces a noticeable difference in practice. A single training day without follow-up virtually never does.

Can you practice giving feedback without colleagues?

Yes. With AI role-plays, you practice a feedback conversation with a conversation partner that reacts realistically, without requiring a colleague to act. That significantly lowers the threshold, especially for conversations that people find intimidating.

Would you like to make the most difficult feedback conversations in your organization actionable instead of just debatable? On the page about feedback training, we show how teams build those conversations from their own practice, how the AI coach feeds back the demonstrated behavior, and how short repetitions prevent it from fading away again after six weeks. Feel free to bring your hardest case study. It is usually the best test.