Most feedback fails in the first sentence. Not because the content is wrong, but because the opening carries a verdict, and the other person spends the rest of the conversation defending against the verdict instead of hearing the content.
“You have been really careless lately.” The person on the receiving end is no longer thinking about the two reports that went out with errors. They are assembling a defense of their character. You have converted a fixable problem into an argument about who they are.
There is a better structure, and it is learnable. I did not invent it. Versions of it show up in nonviolent communication, in situation-behavior-impact models, and in the practice of nearly every experienced executive coach. What follows is the version we teach, why each part exists, and how to practice it so it survives contact with a real, defensive human.
The formula
Five parts, in order:
“When [situation], I noticed [observation], and it made me feel [impact]. Because [need] is important to me, could we try [request]?”
A filled-in example. Your peer rewrote your section of a shared document the night before a review, without telling you, and you found out in the meeting:
“In yesterday’s review, I noticed my section had been rewritten without a heads-up, and I felt blindsided walking into that room. Because coordinating on shared work matters a lot to me, could we agree to flag major edits before the meeting next time?”
Compare that to what most of us would say under the same frustration: “You can’t just rewrite my work behind my back.” One of these starts a conversation. The other starts a trial.
Why each part exists
The situation anchors the feedback to one specific, recent moment. Not a pattern, not “lately,” not “always.” One moment. Patterns feel like prosecution. A single moment feels like a solvable event, and if there really is a pattern, it will surface naturally once the first instance is discussed without bloodshed.
The observation is the load-bearing part, and it is where most feedback breaks. The test is simple: would a camera have recorded it? “The last two status updates went out without the metrics section” passes the test. “You do not take this project seriously” does not. A camera cannot record seriousness. Observations are nearly impossible to argue with, which is exactly what you want, because every argument the other person cannot have is energy that goes toward actually hearing you.
The impact, owned with “I” language, does something subtle. “I felt sidelined” cannot be disputed, because you are the only authority on how you felt. “You sidelined me” is an accusation with a built-in counterargument. Same event, same feeling, completely different conversation.
The need explains why this matters, and it is the part people most often skip. Without it, the feedback reads as a complaint. With it, the feedback reads as information about how to work with you. “Because reliability matters to me” or “because I need us to coordinate on shared work” tells the other person what you are actually protecting, which is almost never their failure and almost always something you both value.
The request must be small, concrete, and about the next occurrence. Not “be more considerate,” which is a personality renovation, but “flag major edits before the meeting,” which is one behavior, one situation, achievable this week. People change behaviors. Nobody changes their personality because a colleague asked them to in a one-on-one.
Timing: the part that happens before the words
The formula fails if deployed at the wrong moment. The rules here are unglamorous but they do a lot of work.
Do not give feedback right before someone’s important meeting. Not when either of you is flooded with emotion. Not in public, ever. Public feedback triggers defensiveness regardless of how well it is phrased, because the audience raises the stakes from “improve this behavior” to “defend your reputation.”
Do choose a moment with privacy, calm, and enough time that the conversation does not get cut off mid-repair. And open by asking: “I’d like to share something. Is this a good time?” That one question hands the other person a small amount of control, and people who feel some control get defensive far less.
There is a failure mode on the other side, though, and it is more common. Waiting for the perfect moment becomes not doing it at all. The signal that you have crossed from patience into avoidance is rehearsal: if you catch yourself running the conversation in your head on your commute, the moment is now. Avoided feedback does not age well. The small thing grows into resentment, and the eventual conversation has to carry six months of freight instead of one incident.
The candor problem
A common misreading of this formula is that it is a politeness technique, a way to make feedback soft. It is the opposite. The structure exists so you can afford to be fully direct.
Feedback fails in two directions. All candor and no care lands as an attack, and the message is lost in the defense. But all care and no candor is the quieter failure: you soften the message so much that the other person does not register it as feedback at all. A month later nothing has changed, and you are now frustrated at them for ignoring something they never actually heard. I have seen this one destroy more working relationships than harshness, because it looks like kindness while it operates as avoidance.
The fix is to make your intent explicit and then say the direct thing. “I want to share this because I think it is holding you back, and you have too much potential for that.” Then the observation, plainly. People do not resist feedback nearly as much as they resist feeling judged. When the intent is visibly on their side, the same direct words land very differently.
After the conversation
Two things separate feedback that changes behavior from feedback that just clears your conscience.
First, end with a next step both people say out loud, and recap it. “So next time, you’ll flag major edits before the review, and I’ll make sure you have my draft by Tuesday.” The recap is where misunderstandings go to die.
Second, follow up. Check in within a week or two. If they have improved, say so, specifically. Recognizing the change is what closes the loop, and it is what makes the next difficult conversation with this person easier instead of harder. A feedback conversation with no follow-up reads, in hindsight, as a drive-by complaint. A conversation with follow-up reads as investment.
Practicing it out loud
Here is the thing about this formula: it is easy to understand and surprisingly hard to execute, because the moment the other person pushes back, your prepared structure evaporates and your habits take over. The person who nodded along to “observations, not interpretations” still opens with “you always” when their pulse is up.
The only fix I know is rehearsal against resistance. Say the feedback out loud to something that pushes back. We built a coach for exactly this on Tough Tongue AI: it teaches the four blocks, timing, the formula, candor, and follow-up, then makes you deliver a complete piece of feedback to a character who gets a little defensive, the way real colleagues do. The first attempt is usually where people discover the verdict hiding in their phrasing. By the third attempt, the words sound like theirs.
However you practice, practice before it counts. The formula is not a script to read in the room. It is a structure to internalize so that when the room gets tense, the structure is what comes out.
