Most feedback is delivered accurately, received politely, and changes nothing. The failure is not usually in the words.
It is that two different instruments are being operated at once, and the person on the receiving end cannot tell which one is in play.
Feedback is not evaluation
Evaluation answers a question about the person: how are they doing, what is their standing, what happens to their compensation. It is periodic, it is recorded, and it has consequences attached.
Feedback answers a question about a piece of work: what happened, what effect did it have, what should be different next time. It is continuous, it is not recorded, and nothing follows from it except the adjustment.
When these arrive together, or when the recipient suspects they might have, the evaluative question captures all available attention. Nobody processes a suggestion about how to run a meeting while working out whether their job is in question. They will nod, agree, and leave having learned only that their standing is uncertain.
Which means the most useful sentence in most feedback conversations is the one that separates them. This is not about your position here, it is about the meeting on Tuesday. Said at the start, it frees the person to actually consider the content.
Behaviour, not attribute
An attribute cannot be acted on. A behaviour can.
You need to be more strategic gives somebody nothing to do on Monday. It also functions as a statement about who they are, which invites either defence or a quiet conclusion that they are unsuited to the work.
In the last three proposals you have gone straight to implementation without setting out why the client should care. Start with the problem you are solving for them is the same observation converted into something a person can do differently on Monday morning.
The test is whether the person could describe, specifically, what they would do next time. If they could not, no feedback has been given regardless of how clear it seemed.
Timing beats phrasing
Feedback given within a day or two of the event needs almost no craft. Feedback given six weeks later needs a great deal and still works less well.
Proximity supplies context both people share, which removes the need to reconstruct what happened. It keeps the matter small, because a single recent instance carries no accumulated meaning. And it arrives while the person can still connect the correction to their own memory of doing the thing.
Delay converts an observation into a pattern. Three instances mentioned together is a case, and a case invites a defence, however carefully it is worded.
The annual review is the wrong instrument
Organizations frequently locate all their feedback inside the evaluation cycle, which guarantees both instruments fail.
The feedback is stale by up to eleven months and inseparable from a conversation about compensation, so it is received evaluatively. The evaluation is contaminated by developmental material that should have been dealt with when it happened.
Nothing raised in a review should be new. If a person learns about a problem for the first time in an annual document, the failure belongs to the leader, who had a year of opportunities and used none of them. The review should be a summary of conversations that already happened.
Positive feedback needs the same discipline
Recognition that names no behaviour teaches nothing, and it is the more common failure because it feels sufficient.
Good work on the Henderson job is pleasant and transmits nothing. You called the client before they noticed the delay, which is why that conversation went the way it did, tells somebody exactly what to repeat.
Specificity is also what makes recognition credible. A person who receives a precise account knows their leader was paying attention, and that knowledge does more than the praise itself.
Checking that it landed
Agreement in the room is not evidence of anything. People agree in order to end feedback conversations, and they do it sincerely.
Two things are better. Ask what they will do differently, and listen for whether the answer is specific. A vague answer means it has not landed, and the moment to discover that is now rather than in six weeks.
Then look at the next instance. Behaviour that has not changed by the third opportunity was not understood or was not accepted, and either way the follow-up conversation is a different one: we talked about this, it has not changed, help me understand what is in the way.
The frequency problem
A leader who gives feedback rarely makes each instance an event. A leader who gives it constantly produces noise and a person managing your reactions rather than doing the work.
The usable threshold is whether the thing will recur and whether it matters when it does. One-off circumstances need no correction. Anything that will happen again next month, in a way that costs something, is worth two minutes now.
Applied consistently, that threshold produces a working rhythm in which small corrections are ordinary, nothing accumulates, and the rare serious conversation is unmistakably serious because it is not competing with a constant stream of minor ones.
Edited by Patrick J. Wolf, PhD