Saying the useful thing early enough to matter, and hearing it without going to war.
Feedback is information about the gap between what someone intended and what actually landed. It's useful only if it arrives while the person can still do something with it, describes something observable rather than a verdict on their character, and is specific enough to act on. Most workplace feedback fails at least two of those. The receiving half is a separate skill and the harder one: the instinct when criticized is to evaluate whether the criticism is fair, which is a different and much less useful question than what the other person actually saw. You can decide a piece of feedback is wrong. Do it after you've understood it, not instead of understanding it.
Constructs each piece of feedback as praise, then the actual point, then more praise, and applies so much bread that the filling becomes undetectable. The recipient leaves genuinely believing the conversation went well, because as far as anything they can recall, it did. The Artisan considers the matter raised and is puzzled six weeks later to find nothing at all has changed, then constructs a second, larger sandwich.
Notices things all year and says none of them, maintaining instead a private ledger opened once, in the review meeting, and read out in chronological order. Produces observations about a meeting in February with the confidence of someone who took notes, which they did. Has delivered a year of feedback at the precise moment nothing can be done about any of it, then records the lack of improvement as a finding rather than as the predictable result of the filing system.
Deals exclusively in altitude. "Be more strategic." "I'd like to see you show up bigger." "Raise your game." Every sentence is unfalsifiable and none contains a verb anyone could perform on a Tuesday. Asked what specifically to do differently, gestures at the general shape of the thing and repeats the phrase slightly slower, as though the problem were audibility. Would be startled to learn the recipient has spent a fortnight trying to work out which meeting this is about.
Receives each observation as an allegation and responds with context, which is accurate, and mitigating circumstances, which are real, and a small correction to the timeline, which is fair. Wins every exchange on the facts. Has established beyond doubt that the feedback was imprecise, and in doing so ensured that the next person who notices something will keep it to themselves. Reports, accurately, that they rarely receive criticism.
A teammate's design review comments read as harsh to everyone but them. You've noticed it three times. What would you do — wait for the review cycle and raise it as a pattern, or say it this week, tied to the specific comment thread while everyone can still remember it?
You get feedback you think is based on a misunderstanding of what happened. What would you do — correct the record first, or first find out what they observed that led them there, and correct the record after you understand which part is actually wrong?
Someone asks you for feedback and you genuinely have a criticism, but they've had a rough quarter. What would you do — hold it until things settle, or give it now, plainly and small, on the grounds that a rough quarter is exactly when unspoken problems compound?