Giving Feedback in 1:1s Without Making It Weird

Why feedback in 1:1s gets weird

Weirdness comes from scarcity. If the only feedback your report ever hears arrives in a scheduled meeting, preceded by a shift in your tone and the words 'so, I wanted to talk about something,' then every 1:1 becomes a threat scan. They spend the first ten minutes trying to figure out whether this is the week something's wrong.

The fix isn't better phrasing — it's volume. Managers who give small, specific feedback constantly (two sentences after a meeting, a quick note on a doc) can raise harder things in 1:1s without ceremony, because on their teams feedback is just a normal thing that happens. If you're currently at near-zero, don't open with the big stored-up item. Start with two weeks of small, real, mostly positive observations, then work up.

One reframe before the tactics: the 1:1 is not a feedback delivery vehicle. It's your report's meeting, and feedback is one agenda item among several — not the hidden purpose of the whole thing.

SBI in plain words

The Center for Creative Leadership's SBI model — Situation, Behavior, Impact — is the rare framework worth memorizing, mostly because it stops you from doing the two things that make feedback go sideways: generalizing and mind-reading.

• Situation — pin it to a time and place. 'In Tuesday's sprint review,' not 'lately' or 'sometimes.'
• Behavior — describe what a camera would have recorded. 'You interrupted Maya twice while she was presenting the numbers,' not 'you were dismissive' — that's a judgment, not a behavior.
• Impact — say what happened because of it. 'She stopped offering ideas for the rest of the meeting, and I need her ideas.'

Full example, maybe fifteen seconds of talking: 'In Tuesday's sprint review, you interrupted Maya twice while she was walking through the numbers. She went quiet for the rest of the meeting, and we lost her read on the launch. What was going on for you?' That closing question matters — SBI ends with their perspective, not your verdict.

SBI works identically for praise, and that's not a gimmick. 'Great job' is empty calories. 'In Thursday's client call, when they pushed back on pricing, you paused and asked what budget they were working with — that flipped the whole call' tells someone exactly what to do again.

The honest answer on praise-to-correction ratios

You've probably seen a magic ratio quoted — five compliments per criticism, or similar. The research behind precise numbers is shakier than the blog posts suggest, so here's the honest version: specific praise should comfortably outnumber correction, and the exact multiple matters less than this — your praise must be as specific as your criticism.

Most managers get it backwards. Their corrections are detailed and their praise is vague, so praise starts to sound like throat-clearing before bad news. If the only time you get specific is when something went wrong, your reports learn that specificity means trouble.

And retire the compliment sandwich. Wrapping a correction in two slices of praise doesn't soften the correction — it poisons the praise. People learn to hear 'you did a great job on X' as an incoming-missile warning. Say the good thing when you mean it. Say the hard thing plainly when you need to. Different moments, different sentences.

In the moment or saved for the 1:1?

Not everything belongs in the 1:1. A rough sorting rule:

• In the moment, or same day: praise, small factual corrections, anything a reasonable person fixes with one sentence. 'Heads up — the numbers on slide four didn't match the appendix' should never wait six days for a meeting.
• Saved for the 1:1: patterns ('this is the third missed deadline this month'), anything touching skills or growth, anything you're still angry about, and anything that needs a conversation rather than a comment.

The failure mode to avoid is the grievance bag: saving up six weeks of small annoyances and unloading them in one meeting. By then each item is stale, your report can't remember the specifics, and the pile feels like an ambush. If something's worth saying, it's worth saying within a week.

And if what you're sitting on is bigger than feedback — performance genuinely off track, trust broken, something that must change or else — that's a different kind of conversation with different rules. Don't try to fit it through the feedback-sized slot.

Scripts you can steal

Word-for-word starters. Adjust them to how you actually talk — a script delivered in someone else's voice sounds like HR.

• Small correction, same day: 'Quick note on standup this morning — when you said the API work was done, QA hadn't started. Next time say done-pending-QA so nobody plans around it. Not a big deal, just flagging.' Twenty seconds, then move on. Don't hover.
• Pattern, in the 1:1: 'I want to bring something up because I've now seen it three times, and once or twice would just be noise. The last three estimates each ran more than double. What's your read on what's happening?'
• Specific praise: 'That escalation doc you wrote Tuesday — the timeline at the top saved me fifteen minutes of reconstructing events, and I forwarded it to my boss unedited. That's the standard now.'
• When you're not sure you're right: 'I might be misreading this, so check me. From the outside it looked like you'd disengaged from the planning work. What does it look like from the inside?'

Notice every correction ends with a question. You're delivering an observation, not a sentence. A third of the time, their answer will genuinely change your view of what happened — which is exactly why the question has to be real and not rhetorical.

The harder half: getting feedback from your reports

Giving feedback is the easy direction — you control the timing, the framing, the script. Receiving it from someone whose raise you influence is where the power gradient does its damage. Ask 'any feedback for me?' and you'll get 'no, all good' from now until one of you changes jobs.

Three things actually work:

• Ask questions small enough to answer safely. 'What's one thing I did this month that made your job harder?' beats 'how am I doing as a manager?' — it presumes a nonzero answer and caps the risk at one item. There are more in this spirit among these 1:1 questions.
• Go first, and mean it. 'I think I hijacked your design review Thursday — I jumped in with my solution before you'd finished presenting yours. I'm working on it.' Naming your own specific miss proves criticism is speakable here.
• Make the first hard truth pay off visibly. The first time a report risks real criticism, your reaction sets the price of the second one. Say thanks, ask one clarifying question, then do something about it within days — and tell them what you did.

Follow up, or it didn't happen

Feedback isn't a delivery; it's a loop. The correction that changes behavior is almost never the conversation itself — it's the moment two weeks later when you say, 'I noticed the last two estimates had buffer built in. That's exactly what we talked about. Thank you.' People repeat what gets noticed.

So write it down. A one-line note — what you said, when, what you agreed would change — turns feedback from a vibe into a thread you can pick back up. Same rule as action items from your 1:1s: if it isn't captured somewhere that resurfaces, it evaporates by Friday.

Close the loop on upward feedback with double care. When a report tells you something hard and you fix it, report back: 'You mentioned my late-night messages created pressure to respond — I've scheduled-send everything since. Was that the right read?' That one sentence buys more candor than a year of saying your door is always open.