Running 1:1s With Engineers That Actually Move Things

Last Tuesday you had a 1:1 with your senior engineer. They said "yeah, things are good." You both glanced at the Jira board, talked about the migration ticket, and they left at the 22-minute mark. You marked it done.

That meeting didn't happen. You spent 22 minutes confirming what was already in Linear. Three weeks of those and you'll wake up surprised when that engineer puts in notice or quietly disengages from a project they used to care about.

The 1:1 is the only 30 minutes a week reserved for things that don't fit anywhere else. If you're using it for status, you've already lost.

Why The 1:1 Is Your Only Signal Channel

Status moved away from you years ago. Linear has the ticket state. Standup has the daily blockers. The async update in Slack has what shipped yesterday. By the time someone shows up to your 1:1, there's nothing left to tell you about the work. That's good. It means the meeting is free for the things that need a human channel.

Career frustration. The architecture decision they're nervous to push back on. The peer they're starting to dread pairing with. The half-formed idea about decomposing the billing service that they don't want to write up yet. None of those land in Linear. If you don't make space for them, they don't surface. They just compound.

The rule I run by: the 1:1 is the only place I'm guaranteed to hear what isn't going to show up on a board. Protect that, or you're flying blind on a team of 7.

The 4-Bucket Agenda

After running 1:1s with about 40 engineers across three companies, I landed on four buckets that cover what needs covering, and exclude what belongs elsewhere.

Their work, 10 minutes. Not status. The hard parts. What's the thorniest decision on their plate? What's the thing they keep putting off? Where are they stuck and not yet asking for help? "What's the part of your current project you're avoiding thinking about?" gets you more than "how's the migration going?"

Their growth, 8 minutes. What skill are they actively building this quarter? What scope do they want that they don't have? What kind of work do they want more of in 12 months? This bucket is where you earn the right to keep them on the team for another year. Skip it for a month and they'll start interviewing.

Manager-of-me, 5 minutes. What can I do better? What's frustrating about how I operate? Where am I getting in your way? Most engineers won't volunteer this; you have to ask, and you have to ask the same way for several weeks before they believe you actually want the answer.

The team, 7 minutes. Dynamics, peer friction, processes that hurt. Who's hard to work with? Where is the team's standard practice broken? What would they change about how we plan?

Total: 30 minutes. Bucket order rotates. The engineer picks where to start. If they always start with "their work," you've got a status problem and the next section is for you.

The "What's On Your Mind?" Opener

"How's it going?" is the most expensive question in management. It costs 30 seconds of nothing, you both know what's coming, and it sets the tone as a polite check-the-box. Replace it with something that names what you actually want.

For the engineer who runs short: "What's the thing this week you'd want me to know that I probably don't?" The phrasing matters. It assumes there is a thing, assumes you don't know it, and gives them permission to bring something forward without having to justify why it's worth your time.

For the engineer who runs long: "Of everything on your plate right now, what's the one decision or conversation you'd want to spend the bulk of today on?" This forces a pick, which is a gift to a high-talker. Without it they'll spend 28 minutes on a codebase tour and you'll learn nothing.

For the engineer who's clearly stewing: "You seem off this week. Want to talk about what's going on, or skip it and pick something easier?" Naming what you see is the move. The out is what makes it safe to take. One in three takes the out; the rest exhale and tell you.

You'll find your own variants. The pattern is the same: name a real expectation, give them a real choice, don't ask a question that can be answered with "good."

When To Push, When To Listen

The default ratio is 70/30. They talk 70 percent. You talk 30 percent. Most new EMs run it the other way for the first six months because pushing feels like adding value and listening feels like nothing's happening. Listening is the value. You just don't see it land until later.

But there are three moments to break the 70/30 rule and push.

They're stuck on a decision. They've been circling the same architecture trade-off for two weeks. They don't need more space; they need a forcing function. "If I told you I needed a call by Friday, what would you pick?" Then ask why. The decision is theirs. The deadline is yours.

They're catastrophizing. "This whole project is going to slip and the VP is going to lose confidence in the team." Maybe. Probably not. Push back specifically: "What's the actual evidence the VP has lost confidence? What did they say last time you talked?" Half the time the catastrophe is invented. The other half it's real, and you both need to know that.

They're avoiding a peer conflict. Same friction with the same teammate, three weeks running, no action. "What's stopping you from talking to them directly this week?" Don't offer to mediate yet. Make them name the block.

The rest of the time, when they're processing, exploring, venting, half-thinking out loud, your job is to ask one more question instead of jumping to advice. Most EMs cut a productive wander short by 90 seconds because the silence felt long. It wasn't. They were about to say the thing.

The "Boring 1:1" Diagnostic

Three weeks running, your engineer says "I don't really have anything." You're tempted to call it efficiency. It's not. It's a signal. There are three causes. They look identical from the outside.

Cause one: psych safety dropped. Something happened. Maybe you reacted badly to a piece of feedback two months ago. Maybe a peer got a harsh review and they're nervous you'll repeat it. The diagnostic question: "If you did have something hard to bring up, would this be the place you'd bring it up, or would you go to someone else first?" Watch for the half-second pause. The pause is the answer.

Cause two: perceived usefulness dropped. You've been giving them generic advice. You've been saying "good question, let me think about it" and never coming back. They've stopped expecting the meeting to move anything, so they've stopped bringing things. Diagnostic: "Looking at the last month of our 1:1s, which one actually changed something for you? Which one was a waste?" Then sit with the answer.

Cause three: things are actually calm. Rare, and easy to over-claim. Calm 1:1s come in clusters of one or two, not three in a row. If you're three deep, default to one of the first two and dig.

The fix depends on the cause. For safety, you eat the cost: bring up something hard about your own management, sit in the awkwardness, let them see you take the harder path first. For usefulness, change format: "Let's drop the agenda for a month and try something different. What would actually help you?" For genuine calm, keep going and accept that not every 1:1 has to move something.

Career-Path Conversations On A Three-Tiered Cadence

Career conversations don't happen in a 30-minute slot once a year. They happen in three layers, each on a different rhythm.

Every 3 months, tactical skill check. "What's the one skill you're actively building this quarter? How will you know you've built it?" Specific. Time-boxed. Concrete. If they say "leadership," push for a thinner cut: "Leadership in code review? Leadership in cross-team decisions? Leadership in hiring?" Pick one. Track it. Bring it back at the next quarterly.

Every 6 months, scope check. "What kind of work do you want more of? What kind do you want less of? What would you do if you had a free day a week to work on the team's behalf?" The third question is the unlock. It tells you whether they want more IC depth, more team leverage, or more scope across teams. Their answer reshapes which projects you route to them next.

Every 12 months, direction check. "Where do you want to be in a year? IC track, tech lead track, EM track, staff track? What's the version of your job 12 months from now that you'd be excited to do?" If they don't know, that's an answer too. It means you're going to need to surface options for them. Your job in this conversation is not to decide for them. Your job is to make sure no door they want is quietly closing while they aren't paying attention.

Don't run all three on the same day. Spread them through the quarter. The tactical one fits in the regular 1:1; the 6-month and 12-month deserve their own slot, scheduled, on the calendar, with the topic in the title.

The "Skip Because We're Shipping" Trap

Crunch hits. You cancel this week's 1:1s because everyone's heads-down on the launch. You tell yourself it's the responsible call. It's not. It's the most expensive false economy in management.

The math: a canceled 1:1 saves 30 minutes. The cost: every cancellation takes two to three weeks of restored cadence to rebuild the candor you had before. Skip three in a crunch month and you've spent the next quarter relearning what your team is thinking. The shipping you protected was real. The signal you destroyed was also real, and it's slower to repair.

If the shipping pressure is genuine, don't cancel. Downsize.

The 15-minute walking version. Walk to coffee. No agenda. Two questions: "What's wearing you down this week? Anything I can take off your plate?" Half the value, none of the relational cost.

The async written version. Three prompts in DM: (1) what's hardest right now, (2) one thing I could do this week to help, (3) anything else on your mind. Give them until Friday. Read carefully. Reply substantively. Works once or twice a quarter; don't make it a habit.

The pattern: under shipping pressure, change the format, never skip the touch.

Written Follow-Up Cadence

Verbal commitments evaporate in 48 hours. You leave the 1:1, both feel good, go back to your respective fires, and by next week neither of you remembers that you said you'd talk to the principal about scope, and they said they'd send the doc on the indexing approach. By the time it comes back up, the moment's gone.

Write it down. Same day, ideally within 24 hours.

The format that holds up: a shared doc per report, three to five lines per session, written by you, edited by them.

Date: 2026-04-22

What we covered:
- Stuck decision on the queueing rewrite: picked option B, spike by Friday.
- Frustrated by the cross-team review loop with platform; will draft proposal.
- Manager feedback: I'm jumping in with answers too quickly; will hold back.

Commitments:
- Engineer: queueing spike doc by Fri 4/26.
- Manager: forward platform process pain to Sara, propose 30-min sync.

Next time: revisit promotion timeline conversation.

Three things this gives you. One: commitments stick because they're written and visible to both of you. Two: in 12 months when promotion or performance comes up, you have a primary source. Not memory, not vibes, a record of what they worked on, cared about, where they grew. Three: writing the summary forces you to identify what actually moved. If you can't write three lines, the conversation didn't have content. Adjust next week.

Engineers can edit the doc. Some will. Most won't. That's fine. The doc isn't a contract, it's a shared memory.

Common Pitfalls

Five patterns I keep seeing from new EMs, in order of damage:

1. Talking too much. You're at 50/50 or worse. Cut your talking by half next session. Just try it once.
2. Fixing instead of asking. They name a problem. You jump to a solution. Replace with: "What have you already considered? What's stopping the option you're leaning toward?"
3. Treating the 1:1 as a mini performance review. Performance review is its own meeting on its own cadence. The weekly 1:1 isn't the place for heavy feedback unless it's time-sensitive. Mixing them poisons both.
4. Never bringing your own topics. Engineers want to know what's on your plate too. Bring one item: a decision you're working through, feedback you got, something you're thinking about for the team. Don't let it be 100 percent their show.
5. Ending early because "we're good." If you're consistently ending at 18 minutes, you're confirming a status, not running a 1:1. Build a question bank. Have one prompt ready for any session that runs short.

What Success Actually Looks Like

You'll know the system is working when four things are true.

The engineer brings at least one topic per session, unprompted. Not because you asked them to prep, but because they've started seeing the meeting as a useful place to think.

You can name each report's current growth goal without checking your notes. Not the title on their job ladder. The actual thing they're working on this quarter to get better at.

The follow-up doc shows commitments closing within two weeks, on both sides. If commitments stack up un-closed for a month, the doc has become decoration. Reset it.

You've canceled zero 1:1s in the last four weeks. Rescheduled, fine. Downsized to a walking version, fine. Outright canceled, no.

That's the bar. Hit three out of four and you're running a real signal channel. Hit two or fewer and you're back in status theater. Re-read this and pick the one bucket to fix this month.

Learn More

• Staff Software Engineer Job Description
• Engineering Career Ladders That Hold Up
• Performance Reviews for Engineers Without the Theater
• Coaching Senior Engineers Who Outrank You Technically