Engineering Manager Coach & Mentor. For the conversation you keep rewriting.

You were the person who could take the ugliest ticket in the sprint and have it in review by Thursday. That's why you got the team. Nobody mentioned that the skill is now maybe 5% of the job, and that the other 95% has no compile step and no tests.

So a Thursday looks like this. Forty minutes with an engineer who isn't shipping and you still can't name why. Four other people who each need twenty minutes you don't have. A hiring plan you owe your CTO. At 6pm you open the repo out of something close to homesickness.

The trap in year one is that you keep score on the old scoreboard. Tickets closed, PRs reviewed, code written. On that scoreboard you're now a mediocre engineer and you feel it daily, which is a large part of why the 12-to-24-month mark is where I watch first-time EMs break.

So we build the new scoreboard before the old one eats you: your two hardest reports by name, the relationship with your own manager, and one delivery number you're willing to defend in a skip-level. Weekly session, small written homework in between, and we work the calls that landed on your desk that week instead of general theory. The theory is free and you've already read it. By week 12 the answer to "how's it going" is one you'd say out loud with your director in the room.

He was your strongest engineer in 2023 and you've got the review to prove it. Then a reorg, a new PM, a stack he didn't pick, and somewhere in there he stopped shipping. Nobody says it out loud. The team just routes around him now, and you can see it in who picks up the tickets he used to own.

You've had the conversation four times. Each time it went well. That's the problem.

In practice this isn't one problem, it's three, and they want opposite treatments. Can't: he never learned the new stack and he's too senior to admit it. Won't: he checked out the day the reorg took his project away. Or can't-here: he'd be good on a different team as opposed to being bad at his job, and the honest move is a transfer. That third one gets diagnosed least, because naming it feels like giving up.

So we name it, and then we write the sentences you'll say. Not a framework, the sentences. Then we run it dry with me playing him, including the part where he pushes back and you soften, because that's the rep that matters. If it ends in a PIP we build one he can survive. If it ends in an exit we design it so the rest of the team reads it as fair, because they're watching and they'll draw a conclusion either way.

Four weeks: the conversation held, the plan in writing, and an escalation line you agreed with yourself in advance so that Monday morning can't talk you out of it.

Completion swings between 40 and 90% and you can't explain the swing. Your PM thinks your estimates are optimistic. In the last skip your CTO said he needs to understand why Q3 shipped 55%, and you had five answers, none of which survives being said out loud in an exec room.

Padding by 30% didn't do the trick. Cutting scope mid-sprint mostly bought you a PM who trusts you less.

The first thing I look at is boring: not the sprint that missed, but the four days before anyone admitted it would. On most teams the slip is visible around day 4 and confessed around day 10. That gap is the whole problem, and it isn't an estimation problem, it's a reporting-upward problem. It usually lives in one recurring meeting where nobody wants to be first to say the number is off.

So we go through your own Jira data weekly. What predicted this sprint, what your director sees, and why those are never the same picture. We install one mid-sprint check that surfaces slip while there's still room to trade scope, which is a conversation cadence and not a Jira config, and costs about ten minutes a week once it runs. By week 12 you're living in the 80s, and you have a sentence about Q3 that holds when the CFO is in the room.

Every grooming is a negotiation. The Product Trio exists on a slide somebody screenshotted, as opposed to anything that happens on your team on a Tuesday. Delivery slips, your PM tells the CPO that engineering is slow, your team tells you the PM is delusional, and both of them are roughly half right with you standing in between.

The reflex is to fix it with process. A longer grooming. A shared doc. A weekly trio sync that becomes a status meeting inside three weeks.

It's rarely a process problem. Nobody wrote down who decides what, so every decision gets relitigated by whoever is most annoyed that day. Two capable people can work side by side for a year without ever agreeing where discovery ends and commitment begins, and then one bad quarter turns that gap into a blame narrative.

We rebuild the working agreement in writing, on one page: discovery cadence, delivery cadence, who calls the meeting when a decision is stuck, and what happens when the two of you disagree. You pick three things you'll change and three you'll ask of your PM, because an agreement where only one side moves is a complaint with formatting.

Then we work the reset conversation itself, sentence by sentence, which is the part everyone skips. By week 6 the agreement exists, both sides have changed one visible behaviour, and the story reaching the CPO has a counterweight.

The laptop closes at 9pm and your head doesn't. Four direct reports, a partner, a small kid, a mortgage. On Monday you'll be the person everyone looks at and you'll smile and say let's dig in. Your CTO would take it seriously if you told him. You haven't told him.

Something unfashionable first: not all of this is fixable, and an industry pretending otherwise is part of why you feel like you're failing at it. Some of the load is simply real. Four reports and a small kid is a lot, and no calendar trick makes it not a lot.

What is fixable is the part that arrived by accident. The meeting you kept because you were in it once. The decision you take back every time you delegate it. The 11pm Slack thread that exists because you answered the first one at 11pm.

So in session one we split those two piles honestly. Then one recovery ritual, one thing you drop and take the L on in public, and one conversation you've been avoiding. Every week moves at least one of the three, and the session itself doubles as the one hour a week where nothing you say has to be managed for somebody else's reaction.

By week 8 there's usually one visible cut in the calendar and one delegated thing that stayed delegated. I'm not going to promise you your Sundays back in month one. Most people get them back somewhere around month three, and some rearrange the job instead.

Some weeks you miss the code. Some weeks the human puzzle is the most interesting problem you've ever been handed. Your director keeps adding scope, which reads as a compliment and a trap at once. And the people close enough to have an opinion either want something from your answer or like you too much to be blunt about it.

This is the one question I won't answer for you, and I say that in session one so you're not sitting there waiting for a verdict that isn't coming.

What I can do is stop you deciding it on a bad Tuesday. We separate three things people mash together: missing the craft, missing being competent, and disliking this particular team at this particular company. They point in completely different directions, and the third is the most common by some distance.

Then we make it concrete rather than philosophical. Both paths written out with a year-2 and a year-5 version. A two-week test drive of each, because a fortnight of doing the thing teaches more than six months of thinking about it. And a way to tell your director what you've landed on that doesn't burn the option you didn't take. Some people finish this and go back to IC. A few work out that they wanted a different company as opposed to a different track. There isn't a right answer here, only a decision you made on purpose.