Leading a team through layoffs

The note I wrote for myself before the team meeting in November was eleven lines long, and I still have it. The first line is “say the number first”. The last line is “don’t say ‘family’”. In between are things like “who to ask about severance” and “no, there isn’t a second round planned”, the second of which I wasn’t sure was true when I wrote it and said anyway.

I’ve been rereading it because of the last ten days. Microsoft announced 10,000 cuts on the 18th, and on Friday Sundar Pichai sent Google’s note, 12,000 people, about 6% of the company, with the line “I take full responsibility for the decisions that led us here.” Amazon had already said 18,000 at the start of the month. The parts of the Google story that went around fastest weren’t the number. They were people finding out from an email at an odd hour, or by their access not working, including people who’d been there a very long time. I don’t know how much of that is accurate and how much is what spreads on Twitter, but I recognized the feeling from the receiving end of other people’s stories, and I recognized the mechanics from the other end of ours.

Ours was small next to those. We’d hired hard through 2021 and the first half of last year, against a plan that assumed the market would keep doing what it had been doing, and by the fall it wasn’t, and the board meeting in October went the way those meetings go when the plan was wrong. I’m not going to go into the numbers. Some of the people who left were on my team, and I told them myself. For the rest I was in a side channel, running the access list.

What I said that day, to the people who stayed, was roughly this. Here’s how many people left and which teams. Here’s why, in two sentences, without the euphemisms. The people who left are getting this much severance and this much health coverage, and here’s who to talk to if you’re worried about someone. Nobody else in this room is affected. I’m going to be around all afternoon, and I’m going to answer any question I’m allowed to answer.

I’m reasonably happy with that list. Saying the number first matters more than I’d have guessed. Everyone in the meeting has already done the math on the Slack channel member count before you open your mouth, and the five minutes you spend on context before the number is five minutes they spend not listening. Saying “nobody else in this room” early is the other one. People can’t hear anything about the company’s runway until they know whether they still have a job.

What I’d change is mostly what I said later in the week, and what I didn’t say at all.

I said the decision was right. I believed it, mostly, and I still do, but I said it on the Wednesday to a team that was still in shock, and it landed as “stop being sad”. A manager on another team said something similar in his one-on-ones, and between us we managed to make it sound like the company had moved on before the people affected had even returned their laptops. What I should have said is that it was a decision I’d have argued against if I’d thought it was wrong, and that I hadn’t, and then stopped. Defending it wasn’t my job that week. The people in front of me didn’t need convincing, they needed to know what happens next.

I didn’t say anything about the work. Everyone who stayed was carrying a piece of something that someone who left had owned, and I left it to the next sprint planning to sort out, which meant a week of people quietly wondering whether they were now on call for a service they’d never touched. A list on the Monday, with every system and project that had lost its owner and a name next to each one, even a temporary one, would have saved a lot of that. We did make that list eventually. It had nineteen lines and three of them said “me”.

And I said there wasn’t a second round planned. That one was true when I said it, and it’s still true today, but I’ve watched friends at other companies hear exactly that sentence in October and then get the calendar invite in January. I can’t promise it, and saying it like a promise borrows trust I might not be able to repay. Next time I’d say what I actually know: what the plan is, how much runway it gives us, and that if the plan changes they’ll hear it from me before they hear it from anyone else.

The other thing I got wrong was more personal. I’d read everything I could find on how to do this well, and I’ve been managing people for a long time, and I’d assumed that between the two I’d be good at it. It made me good at predicting what people would be thinking. It didn’t make me any better at the conversations themselves, and in the first one I talked too much, filling silence because the silence was uncomfortable for me. The person across from me had heard what she needed in the first thirty seconds. Everything after that was me managing my own discomfort at her expense.

Google’s process has gotten a lot of criticism this week, some of it probably deserved, and I’m wary of feeling superior about it. At 12,000 people you can’t have a conversation with each one, and at our size you can, so we did, and it doesn’t make us better people. It does mean that companies our size don’t get to use the email excuse.

The people I find hardest to talk to now are the ones hired in the last year, on the strength of a pitch that I gave them. Several of them took a pay cut from a bigger company to join us, partly because I told them a startup with real revenue was a safer bet than it looked. None of them were affected in November. A couple of them have asked me, privately and politely, whether I still believe that pitch, and I’ve given them honest answers that I’m not sure were the right ones.