You are an engineering manager now. Here is what the job turned out to be.
A new engineering manager is somebody who has stopped being measured on what they produce and started being measured on what their team produces, usually without anyone explaining that the change has happened. The first months are mostly about noticing that, and then rearranging how you spend your week around it.
This is the guide I would hand to somebody three weeks into the job. Not the theory of management, which you can get anywhere, but the specific things that go wrong early and are difficult to reverse once they have set.
The team you inherited has a history
You did not start this team. Whatever state it is in, it got there through decisions made before you arrived, by people who in many cases have left. In fifteen years I have watched teams that never really end, they just get reinvented with fresh faces in waves. Engineers leave, new ones arrive, the team continues, and the reasoning behind past architectural decisions goes with the people who made them. The tell is hearing somebody say, 'Ah, the previous team did that.'
Bruce Tuckman set out four stages of team development in 1965, and Mary Ann Jensen joined him in 1977 to add the fifth. It is the most useful map I know for working out what you have walked into.
The part the model does not tell you is that teams slide backwards. A reorganisation, two people leaving, a new joiner who changes the chemistry, and a performing team is storming again by Wednesday. That is not a failure of your management. Team performance drops at those points, and it is supposed to. Do not give yourself a hard time about it.
Work out which stage you are in before you start changing things. Storming looks like a problem and is usually progress. Forming looks like harmony and is usually silence.
The first ninety days
The instinct is to arrive with a plan. Resist it slightly longer than feels comfortable. The strongest thing you can do early is find out what is actually true, from the people who have been living in it, before you have committed publicly to a view.
That means listening in a structured way rather than a vague one, being honest that you do not know yet, and resisting the urge to demonstrate value by changing something visible in week two. I have written the longer version of this in your first 90 days as an engineering manager, which covers what gets you hired now, the instinct you have to break, and how to tell whether a team trusts its manager.
If you would rather see how your instincts hold up under pressure than read about it, the New Manager Simulator plays a first ninety days out in about three minutes and shows you where yours take you.
The conversation you are avoiding
There is one. There nearly always is, this early. Somebody on the team is not performing, or two people are not speaking, or a senior engineer is quietly undermining the direction, and you have decided to give it another two weeks.
Every new manager looks for a reason to delay that conversation, and the reason always sounds responsible. You want more evidence. You want to be fair. You are new and you do not want to come in swinging. What is actually happening is that the team already knows, and has known for longer than you have. Engineers carry an underperformer for a long time before a manager acts, and they are aware they are doing it.
The cost of waiting is not paid by you. It is paid by the people picking up the work, and they are watching to see how long you will let it run. That is the first real test of whether you are the manager or just the person with the title.
When you do have it, the failure mode is softening it until nothing is communicated. Most new managers open with something like this:
'How are you finding things at the moment? I just wanted to check in, no agenda really.'
They leave the room feeling the conversation went well, because it was pleasant. The person leaves with no idea anything was wrong. The version that works is not harsher, it just contains the actual information:
'I want to talk about the last two sprints. The payments work slipped twice without a heads up, and I found out from Priya rather than from you. I am not looking for an explanation right now. I want to understand what got in the way, and then agree what we do differently.'
Same tone. Specific situation, specific behaviour, and a clear reason for the conversation. The kindness is in the clarity, not in the softening.
Most of what goes wrong early is a variation on this, and the other variations are worth knowing before you walk into them: the mistakes new engineering managers make covers the ones I see most often, including the ones I made myself.
What you do is now the standard
The thing nobody warns you about is how closely you are being read. Your team is watching what you do far more than listening to what you say, and small things set the temperature. If you send messages at eleven at night, you have told them when the working day ends, whatever your policy says.
I deleted Slack and email from my phone. With them installed I found myself on the sofa checking work chat the way you check social media, and it had become a habit before I noticed it was one. That is a small mechanism rather than a principle, and small mechanisms beat good intentions reliably.
The other half of it is what you choose to measure, because the team works out very quickly what you actually care about regardless of what you announce. I worked at an agency with a thirty minute lunch break. I was six minutes over one day, and the lead developer was told to have a word with me about my timekeeping. That same day I had arrived twenty minutes early and stayed an hour past my contracted finish.
Nobody in that building was measuring output. They were measuring presence, because presence is easy to count, and everybody knew it. You will be tempted by the same thing when you cannot yet tell whether the work is going well, because hours are visible and progress is not. Resist it. Ask what shipped and what got unblocked. The moment you start counting who is at their desk, you have told the team that looking busy is the job.
Sarah Allen, Lead Product Manager at Yelp, put the distinction better than I have managed to: a leader can sometimes be a manager, but being a manager does not always mean being a leader. The title arrives on day one. The other thing is decided by a hundred small choices over the following year, most of which nobody will comment on.
Final thoughts
The first year is hard for almost everybody, including the people who turn out to be excellent at it. If you are finding it difficult, that is the normal experience of the job rather than evidence you were the wrong choice.
What surprised me most is how little of it is decision making. I expected the difficulty to be judgement calls, and most of it turned out to be sitting with somebody who is upset, or frustrated, or has just been told something they did not want to hear, and not rushing to make it better. That part does not appear in any job description, and it is most of the work.