Your first 90 days: aim for one honest close

The standard advice for a new engineering manager's first quarter is to instrument things: velocity, cycle time, DORA metrics, a dashboard for the leadership review. All of it measures work in flight. None of it measures the thing your judgement will actually be evaluated on — whether what you promise happens, and whether anyone can check.

So aim your first ninety days at producing one artifact instead: a commitment made with evidence, and closed honestly against what was promised. Everything else in this essay is just the staging for that.

Days 1–30: watch the door

Don't change anything yet. Find where requests actually arrive — the form everyone bypasses, the Slack channel, the sentence at the end of the leadership call — and watch what happens to them. Count two things: how many requests were promised before anyone established an owner, a size and the open questions; and how many were never answered at all. Those two numbers are your baseline, and they are almost certainly worse than anyone in the room believes.

Days 31–60: shape and refuse in the open

Now start holding a bar, visibly. When a request arrives thin, say what it still needs before it can be promised — not as obstruction, as a named list: the problem apart from the proposed solution, an owner, a first capacity figure, the dependencies, the unknowns written down. And when the answer is no or not-now, record it with the reason. You are not building process for its own sake; you are building the evidence trail your one honest close will sit on.

Days 61–90: close something

Pick a commitment that will finish inside the window and close it explicitly: delivered, delivered with exclusions, or not delivered — with what differed from the promise written next to the promise itself, and one lesson worth keeping. If the quarter went badly, close it badly on the record. A visible "not delivered, here is what we learned" is a stronger opening position than a dashboard that stayed green while the date slipped.

Why one close beats a quarter of dashboards

Because it is checkable. Leadership cannot verify a velocity chart and mostly knows it; what they can verify is whether the thing you said in planning matches the thing you said at the close. The first time those two statements line up — including the differences, stated by you before anyone asked — you have something no dashboard produces: a reason to be believed the next time you commit. That compounds. Ninety days is enough to start it, and only if the close is honest.