I coached a newly promoted VP of Engineering through her first quarter last year. Smart, technically respected, well-liked by her former peers, all the things the search-committee deck said. Six weeks in she was on the verge of quitting. She had restructured two teams, fired one senior engineer, redesigned the on-call rotation, and committed her org to migrate the deploy pipeline by end of quarter. None of it was wrong on the merits. All of it had been done too soon, with too little context, and at a velocity nobody around her could absorb. The team was in revolt, her boss was nervous, and she was wondering whether she should go back to being an IC.
She didn't quit. We rebuilt the next 60 days against a framework that's the closest thing I have to a reliable playbook for new department heads. The principle is simple: the first 90 days have three distinct phases, each with a different operating mode, and collapsing them into one is how most new leaders fail. Below is the framework, the questions to answer in each phase, and the unforced errors to avoid.
The arc of the first 90 days
Every successful new department head I've worked with has moved through three phases, even when they didn't have a name for them.
| Phase | Days | Mode | Primary output |
|---|---|---|---|
| Listen | 1–30 | Pure intake, zero changes | A defensible map of the org, the work, the politics |
| Diagnose | 31–60 | Hypotheses + small experiments | A written set of 3–5 priorities, validated against your boss and your team |
| Act | 61–90 | First visible decisions and structural changes | One or two changes that signal direction without overcommitting |
The error mode is starting Act on day 7 because the org has problems that are obvious from the outside and you feel pressure to "show up running." You can see the problems on day 7. You cannot yet see the load-bearing reasons those problems exist. The cost of a poorly-informed change in week one is paid for the next 12 months in trust.
Phase 1: Listen (days 1–30)
The discipline of the first 30 days is resisting the urge to change things you can already see are broken. This is the single hardest behavioral ask for high-performers, because the trait that got them promoted (decisive action, pattern recognition) is the trait they have to actively suppress for a month.
What to do instead:
1:1s with everyone two levels down. Not just direct reports — their reports too. The goal is not to evaluate; it's to listen. Three questions in every conversation:
- What does great look like in your job, and how often do you get to do it?
- If you could change one thing about how this team works, what would it be?
- What do you wish I knew that I probably don't?
Take notes. Don't promise anything. Resist the urge to "fix" any single piece of feedback in real time.
A skip-level with your boss. "Tell me what you're trying to accomplish over the next 12 months, what's blocking you, and how you'd describe this team's role in your plan." Their answer is the only definition of "success in this role" that actually matters. Re-confirm it in writing within a week.
Three customers / three stakeholders. Whatever your function, talk to three people who use the team's output. Engineering: talk to product and sales. Marketing: talk to sales and customer-success. Finance: talk to ops and the CEO. Their answer to "where does this team disappoint you?" is gold.
A formal artifact review. Read everything the team has shipped in the last 6 months — code, docs, decks, ad campaigns, depending on function. Don't critique. Just absorb the team's actual output shape and the implicit standards it represents.
What NOT to do in the first 30 days: announce new processes, restructure, hire, fire, change OKRs, redesign systems, "set the tone." Every one of these is a phase 3 action that loses 90% of its effectiveness when done in phase 1.
Phase 2: Diagnose (days 31–60)
By day 30, you have a thesis. Three or four problems are clearly real, two or three are suspected, and you've started to see the fault lines in the team's structure. Phase 2 is testing the thesis without committing to action.
Write the priorities document. Not for distribution — for yourself. "Based on what I've heard, I think the three things this team has to get right in the next six months are X, Y, Z. The biggest risks to each are A, B, C. The decisions I'll need to make in the next 60 days are…" If you can't write it in two pages, you don't understand the situation yet — go back to phase 1 for another two weeks.
Pressure-test with three people. Show the priorities document to: your boss, one peer department head, one senior person on the team. Not for approval — for friction. Where do they push back? What did you miss? What's the political subtext you didn't catch? Update the document.
Run two reversible experiments. Pick two small things that your thesis says will improve the team's output. A new meeting cadence. A different incident-review format. A change to how status gets reported. Make them small, time-bound ("we'll try this for four weeks and then decide"), and reversible. The point is not the experiment's outcome; it's that you start building the muscle of "this leader proposes, we try, we learn" before you need it for the bigger phase-3 decisions.
Identify the trust deposits to make. You will eventually spend political capital on a change someone disagrees with. You need a balance to spend from. In phase 2, find three concrete things you can do for the team — fix a long-standing annoyance, get them a budget approval, unblock a stuck process at the executive level. These are not strategy; they're trust deposits. They pay back later, with interest.
Phase 3: Act (days 61–90)
By day 60, you've earned the right to make visible decisions. The question is which ones, in what sequence.
The rule: one or two structural changes max in days 61–90. The temptation is to do five — you've been collecting them since day 7 — and the error is doing five. A team can absorb two significant changes per quarter. The third change is the one they remember as "the new leader didn't know what they were doing."
Pick your two from this list, no more:
- A change to who's on the team. A hire, a fire, a restructure, a swap. People-level changes are the highest-cost, highest-signal moves you can make.
- A change to what the team is responsible for. Take something on, give something away, redraw the scope boundary with a peer team.
- A change to how decisions get made. A new operating cadence, a new escalation path, a new ownership matrix.
- A change to what the team measures. A new metric, a retired metric, a redefinition of success.
Two of these, paired well, signal direction. Three or more signal panic.
Announce each change with three components: the observation (what you've seen in 60 days), the why (what the change is meant to achieve), and the what we'll measure (how we'll know in 90 days if it worked). The team needs to see you reason, not just act.
The unforced errors
In ten years of coaching new leaders, the same four mistakes account for most of the early failures.
- Premature firing. Letting someone go in the first 60 days almost always means you didn't have enough context to see why they were structurally underperforming or what role they were actually playing. The fire might be right, but it's almost never right now.
- Importing the playbook from the last job. What worked at your old company worked because of that company's culture, customers, and constraints. None of those transferred with you. Use the patterns; rebuild the playbook from scratch.
- The grand strategy memo in week three. "Here's my vision for this team" — written before you understand the team — is the artifact your team will mock for the next two years.
- Solo decision-making to "show leadership." New leaders sometimes interpret the mandate as "you're now the decider." You are; that doesn't mean you should decide alone. Especially in phase 1, distributing decisions to the people closest to the work earns more credibility than centralizing them.
The new VP of Engineering I worked with eventually made the restructure she'd attempted in week six — but in month four, after she'd done the listening, run the experiments, built the trust, and made the case. It was the same restructure on paper. It worked because by then she'd earned the right to make it. Most of leadership, I've come to believe, is just timing.
