The Interview Spine · Chapter 10 of 10
The first ninety days
You were hired on evidence of past work. You will be judged on the first quarter. These are different things.
Most of this site is about getting the offer. This chapter is about the thing that determines whether the next one is easier: the first quarter, where reputations set quickly and are expensive to change afterwards.
The trap
The instinct is to demonstrate value fast, which usually means proposing changes in week two. The problem is that you are proposing them from a model of the organisation built out of a hiring process, and that model is wrong in ways you cannot yet see. The changes that seem obvious in week two are usually the ones that were tried in 2024 and failed for a reason nobody wrote down.
A structure that works
| Weeks | Goal | What you are actually doing |
|---|---|---|
| 1–3 | Understand | Meeting people, reading incident reviews and post-mortems, and finding out how decisions really get made rather than how the org chart says they do |
| 3–6 | One visible win | Something small, useful and unambiguously yours — it buys the credit you will spend later |
| 6–10 | Form a view | Now you know enough to have an opinion worth having. Write it down |
| 10–13 | Propose and start | The larger change, argued with evidence you gathered rather than instinct you arrived with |
What to do in the first three weeks, specifically
- Read the last six months of incident reviews. Nothing tells you more about an engineering organisation in less time — what breaks, how they respond, whether the actions ship, and whether the language is blameless in practice or only in the template.
- Find out who actually decides. There is a decision-making structure that differs from the reporting structure in every organisation. Ask three people the same question — ‘who would have to agree to change X?’ — and the disagreement in their answers is the map.
- Ask what has already been tried. Before you propose anything. ‘Has anyone looked at X before?’ costs nothing and prevents the most common new-joiner mistake.
- Find the people who will tell you the truth. Usually a long-tenured engineer who is not in management. One such relationship is worth more than five with executives.
- Write down what surprises you. You lose the outsider's eye within about a month, permanently. The notes you take in week two contain observations you will not be able to make in week ten, and they are frequently the most valuable thing you bring.
The visible win
Pick something small, genuinely useful, and unambiguously attributable to you. The best candidates are things people complain about and nobody owns — a flaky test, a broken runbook, a piece of onboarding that wasted a day of your own first week. Fixing your own onboarding pain is close to ideal: you have direct evidence, it demonstrably helps the next person, and nobody will argue about whether it mattered.
What not to pick: a rewrite, a migration, or anything requiring several teams to cooperate. Not because those are wrong, but because you have no credit yet and they take a quarter to show a result.
If you were hired to change something
Sometimes the mandate is explicit — fix delivery, turn around reliability, professionalise the function. That shortens the observation window but does not remove it. What changes is that you should say out loud what you are doing: ‘I am spending three weeks understanding before I propose anything, and I will come back on the 24th with a plan.’ That converts a silence people would otherwise read as passivity into a visible, dated commitment.
The ninety-day review to run on yourself
Write down, in week one, what you believe about this organisation and what you expect to find. Read it in week thirteen.
The gap between the two is the most honest feedback you will get about your own judgement, and it is the thing that makes you faster at the next role. Almost nobody does this, which is exactly why it is worth doing.