There's a moment every new CTO knows. It's your first Monday. You have admin access, a calendar full of introductions, and a head full of opinions about the codebase you skimmed over the weekend.
Every instinct says: prove yourself. Find something broken and fix it. Show the team you're technical, show the CEO you're fast, show everyone the hire was right.
That instinct is how CTOs quietly lose their first 90 days.
I've spent years doing this job as a fractional and interim CTO — walking into companies mid-crisis, mid-growth, mid-political-meltdown, usually under NDA, which is why I publish under a pen name. The pattern is remarkably consistent: the technical work is rarely what kills a new CTO. What kills them is spending their scarcest resource on the wrong problem in the first three days.
Your scarcest resource is not time
On day one, you own something you will never have again: a fresh pair of eyes and zero scar tissue.
You don't yet know which topics are politically radioactive. You don't know whose project is secretly the CEO's favorite. You haven't learned to stop seeing the broken things everyone else has normalized. In three months you'll be just as habituated as everyone around you — you'll walk past the same quietly-broken deploy process, the same meeting that decides nothing, and you won't even register it.
That fresh perspective is a non-renewable asset. Spending it on reading code is like using a satellite to look for your keys.
The codebase will still be there in week three. The terrain — who actually decides things, what your CEO is actually afraid of, which fire is loud and which risk is deep — that's what you can only map cleanly while you're still an outsider.
Hours 0–24: Listen, don't fix
The first day has exactly one job: collect signal without emitting any.
Meet the CEO and your peer executives. Ask three questions, write down the answers, and promise nothing:
- "What does success look like in 90 days?" Not in general — specifically. You're listening for the gap between the job description and the real expectation. There is always a gap.
- "What's the one thing you're afraid I'll miss?" This question does two things. It surfaces the actual anxiety behind your hire, and it tells the person you're someone who asks about risk before claiming victory.
- "Who on the team do you trust, and why?" You're not collecting gossip. You're mapping where credibility already lives, because you'll need to borrow some of it before you've earned your own.
Then comes the hard part: resist every urge to comment on the code, the tooling, the process. You're mapping terrain, not grading it. The moment you start judging, people start managing what they show you — and your fresh eyes are gone, traded for politeness.
One more thing: write down what people repeat. When three people independently mention the same incident, the same team, the same migration-that-never-ends — that repetition is signal. It tells you where the real pain lives, and usually where the real politics live too.
Hours 24–48: Map the real terrain
Day two is about one uncomfortable truth: the org chart is fiction.
Somewhere in the company there's a real decision-making structure — the person whose objection kills projects, the Slack channel where things are actually settled, the engineer everyone routes around. Your job in hours 24–48 is to find it. Ask "who would need to agree for this to happen?" about a few recent decisions and watch the answers triangulate.
While you're at it, separate two things that everyone conflates:
- The loudest fire — the thing everyone complains about in your intro meetings.
- The deepest risk — the thing that could actually take the company down.
They are almost never the same thing. The loud fire has advocates, history, and emotional charge. The deep risk is usually quiet, boring, and owned by no one — an expiring certificate of trust between two teams, a single person who holds all the production knowledge, a billing system nobody has dared touch since 2021.
New CTOs who chase the loud fire get applause in month one and inherit the deep risk in month five.
And find one more thing: the system that is quietly broken but everyone has learned to live with. Every company has one. Nobody will name it in an intro meeting because it stopped being visible years ago. That thing is your future leverage — the fix nobody asked for that everybody feels.
Hours 48–72: Send one signal
By day three you've been listening for 48 hours, and the organization is starting to wonder what you actually do. Time to send exactly one signal — and only one.
Ship one small, visible, low-risk win. Not a refactor. Not a new process. Something a normal human can see: the flaky CI step that wastes everyone's morning, the on-call alert that pages people for nothing, the meeting that can be a document. Small enough to be safe, visible enough to be felt. The content of the win matters less than what it proves: you observe, and then you act.
Put exactly one expectation in writing. A short note to the CEO: here's what I'm going to focus on first — and, just as important, here's what I'm deliberately not touching yet. The second half is the part most new CTOs skip, and it's the half that protects you. Unspoken expectations are the raw material of month-three disappointment.
Lock the rhythm. One recurring slot with the CEO. One with your team. Before the calendar fills with everyone else's priorities. Cadence beats heroics — the CTOs who survive are rarely the ones who pulled the legendary all-nighter; they're the ones whose updates arrived every Friday, boring and on time, until boring became trust.
The three traps
If the first 72 hours have a failure mode, it's premature certainty. Three moves to explicitly ban:
- Announcing a reorg before you understand the terrain. You're redrawing a map you haven't read.
- Promising a big rewrite or migration to look decisive. You're signing a check your month-six self has to cash, with information your day-three self doesn't have.
- Taking a side in a political conflict you haven't mapped. Both sides are auditioning you. Declining to be recruited is a position, and it's the right one.
72 hours is the on-ramp, not the road
Everything above gets you to day four with your credibility intact and your map drawn. It does not get you a mandate, a budget, a diagnosis the CEO believes, or a team that ships. That takes a system, not a checklist — the first 90 days are won by operating cadence, not by inspiration.
I've packaged the on-ramp — the first-evening checklist, three fill-in worksheets, and a full chapter on running the 30-day diagnosis — as a free starter pack at the90daycto.com. It's the same material I use when I walk into a company cold. The rest of the system — securing the mandate in writing, modeling real capacity, winning the budget conversation, killing reliability theatre — lives in The 90-Day CTO Playbook.
But start with the 72 hours. They're free, they're concrete, and they're the part almost everyone gets wrong.
Mandate, not theatre.
— The Shadow CTO
(Why the pen name? I still run engagements under NDA and can't put client logos on a book — so there's no face and no case-names. Just the system.)
Top comments (0)