The First 90 Days as Chief of Staff: Playbook

7 Mistakes New Chiefs of Staff Make in the First 90 Days (And the Sequence That Avoids Them)

You weren’t hired to learn the company. You were hired to install the structure that lets the company stop being the CEO.

I’m sitting across from a Chief of Staff who’s been in the role for eleven weeks. She’s brilliant. Top-15 MBA. Two prior ops roles. She was promoted into this seat with a mandate to “transform how we operate.”

Photo by LinkedIn Sales Solutions on Unsplash

She tells me she cried in the bathroom yesterday. Eleven weeks of fourteen-hour days, and she still couldn’t name a single structural thing she’d built. Forty-seven unfinished tasks on her list. A CEO who kept adding more. 5 meetings a day. A leadership team that treated her like the new note-taker.

She’s not failing — she’s doing what most new Chiefs of Staff do in their first ninety days. And it’s exactly what will bury her by month six.

Nobody told her the part that actually matters.

The first 90 days aren’t about learning the company. Right now, every decision routes through your CEO. Your job is to install the structure that takes their place.

Most Chiefs of Staff become indispensable. The best ones become unnecessary.

If you got that job offer last month — congratulations and welcome.

The First 90 Days as Chief of Staff — Novatools.org

This is the playbook I wish someone had handed me in week one. Everything I’ve watched work across 40 companies, distilled into a phased, honest plan. Not a list of “tips for new CoS.” A field manual you can actually run.

Let’s start with what’s true.

The Brutal Truth Most CoS Onboarding Documents Don’t Say

You’re scared, and the title makes it harder to admit it. People expect you to know what you’re doing — you have “Chief” in your title — but you’re three weeks in and you can’t tell whether you’re succeeding or drowning.

You’re scared that the senior leaders quietly see you as the CEO’s errand-runner. Scared you don’t have permission to push back when they pile on more. Scared of the meetings where you’re expected to add value but don’t yet have context, so you fake it and worry every day about being found out.

Brené Brown calls this a vulnerability hangover — the morning-after of every high-stakes new role. You took the job because you wanted to be in the room. Now you’re in the room. And you have no idea what you’re supposed to be saying.

Name it. It doesn’t make you weak. It makes you a person doing a hard thing.

What month one through three actually feels like

If you’re in the first three months and any of this is familiar, you’re not broken. This is the role.

• The contradictory demands. You’re expected to be strategic and detail-focused. Independent and collaborative. Assertive and invisible. The job description told you what to do — nobody told you it came with built-in contradictions you’d feel in your body by 11am every day.
• The “what do you actually do” question. In your first month, three different people will ask you, in slightly different words, what your role actually is. Some of them senior. Most companies have never had a CoS before, and your colleagues genuinely do not know.
• The peer vacuum. You don’t have a peer at your level. The CFO is senior to you in their domain. The CRO too. So is the CTO. You can’t ask them for help without feeling like you’re admitting you don’t belong. There is no one to think out loud with at 4pm on a Tuesday.
• The lackey effect. A senior leader asks the CEO a serious question, and the CEO routes it to you. The senior leader takes the meeting, looks at you, and visibly decides you’re “the lackey” — there to take their answer back, not to be one. You’ll feel this in your body. You won’t always know what to do with it.
• The invisible-when-it-works paradox. When you do your job well, nothing breaks. When nothing breaks, nobody notices. You’re doing structural work that will not pay off in visible wins for two months, and in the meantime you have nothing to point at when someone asks what you accomplished this week.
• Imposter visibility. You’re suddenly in CEO-level conversations where you’re expected to add value but don’t yet have the context. So you nod, you take notes, you say “let me come back to that,” and you spend the rest of the day worrying you’ll be found out.
• The Sunday spiral. You haven’t really left work since you started. Every Sunday at 8pm, the dread starts. Not because the job is hard — because you can’t tell whether you’re winning.
• The 100%-certain truth nobody says out loud. Some people will be unhappy with you. Not might be — will be. You’ll deliver a “no” from the CEO and someone will resent you for it. You’ll redirect a senior leader’s project and they’ll go cold for two weeks. The role asks you to be a heat sink for hard decisions, and that comes with a real emotional bill.

None of this is failure. It is the texture of the role. The Chiefs of Staff who survive the first year don’t avoid these — they recognize them, name them, and stop reading them as evidence they don’t belong.

The fear is also what makes you cautious in the wrong ways. You overcompensate by being maximally available. You say yes to every meeting. You take on every action item. You smile and nod when the CFO patronizes you. You become indispensable through volume, not through structure.

That’s the trap!

Indispensable feels like the goal. It’s actually a cap on your career.

The Chiefs of Staff who go on to become COOs do something different. They build the operating system instead of becoming it. Within ninety days they make themselves unnecessary — in the best sense of the word — and the team can run without them in the middle of every conversation. Because they built that, they get the credit, and they get the next role.

This manual is how you do that.

Seven Principles for Your First 90 Days

Before any tactic, the principles. Following Ray Dalio’s style — write them down, test them, change them, but operate from them. Tactics without principles is just busywork in a nicer shirt.

1. Your job is to make yourself unnecessary. The team can run, decide, and execute without you in the room. That’s what unnecessary means here — not that you’re invisible, but that the structure carries the weight you used to carry alone. Optimize for that.

2. The first 90 days aren’t about learning the company. They’re about installing the structure that lets the company scale. You can learn the company forever. Structure is what changes things. Build the smallest piece of structure first.

3. Decisions are artifacts, not events. A decision made in a meeting and not written down didn’t happen. It will be re-litigated in 11 days. If you only do one thing in your first 90 days, install a decision log.

4. Behaviour follows structure. Don’t try to change behaviour. Change the structure that produces the behaviour. Telling people to “communicate better” never works. Giving people a meeting template that requires a written pre-read does.

5. Discipline beats motivation. Rituals beat both. Motivation is unreliable. Discipline is exhausting. The teams that scale don’t run on either — they run on rituals. Repeated practices that become “how we work here,” until the right behaviour is automatic and culture is the byproduct. That’s what most companies miss when they try to change culture through speeches and values posters. Culture isn’t declared. It’s built, ritual by ritual. Nova is where those rituals live.

6. Structure isn’t bureaucracy. Bureaucracy is structure without purpose. The fear most new CoS have is “I don’t want to be the bureaucrat.” Then don’t be one. Every structure you install must have a clear purpose people can name in one sentence. If you can’t, don’t ship it.

7. The system replaces you. That’s the point. You aren’t being replaced. The work you used to do — the chasing, the summarizing, the translating — is being replaced. You move up the stack. You design instead of execute. That is the upgrade you came here for.

Tape these on your wall. Read them on Sunday nights. They’re the test for every decision you make.

The 90-Day Map

Here’s the architecture before we get into the specifics. Six phases, fifteen days each, each with a clear key objective and a concrete deliverable.

That’s the sequence. It is not the only sequence.

Most onboarding plans get this wrong by trying to install OKRs in week 2.

By week 6 the OKRs are dead, the team resents you, and you’ve lost the political capital to install anything else. Sequence matters more than ambition.

You can find this playbook and all the processes I will be mentioning here:

→ First 90 days as Chief of Staff playbook: The pre-built operating system that turns a new CoS’s first 90 days from chaos and overwhelm into a phased, structural install — six 15-day phases, every template ready, every cadence wired up.

→ OKRs + Operating Cadence: The structured 4-week sequence to install company-level OKRs that don’t die in week six — from CEO-set Objectives, through honest cascade, to embedded check-ins on your existing cadence.

→ Good Management Rituals: This one is the manager’s operating system for their own pod or department — the rituals and habits that make a team of 5–15 actually function as a team rather than a collection of individuals.

Now, phase by phase.

Phase 1 · Days 1–15 · The Listening Tour

Key objective: Map reality without committing to anything.

This is the hardest phase to do well because it will feel like nothing is happening. You’re meeting people. You’re taking notes. Your CEO is starting to wonder when you’ll “do something.” Your friend asks how the new job is going and you don’t have a satisfying answer. Your instinct will scream at you to make a visible move — to prove the hire, to justify the title, to give yourself something to point at on Friday.

Don’t.

The first phase isn’t about doing. It’s about becoming someone who can see the system clearly enough to change it. Most new CoS skip this phase because passive feels like failure. Passive isn’t failure here. Passive is the work. You’re gathering the map before the march.

If you start solving in week one, you’ll solve the wrong thing — because you don’t yet know what’s actually broken. And worse, you’ll commit in week one. Committing without context is how new CoS lose credibility. The CMO will say “Sarah promised me X.” The CEO didn’t know. Now you have a political problem in week three you didn’t need.

Here’s what you do instead.

The method: 30 conversations in 15 days

Schedule 30 thirty-minute conversations across the first fifteen working days. That’s roughly two a day. Mix of stakeholders, in this order:

1. The CEO — three conversations, not one. Day 1 is “what do you want me to do.” Day 5 is “here’s what I’m hearing.” Day 12 is “here’s what I’d recommend.” Different conversations, different muscles.
2. Each member of the leadership team — one conversation, deep. (or a few people from each department)
3. A handful of senior individual contributors — usually 5–8 people. Engineering leads, senior PMs, top sellers. The people who actually know what’s broken.
4. Two or three frontline employees — pick at random. They will tell you things nobody else will.
5. Two recent leavers, if you can get them on the phone — they have nothing to lose by being honest.

The order matters.

Don’t start with the leadership team.

Start with your CEO so you understand the framing they want, then go wide. Going CEO-first prevents you from accidentally surfacing things in your CEO meeting that contradict what their own team has been telling them.

The 5-question listening tour script

Use the same five questions for every conversation. Variation kills synthesis. Five questions, in this order:

1. What is working really well right now that we should not break?
2. What’s the one thing that, if it were fixed, would make the biggest difference?
3. Where do decisions get re-made or re-litigated? What’s an example?
4. Where do you spend time you wish you didn’t?
5. What’s a thing nobody is naming that you wish someone would?

Question 5 is the one that matters most. It pulls out the unsaid thing — the dynamic between two VPs that’s blocking everything, the founder’s blind spot, the hire that isn’t working. **Things nobody names don’t get fixed. Your job for the next 90 days is to start naming them.**

The decision log starts on day one

Open a single document. Title it “Decision Log — [Company]”. Five columns: Date · Decision · Owner · Context · Status or if you are using Nova open a project and use the decision log.

Every decision you hear about — even ones from before you joined that are still live — goes in. By day 15 you’ll have 30–50 decisions logged, and roughly a third of them will be ones the team thinks are settled but actually aren’t. That’s gold!

What to never do during the listening tour

• Don’t promise anything. Not even small things. Especially not small things — small promises become how people characterize your judgment.
• Don’t share what one person said with another, even neutrally. People are testing whether you keep what they said in confidence. You’re either trustworthy or you’re not, and that gets decided in your first three weeks.
• Don’t use the word “we” yet. You aren’t “we” until you’ve earned it. You’re “I’m new, I’m listening.” Earn the “we” by week six.
• Don’t have a final opinion in meetings. Not one. Not even when asked. “I’m still in listening mode, I’d like to come back to you on that” is your most-used sentence for two weeks.
• Don’t try to fix anything. Even obvious things. Your credibility comes from the system you install, not from one heroic save.

Must-do in Phase 1

• Three conversations with your CEO, not one
• 25–30 stakeholder conversations completed
• Decision log live with at least 30 entries
• A written one-page synthesis of what you heard, kept to yourself for now
• Zero commitments made

What “good” looks like at day 15

You can answer three questions in writing, in one paragraph each:

1. What’s the company actually trying to do this year, in plain language?
2. What are the three things blocking it that nobody is naming?
3. What’s the one piece of structure that, if installed, would unlock the most?

You don’t share these answers yet. You sit with them. Phase 2 is where they become visible.

Phase 2 · Days 16–30 · The First System You Install

Key objective: Install one small structural ritual that holds.

Most new Chiefs of Staff try to install OKRs in week two. It fails every time. Installing OKRs requires political capital you haven’t earned and a baseline of operational hygiene the company doesn’t have. OKRs aren’t the problem — the timing is.

The thing that installs first, and best, is far less ambitious. It’s the weekly leadership sync.

You won’t think of this as ambitious. It is. Done right, it changes the company in ten weeks. Done wrong, it dies in three.

The Weekly or Monthly Leadership Sync

The method: rebuild one meeting

Pick the most important recurring leadership meeting in the company. Usually it’s a Monday or Tuesday meeting attended by 6–10 people, runs 60–90 minutes, and currently consists of round-robin status updates that no one remembers afterward.

That meeting is your beachhead.

You’re not adding a new meeting. You are taking a meeting that already exists and rebuilding it from the inside. This is critical: new meetings die. Restructured meetings stick.

The three-pillar structure

The three-pillar structure — Novatools.org

Replace the round-robin with three pillars, in this order:

1.Decisions (40 minutes) — items that need a decision this week, with the decision owner identified before the meeting. Pre-read circulated 24 hours in advance.

The meeting makes the decision — it does not “discuss it further.”

— The pre-read rule: Nothing is decided without a pre-read. This is non-negotiable in both models, but it’s especially important here because the decision surface is smaller. The pre-read format is simple: here is the decision we need to make, here is the context, here is what we recommend, here is what we need from the group. One or maybe 2 pages. (I also use video) Circulated 24 hours in advance. If the pre-read doesn’t arrive on time, the item is pulled from the agenda. That rule — pull the item if the pre-read is late — will feel aggressive the first time you enforce it. It will be respected by week four. It teaches the organization that the meeting’s time is not available for underprepared thinking.

2. Blockers (15 minutes) — what is one team blocked on that another team can unblock. Specific. Named.

3. Signal (10 minutes) — one metric or qualitative read from each leader. Not status. Signal. “What changed in your part of the world this week that the rest of us should know.”

Status updates die. They are replaced by an async written update people send Friday afternoon — two paragraphs max — that the leadership team reads before Monday. The meeting is for the things async cannot do.

An Alternative Worth Knowing: The Working Rhythm Model — A Full Breakdown

The model rests on a single premise: most leadership teams make poor decisions not because they lack intelligence, but because they lack shared context. Each leader arrives to a meeting knowing their own function deeply and everyone else’s function barely. The three-pillar model accepts that condition and routes around it. The Working Rhythm Model tries to fix it.

The Working Rhythm Model tries to fix it.

Here’s what each element is actually doing.

1. Pulse (15 minutes)

The mechanism

Each leader gives their team a number from 1 to 5. One sentence of context. That’s it — unless the number is a 2 or below, at which point the group has a brief conversation.

The scale isn’t defined in advance beyond the poles. 1 is we are in trouble. 5 is we are strong. The space in between is left deliberately loose, because the goal isn’t precision — it’s signal. A leader who says “3 — we shipped but morale is soft” is giving the room something they couldn’t get from a dashboard.

Why it comes first

Most leadership meetings begin with agenda items. The implicit assumption is that everyone is arriving in roughly the same emotional and operational state. They’re not. Someone’s team just lost a key person. Someone’s product just had a bad week with customers. Someone is carrying a conversation they had yesterday that they haven’t processed yet.

Pulse surfaces all of that in four minutes per person — without making the meeting about it. It does two things simultaneously: it creates an honest baseline, and it tells you where the decisions later in the meeting will be distorted by pressure that hasn’t been named.

If you skip Pulse and go straight to decisions, those pressures don’t disappear. They show up as defensiveness, silence, or false agreement. Pulse is how you metabolize them before they infect the agenda.

The discipline it requires

The rule — no elaboration unless the number is 2 or below — is not arbitrary. It’s load-bearing. The moment you allow elaboration above 2, Pulse becomes a status update. Every leader will want to explain their 3. Some will perform a 4 when they’re a 2. The facilitator’s job is to hold the format without embarrassing anyone.

Over time, Pulse does something more interesting: it creates a longitudinal record. If the same function reads 2 or 3 for six weeks in a row, that’s a pattern the leadership team owns together — not a problem one person is hiding.

The failure mode

Leaders who are new to this format treat it as a performance. They give 4s when they’re 3s because they don’t want to look weak. The fix isn’t confrontation — it’s patience. When the first leader who honestly says “2 — we’re not going to hit this quarter’s number” is met with support instead of judgment, the room recalibrates. It takes a few cycles.

2. Focus (30 minutes) — one pre-selected topic per week, rotated by function. Not a decision. A deep read. The goal is that every leader understands every function well enough to make better cross-functional decisions the rest of the week.

Why this is structurally different from a presentation

A presentation is designed to inform. A Focus session is designed to create empathy and operational understanding.

A good Focus session from the Head of Sales doesn’t begin with pipeline numbers. It begins with something like: “I want to show you how we actually think about a deal from first call to close, because I think most of you have a mental model of this that’s about 60% accurate, and the 40% gap is where we’re creating friction.”

Then they walk through it. The questions that come from engineering, from product, from finance during that 30 minutes are usually the most valuable cross-functional conversation the company has had in months. Not because the problems are new — but because they’ve never been named in a room where the right people were present.

The pre-read rule

Nothing is decided without a pre-read. This is non-negotiable in both models, but it’s especially important here because the decision surface is smaller. You have 20 minutes, not 40. A decision that arrives cold — without a written framing, a clear recommendation, and a named decision owner — will consume the entire block in orientation.

The pre-read format is simple: here is the decision we need to make, here is the context, here is what we recommend, here is what we need from the group. One page. Circulated 24 hours in advance. If the pre-read doesn’t arrive on time, the item is pulled from the agenda.

That rule — pull the item if the pre-read is late — will feel aggressive the first time you enforce it. It will be respected by week four. It teaches the organization that the meeting’s time is not available for underprepared thinking.

3. Decisions (20 minutes) — same as pillar one above, but with a smaller surface area because the Pulse has already surfaced the urgent.

4. Commitments (5 minutes) — each leader states one public commitment for the week. Not a task. A result. Written down. Reviewed the following week, first thing.

The key difference: the three-pillar model is a decision engine. The Working Rhythm Model is a shared consciousness engine. Neither is superior. Choose based on the company’s failure mode. If leadership makes bad decisions, install the three-pillar model. If leadership makes decisions in silos, without enough context, install the Working Rhythm Model.

In either case, the same rule applies: don’t start from scratch. Take the meeting that exists. Rebuild it from the inside.

The Monthly Ritual: Deliverables & Priorities

Install this in week three, but don’t run it until week five. You need one full cycle of the weekly sync before this ritual has anything to land on.

What it is

A monthly working session — 45 to 60 minutes per manager-team pairing — that does one thing the weekly sync cannot: it forces deliberate prioritization at the individual level, and it puts the manager on record about what actually matters.

This is not a check-in. It is not a performance review. It is a priority contract.

Why it works

Most managers carry a mental model of what each person on their team is doing. It’s incomplete, outdated, and never stress-tested. This ritual forces the manager to externalize that model before the meeting — and the act of writing it down exposes every gap. When you have to name what someone’s top deliverables are for the month, you quickly discover how little precision you’ve been operating with.

The team member does the same. When the two lists meet, the conversation almost writes itself.

The structure

Before the meeting, the manager fills in their section of a shared document — one per team member. No templates. Just this:

• What I believe your top deliverables are this month
• The one thing that, if delivered, would be a significant achievement
• What I think you need to start now that you may not have started

The team member opens the document at the start of the meeting. They read in silence — no speaking. Then they add their own deliverables:

• Additional deliverables I have on my plate that aren’t captured above
• On my radar: things I’m tracking that I haven’t flagged yet

Then the conversation begins. Not about tasks. About deliverables. The distinction is load-bearing: a task is something you do; a deliverable is something that exists when you’re done. Managers who conflate the two build teams that are busy but not productive.

The four elements in the document

Top Deliverables — the primary outputs expected this month. Written by the manager first, reviewed and amended by the team member. These become the single source of truth. If priorities change mid-month, this document is updated. Not the manager’s memory. Not Slack. This.

Plant the Seed — things that must be initiated this month even if they won’t be completed this month. Long-lead work, relationship-building, early research. The trap for most teams is that they only count things that finish. Plant the Seed makes starting count.

Ambition of the Month — one deliverable that, if achieved, would be considered a real win. Not a stretch goal in the OKR sense — this isn’t a target. It’s a flag in the ground. It says: this month, I’m aiming at something that matters, and I want you to know it. It changes how the team member makes tradeoffs during the month.

On My Radar — a free field. Things the team member is watching, thinking about, or quietly worried about. Not commitments. Not yet. But surfacing them in a structured context allows the manager to either elevate them, deprioritize them, or simply acknowledge that the team member is thinking ahead. This field, more than any other, reveals how your best people think.

The Monthly Ritual: Deliverables & Priorities — Novatools.org

The operating rule

The document or session is not reviewed once and filed. It is the single source of truth for that manager-team member relationship during the month. When something changes — a new project lands, a deadline moves, a deliverable gets deprioritized — the document or session is updated. Not in a meeting. Not in a retrospective. In the moment, by whoever owns the change.

This removes a category of organizational dysfunction that most companies treat as inevitable: the gap between what was agreed and what was remembered. The gap is not a memory problem. It is a structural problem. This ritual closes it.

What You’re Actually Installing

By the end of Phase 2, you have two rituals running:

Weekly — a leadership meeting that produces decisions, surfaces blockers, and keeps the team reading the same signals.

Monthly — a manager-team ritual that produces clarity, forces deliberate prioritization, and creates a shared written record that outlasts anyone’s memory.

Neither ritual is complicated. Neither requires new software, new frameworks, or a culture change initiative. They require consistency, and they require you — for the first month at least — to be the person who holds the format when the impulse to drift back to the old way is strong.

That impulse will come. It always does. The meeting will try to become a status update again. The monthly document will get skipped because the month was busy. Your job in Phase 2 is not to install rituals. It is to install the discipline that lets rituals survive.

The decision-log connection

Every decision made in this meeting goes into the decision log. An any action will have an owner. With a deadline. In the room, not after.

This is the moment you stop being a meeting-summarizer and become an operating-system-builder. The summary used to live in your head and your notes app. Now it lives in a place the team owns. You are no longer the database. You are the architect of the database.

This single shift — meetings produce logged decisions instead of forgettable conversations — is worth more than any other structural change you’ll make in your first year.

What to never do in Phase 2

• Don’t make it a “you” meeting. The CEO runs the meeting. You run the structure. There is a difference and the team will read it instantly.
• Don’t add a second meeting. Your CEO doesn’t need another calendar item. Replace, don’t add.
• Don’t bring a 30-page agenda template. A one-page agenda. Three sections. Seven minutes to read.
• Don’t take notes for everyone. The notes are the decision log. Owners log their own tasks and decisions related to those tasks. You’re not the scribe.

Common Phase 2 mistakes

• The ambition trap: trying to install OKRs, decision rights, a new tool, AND the meeting redesign all at once. Pick one. One small thing that holds beats five things that collapse in week six.
• The CEO-as-spectator trap: designing the meeting and asking the CEO to “just show up.” They need to introduce it, defend it, and own it. Otherwise it’s “Sarah’s meeting,” and Sarah’s meeting dies the first time Sarah is sick.

Phase 3 · Days 31–45 · Building the Rhythm of Business

Key objective: Make the cadence visible, layer by layer.

Once one meeting is fixed, the temptation is to fix all of them. Don’t. Fix the cadence instead. The cadence is the system of meetings, not any single meeting.

Get the cadence right and the meetings get easier to fix one by one.

The four-layer cadence

Most companies have meetings. Few have a cadence. The difference: a cadence is a deliberate set of forums at four time horizons, each with a clear job, each connected to the others.

OKRs & Operating Cadence — Novatools.org

These are the only four meetings that should exist at the leadership layer. Everything else should be a 1:1, a project meeting, the monthly priority meeting or async.

If you look at your CEO’s calendar today and they have 14 different recurring meeting types, you have a cadence problem.

Your job over the next 15 days is to compress those 14 into 4 (max 6)

→ OKRs + Operating Cadence: The structured 4-week sequence to install company-level OKRs that don’t die in week six — from CEO-set Objectives, through honest cascade, to embedded check-ins on your existing cadence.

How to actually do this without firing 10 meetings in week one

You don’t kill meetings. You consolidate them. Quietly. One at a time.

• Find a recurring meeting that overlaps with the weekly sync’s job. Propose folding it in.
• Find a recurring meeting that has no clear job. Ask its organizer what its purpose is. Often they don’t know either. That’s how the meeting dies — politely, by making someone admit they’re not sure why it exists.
• For each meeting you keep, give it the three-pillar treatment. Decisions, blockers, signal. Or kill it and replace it with async or use the methodologies about how to lead great meeting: OFQ Meetings.

Aim to remove three to five recurring meetings in this phase. Not ten! Three to five is enough to free 4–6 hours per leader per week. That is enormous. That is the gift that buys you political capital for the next six months.

The async-first rule

If a meeting’s job is “share information,” it should not be a meeting (90% of the time) . It should be a written update or async.

This is harder than it sounds. The instinct of senior teams is to “talk through it.” Talking through it produces no artifact, no decision, no follow-up. No progress. The conversation evaporates. Decisions are artifacts, not events. If you want it to count, write it down.

Your weekly sync’s pre-read becomes the rule-setter for the rest of the company. When the leadership team starts sending two-paragraph written updates on Friday and showing up on Monday already aligned, the rest of the company notices. Behaviour follows structure. Within six weeks, the VPs are running their own teams the same way.

The personal cadence layer

Don’t forget your own cadence. The CoS who builds a perfect company-wide cadence and lives in chaos is a common failure mode. Block these on your own calendar by day 35:

• 15 minutes every morning, before Slack — what’s the one outcome today, what are the three decisions to push, what blocks me?
• 5 minutes after each meeting — log decisions. Just log. Don’t process. Just log.
• 30 minutes Friday afternoon — close the week. What was decided. What carries forward. What was learned, and what do I need to “leave ready for Monday.”
• 60 minutes monthly, last Friday — your own retrospective. What’s working in your role. What isn’t. What needs to change.

Most CoS skip these. They feel selfish. They aren’t.

You can’t install structure for the company if your own life is unstructured. Write Brené Brown words on a post-it if it helps: you can’t pour from an empty cup.

Make your own cadence, habits and rituals non-negotiable.

What to avoid in Phase 3

• The cadence purist trap: trying to install all four layers in fifteen days. Pick the weak link. Usually it’s the monthly review. Fix that one. The others can wait until Phase 5 or 6.
• The meeting-killer trap: announcing a “meeting cleanup initiative” and slashing things publicly. People defend their meetings. Kill them quietly, by making the alternative obviously better, provide support, training and give them the tools they need.
• The personal-cadence-skipper trap: building everyone else’s cadence while running your own life on adrenaline. Three months of that and you’re done. The model is the message. If your life is chaos, no one believes the structure works.

Must-do in Phase 3

• Four-layer cadence written down and visible to the leadership team
• 3–5 meetings consolidated or killed
• Weekly sync still running cleanly (don’t break Phase 2’s win)
• Personal cadence on your own calendar

What “good” looks like at day 45

A new VP joins the company on day 46. You can hand them a one-page document called “How we operate.” On it: the four-layer cadence. The decision log link. The weekly sync template. The async update format.

They start running on day one because the system is legible.

That is the payoff.

Phase 4 · Days 46–60 · Your First Real Test

Key objective: Navigate your first cross-functional break and produce a logged decision the org accepts.

Around day 50, something will break.

Always.

It looks different every time. Two VPs in open conflict over a roadmap call. A senior hire who’s underperforming and nobody wants to address. A customer escalation that needs an executive call nobody wants to make. A budget reallocation request that pits Engineering against Marketing.

This is the first real test of whether the structure you’ve installed actually holds. Up to this point, you’ve been working on calm-water exercises. Now the wind picks up.

Most new CoS make the same mistake here: they try to mediate it personally. They schedule a 1:1 with each person. They translate between them. They craft compromise proposals. They become the diplomat. And the conflict gets “resolved” — meaning it goes underground, where it festers for two months and resurfaces sharper, and by then you’re complicit in not having addressed it the first time.

There’s a better way.

Create a vision that clears the way forward — build the rails for progress instead of being the rescue boat for every stumble.

Run the conflict through the structure, not around it

The structure you’ve installed is a decision-making structure.

Use it.

1. Name the conflict explicitly, in writing. “VP A and VP B disagree on X. The decision needs to be made by [date]. The two positions are [position A] and [position B]. The decision criteria are [criteria].” You write this. Not them. Send it to all involved before the meeting.
2. Bring it to the weekly leadership sync as a Decision item. Time-boxed to 20 minutes. Pre-read goes out 24 hours in advance with both positions clearly stated and the trade-offs honestly listed.
3. The CEO makes the call in the room — or explicitly delegates the call to one of the parties. Either way, one person owns the decision and the decision is logged.
4. The decision goes in the log with the rationale. Including the dissenting view. Especially the dissenting view. Future-you, six months from now, will need to know why the decision was made the way it was.

This sequence does three things at once. It forces the conflict above the surface where it can be resolved. It demonstrates the structure can hold under stress. And — this matters more than people realize — it gives the losing side a way to lose with dignity. Their position was heard, written down, considered, and explicitly recorded as the path not taken. That is the difference between someone who carries resentment and someone who commits to the chosen path.

Brené Brown’s research on workplace courage applies directly here: people don’t leave companies because they got a “no.” They leave because the “no” was given without acknowledgment. Honest disagreement, fairly resolved, builds more trust than easy consensus.

Here is a great rule: believability-weighted decision-making. On any contested call, the people whose past track record on this kind of decision is strongest should weigh more. The CFO has higher believability on a budget call than the CMO does. The Head of Engineering has higher believability on a technical architecture call. That doesn’t mean the others don’t get heard. It means the call defaults to the highest-believability voice unless someone makes a credible counter-case.

This sounds harsh. It is much kinder than the alternative — which is endless re-litigation by people whose stake in the decision is opinion rather than expertise. Decisions get re-litigated when ownership is unclear. Believability weighting makes ownership clear.

What to never do in a cross-functional break

• Don’t take sides privately while staying neutral publicly. Everyone will eventually find out. Your credibility never recovers.
• Don’t mediate in 1:1s. It feels productive. It teaches the org that conflicts get resolved through Sarah’s diplomacy, which guarantees you mediate every conflict for the next two years.
• Don’t let the conflict drift past the next sync without being formally addressed. Drift is what kills companies. Force the decision into the log.
• Don’t go to the CEO in private to “give them a heads up” about something that should be addressed in the room. That’s how leadership teams become CEO-and-the-CoS versus everyone else.

Must-do in Phase 4

• Surface the conflict in writing, with both positions and the trade-offs
• Run the decision through the weekly sync, not around it
• Log the decision with rationale, owner, and the dissenting view
• Don’t take sides privately

What “good” looks like at day 60

A real conflict was named, surfaced, decided, and logged — and the team is still functional. You did not personally resolve it. The structure resolved it, and you held the structure. The losing side accepted the outcome because the process was honest. The CEO trusts the decision-making process more than they did at day 45.

That is when your job security stops being a question.

Phase 5 · Days 61–75 · Installing OKRs (Now You Can)

Key objective: Install goals — now that the foundation can start holding them.

You’ve been waiting two months for this. OKRs are the shiny thing. Every leadership book says install them in week one.

Don’t.

OKRs without the foundation you’ve spent 60 days building are theater. They become a Q1 ritual everyone forgets by week three. OKRs require a working decision-making rhythm to function — without the weekly sync, the decision log, and the cadence underneath them, OKRs are just a list of aspirations. With those, OKRs become a steering mechanism.

You have the foundation now. Install the OKRs.

The four-week OKR install

You don’t roll out OKRs in a meeting. You roll them out across a four-week sequence, the same way you’d roll out any structural change.

Week 1 (days 61–67): Set the constraint. The CEO writes 3 company-level OKRs. Three. Not seven. Not twelve. Three. (MAX 5) You help draft them. They’re aspirational but specific. Each O has 2–3 KRs. Each KR is a number, not an adjective. “Improve customer satisfaction” is not a KR. “Increase NPS from 32 to 45” is.

Week 2 (days 68–74): Cascade with friction allowed. The leadership team takes the company OKRs and sets their own. Friction is allowed and welcome. If the VP of Engineering says “we can hit two of these but not three at this resourcing,” that’s a conversation about prioritization, not insubordination. Capture the conversation in the decision log.

Week 3 (days 75–81): Each team writes their OKRs publicly. Not in private documents. In a shared place, where every team can see every other team’s OKRs. This is uncomfortable and important. Misalignment is easier to fix when it’s visible. If Marketing’s OKRs and Sales’s OKRs contradict each other, you want to know that in week three, not in month three.

Week 4 (days 82–88): Lock and start. OKRs are final. Each team has a monthly priority check-in cadence, and every person in the team got their personal goals. Each leader reports OKR signal at the monthly business review. Nothing is secret.

Sometimes this phase can take 2–3 more weeks.

How many OKRs (the number that’s actually right)

The single most common OKR mistake at Series A and B is having too many. The right number for the company:

• 3 company-level Objectives. Not five. Three.
• 2–3 Key Results per Objective. No more.
• Each leader’s team has 2–4 Objectives that ladder up.

OKRs Discussion Wall — Nova

If the CEO has seven priorities, they don’t have priorities. They have a wishlist. Your job in this phase is to be the person who says “we’re picking three.” It is one of the highest-value things a Chief of Staff ever does.

Most CEOs have never had someone in their lives whose role was to enforce focus and remove distraction.

You are now that person.

→ OKRs + Operating Cadence: The structured 4-week sequence to install company-level OKRs that don’t die in week six — from CEO-set Objectives, through honest cascade, to embedded check-ins on your existing cadence.

Why OKRs die in week six (and how to prevent it)

OKR rollouts fail in a predictable way. Week 1 is exciting. Week 4 is launched. Week 6 is silent. By month two, no one mentions them.

The cause is structural: OKRs were treated as an event, not a ritual. The fix is to embed OKR review into the cadence you already built.

• Monthly business review = OKR signal review. Each leader reports on their OKRs in 5 minutes. Green / yellow / red. Plain language about what’s changed.
• Quarterly planning = formal OKR retro and reset.
• Monthly leadership sync = surface OKR-related decisions and blockers as they arise.
• Monthly deliverables and priorities with each team

Notice you’re not adding new meetings. You’re loading OKR review into the meetings that already exist. Behaviour follows structure. The structure already runs. OKRs ride on top.

What to never do during OKR install

• Don’t write the OKRs for the CEO. Help draft. Pressure-test. Refine. Don’t author. They have to own them or they don’t survive contact with reality. Another way to lead OKRs is to start with a workshop and then after that discussion the CEO will share the proposal.
• Don’t use OKR software in week one. Run it in a doc or spreadsheet for the first quarter or use a simple whiteboard for the first one. Find the perfect way to do it later once you define the process to get there.
• Don’t set KRs that are activities. “Launch the new product” is not a KR. “Generate $500K in pipeline from the new product by end of Q1” is.
• Don’t allow more than three company-level Objectives. Even if everyone wants them. Especially if everyone wants them.

Two primary reasons OKRs fail are when resources and company reality do not support the objectives. Specifically, if processes, leadership support, and budgets are not aligned, goals are set but teams fail to have the necessary conversations about what must happen systematically to achieve them. This includes identifying the processes needed for speed and clarity, how leaders must support the team, and ensuring investments are approved. This alignment is the “real work” of OKRs; setting goals is easy, but the difficult part is understanding the reality, strengths, and weaknesses of the system.

The second biggest mistake is failing to translate goals to the individual level. Goals are often communicated, but the day-to-day work lacks the systems to support the team. As a result, employees “spin their wheels” — spending more time coordinating or trying to get answers from another department than actually collaborating or doing the work. This leads to a loss of clarity regarding priorities and how success is measured.

It must be non-negotiable for leaders to translate company goals into individual ones. In one-on-one meetings, leaders should specify: “This is your role and your responsibility; it will impact this company objective. If done well, we will move from [where we are now] to [the desired result].” Finally, monthly team meetings are essential to maintain priority alignment. Trust me, priorities are always under attack, so that monthly “source of truth” is something you cannot skip.

Common Phase 5 mistakes

• The seven-objective mistake: the CEO insists they need more. Hold the line. Or rather, help them hold it. Frame: “If we hit all seven we’ve succeeded. If we hit only three, can we still call this year a win? Then those three are the OKRs.”
• The cascade-without-friction mistake: rolling out company OKRs and demanding teams accept them silently. Friction is information. Lack of friction at OKR rollout means you set them too low.
• The vanity-metric mistake: KRs that look impressive but aren’t tied to outcomes. Followers. Traffic. Hours of training delivered. None of those are KRs. Measure outcomes, not effort.

What “good” looks like at day 75

A new senior IC joins the company on day 75. By the end of their first week, they can answer three questions without asking anyone:

1. What are we trying to do this quarter?
2. How does my team’s work connect to that?
3. How will we know if we won?

If they can answer those, the OKRs are working. If they can’t, you have more work to do — but at least you’ll know exactly where the gap is.

Phase 6 · Days 76–90 · Your First Quarterly Review

Key objective: Run a complete planning cycle — the meeting, the retro, the carry-forward.

You’re now in the final phase of your first 90 days. The cadence is running. The OKRs are live. The decision log has 100+ entries. The leadership team trusts the structure more than it did when you arrived.

Quarterly planning is the moment when everything you’ve built either compounds or collapses. Done well, this is the meeting that makes your job impossible to do without.

The 4-week prep, compressed into 2

A real quarterly planning cycle takes four weeks of preparation. You don’t have four weeks. You have two. Here’s the compressed version.

Week 1 (days 76–82): Pre-read. Each leader sends a one-page pre-read. Three sections:

• What did we commit to last quarter? What did we hit? What did we miss? Why?
• What’s changed in our world that matters?
• What do I want from this planning cycle?

You aggregate the pre-reads into a single packet. Sent to the leadership team 48 hours before the planning meeting. This is non-negotiable. No pre-read = no participation.

Week 2 (days 83–89): The meeting itself + retro + carry-forward.

The meeting is half a day. Four hours, broken into:

1. Hour 1: Retro. What worked. What didn’t. No new ideas yet. Just look at last quarter honestly. The decision log is open on screen. Decisions that didn’t ship — why? Decisions that were re-litigated — why?
2. Hour 2: New OKRs. Draft. Iterate. Pressure-test.
3. Hour 3: Trade-offs. What are we not doing this quarter? Make this list explicitly. Write it down. The list of what you’re not doing is more important than the list of what you are.
4. Hour 4: Cascade plan. How does this get to teams in the next two weeks. Who tells whom. What artifacts go to the all-hands.

Then a one-hour retro on the planning meeting itself. What worked about how we just planned? What should we change next quarter? This is the meta-loop that makes your operating system get better every quarter instead of staying frozen.

The agenda template

Steal this. Use it next quarter. Iterate it after.

QUARTERLY PLANNING — [Date]
Attendees: [Leadership team]
Pre-read: [Link to packet — REQUIRED reading]

1. Retro on Last Quarter (60 min)
   • What we committed to
   • What we hit / missed
   • Decisions that worked / didn't
   • Patterns we want to keep / kill

2. New OKRs (60 min)
   • Company-level — 3 Os, 2–3 KRs each
   • Trade-off discussion: what doesn't make the cut

3. Cascade & Communication (60 min)
   • Who owns the cascade to each function
   • Communication artifacts (all-hands deck, FAQ, written memo)
   • Timing of rollout

4. Meta-retro: How did we plan? (30 min)
   • What worked about this meeting
   • What we change next quarter

That’s it. Four sections. Half a day. The meeting that used to take two days because it had no structure now takes four hours and produces more.

The carry-forward — the part most companies skip

After the planning meeting, do not let the team disperse without a carry-forward document. One page. Three sections:

1. What we decided this quarter (with owners)
2. What we explicitly chose not to do (the kill list)
3. What we want to do better next time we plan

This document goes in the decision log. Next quarter, the first thing we do in the new planning meeting is read this document. That is how operating systems compound. Each quarter is built on the explicit lessons of the last one. Without it, every quarter starts from a blank page.

What to never do in Phase 6

• Don’t run quarterly planning without a pre-read. It becomes a 4-hour brainstorm. Brainstorms produce no plan.
• Don’t skip the retro. This is where the company learns. Pain plus reflection equals progress.
• Don’t let the kill list be private. What you’re explicitly not doing is the most important thing you can communicate to the org.
• Don’t try to plan beyond the next quarter. This is a quarterly meeting. Annual planning is a different beast. Don’t conflate them.

Common Phase 6 mistakes

• The all-hands-as-planning mistake: trying to do company-wide planning in the leadership planning meeting. Two different meetings. Plan with leadership. Communicate to the company afterward.
• The status-update-disguised-as-retro mistake: the retro becomes a round of “here’s what my team did.” That’s not a retro. A retro asks why things went the way they did. Why is the unit of learning.
• The “we’ll figure out cascade later” mistake: ending the planning meeting without a cascade plan. The team disperses. Each leader interprets the OKRs differently. By the time it gets to ICs, the message is fragmented. Cascade is part of planning, not aftermath.

Must-do in Phase 6

• Pre-read circulated 48 hours in advance
• Half-day planning meeting with the four sections
• Carry-forward document in the decision log
• Cascade plan with named owners
• Meta-retro on how you planned

What “good” looks like at day 90

The new quarter starts. Every team can name the company’s three OKRs and how their work ladders up. The first weekly leadership sync of the new quarter opens with a check on the OKRs — naturally, because OKRs ride on the cadence you built. Your CEO, in conversation with their board, mentions: “We’ve completely changed how we operate this quarter.”

That sentence is the receipt for everything you’ve built.

Day 90 · What Success Actually Looks Like

You have arrived at the symbolic finish line. Most CoS field manuals end here. This one doesn’t — because the real test of your first 90 days is whether the structure you built can survive without you holding it up.

Here is the day 90 self-audit. Answer honestly. Pain plus reflection equals progress.

Operating signals — green if true, red if not:

• The weekly leadership sync runs in 60 minutes or less, every week, and produces a logged decision.
• The decision log has 100+ entries. New ones are added weekly without you reminding anyone.
• 3–5 recurring meetings have been killed or consolidated.
• Three company OKRs are live. Every leader can name them.
• A real cross-functional conflict was surfaced, decided, and logged — and the team is still functional.
• A quarterly planning meeting has happened, with a pre-read, a retro, and a carry-forward.
• You took a Friday off and the company didn’t burn down.

Career signals — green if true, red if not:

• Your CEO has, unprompted, used the phrase “I don’t know what I’d do without this” in the last 30 days.
• A peer (CFO, CRO, CTO) has come to you for advice on how to run their function’s operating cadence.
• You’ve been included in a board prep conversation as a participant, not a note-taker.
• You can articulate, in one paragraph, the three things you’ve structurally changed about how this company operates.
• You haven’t replied to a Slack message after 9pm in the last two weeks.

If most of these are green, you’ve done the job. The role becomes much easier from here. The structure is the asset. You’ll spend year two compounding it.

If most are red, that’s information, not a verdict. Pick the two most red, ask why honestly, and adjust. Most red signals at day 90 aren’t because you did the wrong thing — they’re because you did the right thing in the wrong order, or before the political capital was there. Sequence matters more than ambition. Adjust the sequence.

Some things this manual didn’t tell you, that I want you to hear before you run it.

You will doubt yourself. Especially in weeks 4–7, when the listening tour is over and the system is half-built and nothing visible has happened. That’s the dip. Everyone has it. The structure compounds quietly until week 8 or 9, when it suddenly snaps into visibility. Don’t quit in week 6.

You will get the politics wrong somewhere. A meeting will go sideways. A leader will feel undercut. The CEO will surprise you. When this happens, name it, own it, and let the structure absorb it. Mistakes inside a working structure are recoverable. Mistakes inside a chaotic one are not.

Your CEO will not always understand what you’re doing. That’s normal. They hired you because they couldn’t do it themselves. They cannot evaluate the work in real time — they can only evaluate the outcome. Trust that the outcome will speak. It will.

You are not “just” anything. Not “just a CoS.” Not “just supporting.” You are installing the operating system that lets a company go from 50 people of held-together chaos to 200 people of compounding execution. That is not assistant work. That is the most strategic role in the company. You’ll know this in your bones by month nine.

One last thing.

What makes this hard isn’t the work itself. It’s that the work asks you to do something visible without being able to control whether anyone appreciates it in real time. You’ll build something that looks like nothing for two months and then becomes load-bearing for the company forever. That’s a vulnerable place to operate from. It’s also, statistically, what the best Chiefs of Staff do.

You’re going to be fine.

Better than fine.

You’re going to build something :)

One more thing,…everything in this field manual can be built from scratch with documents, and discipline, as an alternative you have the entire 90-day operating system above as a single multi-session process inside Nova called Build Good Leadership Habits + Listening Tour. It’s the playbook above, pre-built — every template, every cadence, every decision log structure, already wired up and ready to run.

What you get inside Nova(novatools.org):

• The 5-question listening tour, structured as a guided session you can run with each stakeholder
• The decision log, native to every meeting — decisions are logged in the room with owner and date, not after
• The weekly leadership sync template with the three-pillar structure already built in
• The four-layer cadence (daily / weekly / monthly / quarterly) running on autopilot
• The quarterly planning multi-session that carries last quarter’s decisions and lessons forward automatically
• AI participating with a defined role inside the structure — handling pre-reads, summaries, and decision research, while humans review and decide

You can start running the leadership sync inside Nova within 5 minutes. The full 90-day operating system unfolds as a guided process — you don’t have to remember which phase comes when, because the process does.

Free trial. No credit card. The fastest way from “I just took the role” to “the team runs without me in the middle” — pre-built.

→ First 90 days as Chief of Staff playbook: The pre-built operating system that turns a new CoS’s first 90 days from chaos and overwhelm into a phased, structural install — six 15-day phases, every template ready, every cadence wired up.

→ OKRs + Operating Cadence: The structured 4-week sequence to install company-level OKRs that don’t die in week six — from CEO-set Objectives, through honest cascade, to embedded check-ins on your existing cadence.

→ Good Management Rituals: This one is the manager’s operating system for their own pod or department — the rituals and habits that make a team of 5–15 actually function as a team rather than a collection of individuals.

One last thing.

If you read this whole field manual and you’re sitting at your desk right now thinking “I don’t know if I can do this” — that doesn’t mean you’re not the right person for the job.

It means you’re paying attention.

The Chiefs of Staff who go on to do this well are the ones who feel the weight of the role honestly, on day one. The ones who don’t feel the weight are the ones who flame out, because they didn’t realize they were being handed something heavy.

You were handed something heavy.

You’re also being handed the manual to carry it.

Now go.

— Ro

[embed]

If you found this useful, share it with the next CoS you meet. The role is harder than it looks. We’re better when we share what works.