Month three of a new PM job is where most onboarding advice goes vague. "Start contributing more." "Take on bigger things." That is not a plan, that is a horoscope.
Month three has exactly two moves, and both put your name on something. One falsifiable bet, and one public subtraction. The bet shows the org what you will build. The kill shows the org what you will protect them from. A PM who only ever adds is a backlog with a pulse. This is the final deep dive on The PM 30/60/90, and it assumes month one built your evidence base and month two shipped your artifact.
The short version
PM days 61 to 90 convert standing into ownership through two moves. First, own one bet: a single outcome sized to a quarter, backed by evidence from your own archaeology and signal map, with a falsifiable measure and a date, written as a one-pager and pre-wired through your signal map relationships before any meeting. Second, kill or renegotiate one inherited commitment, the safest one, with an evidence memo and a replace-not-remove script. Close the quarter with a day-90 note to your manager: found, shipped, betting, killed, measure-me-on. The method for the kill lives in The Anti-Backlog, and the templates are in the PM First-90 Kit.
Choosing the bet
Not the whole roadmap. One bet. Four filters, applied in order:
Sized to a quarter. Your first bet has to resolve while people still remember you made it. A year-long bet is indistinguishable from no bet at all, because by the time it resolves the org has reorganized twice. One quarter: long enough to matter, short enough to be accountable.
Evidence from your own digging. The bet must rest on your month-one archaeology and signal map, not on a deck someone handed you. This is not pride, it is mechanics. When the bet gets pushback, and it will, you need to defend it from primary evidence you personally collected. "The last six churn surveys all mention this" survives a hard meeting. "The strategy doc said so" does not.
Falsifiable. A measure that can prove you wrong. "Improve the onboarding experience" is not falsifiable. "Week-one activation for the self-serve segment moves from A to B" is. If no result could make you abandon the bet, you do not have a bet, you have a preference with a slide.
A date. Written down, said out loud, in the one-pager. The date is what makes it yours.
The common mistake is betting on a feature instead of an outcome. "Ship the bulk-export flow by March" is a project, and projects can succeed while the product fails. "Reduce export-related support tickets by half by March, and bulk-export is my current best lever" is a bet. The feature can change if the evidence changes. The outcome cannot.
The bet one-pager
Five sections, one page, no appendix:
- The claim. One sentence. Outcome, measure, date.
- The evidence. Three to five items, each one line with a source: a recurring theme from your call synthesis with a verbatim quote, a number from the analytics, a pattern from the decision timeline. Sourced evidence only. This is where month one pays its rent.
- The measure. Current value, target value, where it is tracked. If the measure does not exist yet, instrumenting it is week one of the bet.
- The date. A date, not a quarter. "End of Q3" is a shrug. "September 12" is a commitment.
- What would change my mind. Two or three results that would kill the bet. "If the prototype tests flat with the target segment, I am wrong about the lever." This section does the most work in the room. It tells everyone you are running an experiment, not defending an ego, and it makes disagreeing with you safe, which means people actually tell you things.
Pre-wiring through the signal map
The worst way to socialize your first bet is to unveil it in a meeting. Meetings are where pre-wired decisions get confirmed, not where new ideas get fair hearings. I have watched good bets die in rooms because the PM treated the meeting as the first conversation instead of the last one.
The sequence, roughly ten days before the decision meeting:
First, the engineer who knows everything, the one you found in month one. Not for permission, for the landmine check: "what am I missing, what has been tried, what breaks?" Their corrections go into the evidence section, with credit.
Second, the people the bet affects, one at a time, fifteen minutes each: "I am about to propose this. Here is the evidence. What would make this wrong?" You are collecting objections while they are cheap and private. Every objection you hear in a hallway is one that does not ambush you in the room, and half of them improve the bet.
Third, your manager, with the near-final one-pager, framed against the week-one contract: "you said good at day 90 looks like X. This is my bet on X." No manager should ever see your bet for the first time in a group setting. None. Ever.
On framing the claim itself: Wes Kao's "spiky point of view" idea applies directly. A defensible bet is specific enough that a reasonable person could disagree with it. "We should improve activation" is mush nobody can oppose and nobody will fund. "Activation is gated on the empty-state experience, not the signup flow, and here is why" is spiky: falsifiable, evidence-backed, and worth a real argument. Your first bet should be spiky or it is not a bet.
By the time the meeting happens, the room has already read the one-pager, the objections are in it, and the meeting is twenty minutes of confirmation. That is what a decision meeting is supposed to be.
Killing one inherited commitment
Every PM inherits at least one commitment the evidence no longer supports: a feature promised to a customer who has since churned, a weekly report nobody reads, a ceremony that outlived its purpose. Killing one, calmly and openly, teaches the org how you operate and makes every future no cheaper. The full method is The Anti-Backlog; here is the month-three version.
Pick the safest one. Three criteria: low political heat (the original sponsor has moved on, or privately agrees), weak evidence (the original justification visibly no longer holds), and a clear beneficiary (someone gets time or focus back when it dies). Your first kill is a demonstration of process, not a war. The contested kills come later, after this one establishes that you kill with receipts, not opinions.
Write the evidence memo. Half a page: what was committed and when (from your archaeology), what the evidence was then, what the evidence is now, and what the commitment costs to keep, in hours, focus, or roadmap slots. No adjectives. The memo should read like an autopsy, not an argument.
Renegotiate before announcing. In person, with whoever owns the expectation, before anything is public. The script, close to verbatim: "We committed to X about six months ago. Since then, three things changed, and they are in this memo. I want to propose we retire it, and here is what you get instead: the two days a sprint it costs go to Y, which the evidence says you need more." Replace, never just remove. People do not mourn commitments, they mourn the thing they thought the commitment would get them. Give them a better path to that thing and most kills take one meeting.
If they push back hard, renegotiate scope instead of forcing the kill: "monthly instead of weekly," "the report becomes an automated page." A partial kill that lands beats a full kill that starts a feud you cannot afford in month three.
The day-90 note
One page to your manager, closing the contract you wrote in week one. Five lines of structure:
- Found: the two or three biggest gaps between the official story and the evidence, stated neutrally.
- Shipped: the artifact, with one line on who is using it now.
- Betting: the bet, the measure, the date.
- Killed: the commitment you retired and what the freed capacity now funds.
- Measure me on: what you want to be evaluated against for the next year. Write it yourself. If you do not, someone else will, retroactively, in a calibration meeting you are not in.
Send it as a document a day before your one-on-one, then discuss it live. Quietly, this note is the first draft of your next promotion case, and the discipline of writing one per quarter afterward, the way the weekly review system works at a smaller cadence, compounds the same way the artifact did.
What changes after day 90
The window closes. You have absorbed the org's story now, you reference history in meetings, you have a stake in past decisions. The fresh read does not come back.
But the practices outlive the window. The decision ledger keeps growing. The signal map keeps routing into your digest. The artifact keeps compounding. The bet gives you a scoreboard, and the quarterly note keeps the scoreboard honest. The 90-day plan ends. The operating system it installed does not.
Failure modes
Betting on a feature instead of an outcome. The most common failure, so it earns repeating. If your bet can succeed while the metric stays flat, you bet on output. Restate it until the metric is the bet.
The kill that picks a fight you cannot win yet. The tempting kill is the big one, the zombie project everyone privately mocks. It is also the one with a sponsor who has skin in it and tenure you do not. Month three is too early for that fight. Kill the safe one, bank the precedent, and come back for the zombie in month six with a track record.
The day-90 note as humble-brag. "Thrilled to share what an incredible journey these 90 days have been." Delete it. The note is an operating document, not a LinkedIn post. Findings stated neutrally, the bet stated with its kill criteria, the ask stated plainly. Your manager reads a hundred self-promotional updates a year. The PM who sends a scoreboard instead is the one they remember in calibration.
Pick one thing this week
If you are in month three: draft the bet one-pager today, all five sections, especially "what would change my mind." Then book the first pre-wiring conversation, the engineer, for this week.
If the bet is socialized: pick the kill. Safest one, evidence memo this week, renegotiation next week.
If day 90 has passed and none of this happened: the window for fresh eyes is gone, but a bet one-pager and one clean kill work at any tenure. Start with the bet, and pull the evidence from whatever your version of the signal map is. If you never built one, the first 30 days post works just as well at day 400.
Sources: Michael Watkins, The First 90 Days on securing early wins and alliances, Wes Kao on spiky points of view for the bet framing, Marty Cagan, Empowered on outcome ownership over feature ownership.
Further reading
- The PM 30/60/90, the full arc this series expands
- The Anti-Backlog, the full kill method
- The PM First-90 Kit, the one-pager and day-90 note templates
Frequently asked
What should a PM do in days 61 to 90 of a new job?+
Two moves: put your name on one falsifiable bet (one outcome, a measure, a date, sized to a quarter), and kill or renegotiate one inherited commitment the evidence no longer supports. Then close the quarter with a one-page day-90 note to your manager. The bet converts standing into ownership; the kill teaches the org how you operate.
How do I choose my first bet as a new PM?+
Four filters: it is sized to a quarter, not a year; the evidence comes from your own month-one archaeology and signal map, not someone else's deck; the success measure is falsifiable, a number that can prove you wrong; and it has a date. If you cannot describe the result that would make you abandon the bet, it is not a bet, it is a preference.
What goes in a bet one-pager?+
Five sections: the claim (one sentence, outcome plus measure plus date), the evidence (three to five items from your archaeology and signal map with quotes and data), the measure (current value, target value, where it is tracked), the date, and what would change my mind (the two or three results that would kill the bet). The last section is what separates a bet from a pitch.
How should a new PM kill an inherited commitment?+
Pick the safest one first: low political heat, weak evidence, an owner who has moved on or quietly agrees. Write a short evidence memo, then renegotiate in person before any public announcement. The script is replace, not just remove: here is what we promised, here is what changed, here is what you get instead. One calm kill with receipts makes every future no cheaper.
What is in the day-90 note to your manager?+
One page, five lines of structure: what I found, what I shipped, the bet I own, what I killed or renegotiated, and what to measure me on for the next year. It closes the manager contract from week one and quietly drafts your next performance review on your own terms. Send it as a document, then discuss it live.
What changes after day 90 in a new PM job?+
The fresh-eyes window closes. You have absorbed the org's story and can no longer see the product without it. But the practices survive the window: the decision ledger keeps growing, the signal map keeps routing, the artifact keeps compounding, and the bet gives you a scoreboard. The 90-day plan ends; the operating system it installed does not.
Comments (0)
Sign in with LinkedIn to leave a comment.
Sign in with LinkedIn