How-to: Pick an Early Win in Your First 90 Days
An early win is a credible improvement you can finish soon enough that people trust your judgment, not a theatrical rebrand of someone else’s work. The point is learning the system while delivering a bounded outcome: a cleaner handoff, a decision unblocked, a broken report fixed, a customer pain reduced. Pick wrong and you either vanish into busywork or pick a fight with the org chart.
Time box: 90 minutes to choose (spread over a few days of listening), then a separate execution plan. This how-to is the picker, not the entire project.
Place it in the New hire hub and first 90 days arc after you have intros and a week of context. ICs: Individual Contributor hub. New managers: prefer team-level enablers over solo heroics (New Manager hub).
When to use / when not
Use when:
– You have enough context to see repeated friction (usually after 1–3 weeks of listening)
– Your manager wants signal that you can ship, not only absorb
– There is a painful, local problem with a clear “done”
– Stakeholders will notice the improvement without a press release
Do not use when:
– You just arrived yesterday and are still unlocking email
– The only “wins” on offer are political takedowns
– The work requires authority you do not have yet (org redesign, comp changes)
– You are being nudged into a fake win for optics while ignoring a real safety or ethics issue
If HR, ethics, or customer harm is in play, escalate properly via how HR works rather than hunting a trophy.
Prep
- Review notes from listening chats, standups, and your manager’s week-one priorities
- List friction points with rough effort and who benefits
- Know constraints: tools you can touch, approval paths, quiet periods (freezes, audits)
- Ask who owns adjacent surfaces so you do not “win” by creating downstream mess (how departments work together)
- For revenue-tinged ideas, sanity-check against how go-to-market actually runs (how sales works)
Bring a draft shortlist to your manager before you announce anything widely.
Steps
- Collect candidate problems (20 min). Aim for 8–12 raw items from your notes. Phrase each as outcome, not task (“reduce double data entry between X and Y,” not “build a spreadsheet”).
- Score with four filters (25 min). Impact visible to others; feasibility in 2–4 weeks; learning value about how the org works; low blast radius if imperfect.
- Kill vanity and landmines (15 min). Drop items that are purely cosmetic, depend on absent executives, or reopen healed scars without mandate.
- Pick one primary and one backup (10 min). Primary should survive a vacation day and a single dependency slip.
- Define done in one sentence (10 min). Include owner (you), reviewer, and date.
- Socialize with manager and one peer (10 min). Adjust scope. Confirm you are not stealing someone else’s active project.
- Only then schedule the work. Put milestones on the calendar. Keep your day-one curiosity habits: keep listening while you ship.
Leaders with broader scope should still pick a bounded win first; giant strategy resets belong later (Director / VP hub).
Script (what to say / send)
Manager alignment:
“I shortlisted early-win options after listening. Primary: [outcome] by [date]. Why: [impact]. Feasible because [access/skills]. Risks: [one]. Backup: [outcome]. Does this match what would build trust here?”
Peer check:
“I am considering fixing [pain]. Have you or anyone tried recently? Any landmines I should know before I propose it?”
Announcement after approval (keep modest):
“I am taking [scoped outcome] this [sprint/month]. Done means [sentence]. I will share a short before/after note when it ships. Ping me if this collides with work you already own.”
If pressured into a fake win:
“I can do a quick cosmetic pass, but it will not change outcomes. Option B is [real fix] on a slightly longer clock. Which do you want people to feel?”
If the win touches budget or tooling cost:
“This needs a small spend / vendor change. I will bring a thin ask with cost and benefit rather than sneaking it through.” Use framing from finance for non-finance managers when you write that ask.
Output artifact
Early-win one-pager:
| Field | Content |
|---|---|
| Outcome | One sentence |
| Why now | Trust + pain |
| Done means | Testable |
| Timebox | Start / end |
| Dependencies | Names |
| Non-goals | Explicit |
| Backup win | One line |
| Share-out plan | Who / where |
Also keep the scored shortlist. It becomes backlog fuel for days 60–90, not a museum of rejected ideas.
Watch-outs
- Choosing something nobody feels. If only you notice, it is a learning project, not a trust-building win.
- Boiling the ocean (“fix onboarding,” “rebuild culture”). Slice until a two-to-four-week path appears.
- Stealing credit. If you finish a nearly done effort, name the prior work.
- Optimizing your old company’s playbook onto a new context without listening.
- Ignoring second-order load on support, sales, or finance partners.
- Announcing before aligning. Quiet alignment prevents public walk-backs.
Done looks like: one scoped outcome agreed with your manager, non-goals listed, a backup ready, and a calendar path to a real finish line. Early wins are proof of judgment. Pick a problem worthy of your attention and sized for your runway.