The 90-Day Sprint: the goal system for people whose annual goals are already dead

Annual goals fail for a mechanical reason, not a moral one: the deadline is too far away to generate any feedback. Miss a week in March and the ledger doesn’t flinch — December will absorb it, you tell yourself, and December disagrees. Daily to-do lists fail in the opposite direction: endless feedback, zero direction. You can clear your list for 200 straight days and build nothing.

The quarter is the bridge. Thirteen weeks — roughly 65 working days — is the largest block of time you can see the end of from the start, and the smallest that fits something real: a product shipped, an offer sold, a body changed, a skill installed. The system below is everything we know about running one well. It has five parts, and every part has a clock on it. Start Monday. Any Monday.

Step 1: Pick one outcome you will ship

Not a theme. Not an intention. An outcome — something a stranger could verify happened. “Get better at marketing” fails the stranger test. “Publish 12 newsletter issues and reach 300 subscribers” passes. “Get in shape” fails. “Run a 5K without stopping” passes.

One outcome. Two if they don’t compete for the same hours (one work, one health survives; two work goals rarely do). Three is zero — you’ll rotate between them just fast enough to keep each one warm and none of them alive.

Step 2: Write the one-page charter

One page, six fields, written before the sprint starts. If a field takes you longer than ten minutes, that’s the field telling you something.

1. Outcome. One sentence. Stranger-verifiable.
2. Why now. One sentence. If you can’t finish it, this isn’t the quarter for this goal.
3. Definition of done. The demo: what exactly exists on the last Friday that doesn’t exist today?
4. Lead measures. The two weekly inputs you’ll score (Step 3).
5. Not doing. Three named things you are explicitly benching for 13 weeks. This field does more work than any other.
6. Kill criteria. What evidence, by week 7, would mean this deserves to die?

Here’s a filled one, from the world of our productized-service playbook: Outcome: sell a fixed-scope website audit at $1,500. Why now: two clients asked for it unprompted last month. Done: four paying customers and a repeatable delivery checklist. Lead measures: 20 outreach touches a week; one audit delivered a week, discounted if necessary. Not doing: redesigning my own site, starting a podcast, custom proposals. Kill criteria: fewer than two buyers by week 7 despite 100+ touches.

Step 3: Choose lead measures, not lag measures

You cannot will an outcome into existence in any given week. You can only control inputs. The test: if you can’t hit the number purely by deciding to, it’s a lag measure and it doesn’t belong on your weekly scoreboard. “Three sales” is lag — the market votes on it. “Twenty-five outreach messages” is lead — only you vote on it.

Two lead measures, maximum. For most goals the right pair is one production number (things made: workouts done, issues written, audits delivered) and one distribution number (things pushed into the world: pitches, posts, applications, invites). If both go green for weeks and the outcome doesn’t move, that’s not failure — that’s the system generating the exact information the week-7 gate needs.

Step 4: The Friday scoreboard — 15 minutes, non-negotiable

Same time every Friday. It has four moves and no ceremony:

• Score each lead measure: hit or miss. No partial credit — partial credit is how fiction gets into the books.
• Mark the week green (both hit), yellow (one), or red (none).
• Write one sentence: what made it green or red. One. The discipline of one sentence beats a journal you’ll abandon.
• Set next week’s two numbers.

Two reds in a row means the plan is fiction. Don’t summon willpower — shrink the numbers until they’re boring, because a boring number you hit every week beats a heroic one you hit twice. The scoreboard slots neatly into the front of a 30-minute weekly review if you already run one.

Step 5: The week-7 gate

Put it in the calendar the day you write the charter: one hour, midpoint of the sprint. You are allowed exactly three verdicts, and you write the sentence for whichever you choose:

• Persevere. “Lead measures are mostly green and the outcome is moving. Change nothing.” The boring verdict is correct more often than feels satisfying.
• Pivot. “Inputs are green, outcome is flat — my method is wrong, not my effort.” Change one variable: the price, the channel, the audience, the format. Never all of them — a total reset at week 7 is just a new sprint wearing the old one’s number.
• Kill. “Kill criteria hit. Stopping.” Write a five-line post-mortem and bank the lesson. Killing on evidence at week 7 is a win; drifting to week 13 on hope is the actual loss. If the sprint died because you fell off rather than the idea failing, that’s a different problem with a different fix — the 48-hour reset protocol.

Step 6: Ship week, then the retro

Weeks 12 and 13 have one job: stop improving, start finishing. The rule that protects them is the ratchet rule — scope may shrink, the date may not move. Cut features, cut polish, cut the nice-to-haves; deliver the demo on the last Friday to a real audience, even an audience of one.

Then a 30-minute written retro, four questions: What did I plan? What actually happened? What surprised me? What do I know now that the January version of me didn’t? Keep every retro in one document. After three sprints that document reads like a user manual for your own operating tendencies — and it will tell you things no productivity book can.

The six failure modes

• Three goals. The most common death. Three goals is zero goals wearing a trench coat.
• Lag measures in lead clothing. “Gain 500 followers a week” is not an input. If the market has a vote, it’s not a lead measure.
• The week-3 skip. The first missed scoreboard is the sprint’s highest-risk moment — the streak is the system. If Friday explodes, do a five-minute version. Never zero.
• Moving the definition of done. Mid-sprint “upgrades” are procrastination with better branding. Ratchet rule: scope shrinks, date holds.
• Sprinting on an unchanged calendar. A sprint adds hours of work to your week; if nothing was removed, the sprint is the thing that gets removed. Clear the space first — that’s an attention-architecture problem, and it’s solvable.
• Back-to-back sprints. Week 13 into week 1 with no gap means sprint two inherits sprint one’s fatigue and none of its lessons. Take a one-week cooldown: retro, rest, write the next charter slowly.

That’s the whole machine. One page to start it, 15 minutes a week to run it, one honest hour in the middle, one shipped thing at the end. Thirteen weeks from now either the thing exists, or you know exactly why it shouldn’t — and both of those beat where the annual plan was going.