founder-week

name	founder-week
description	The Friday 30-minute operating ritual — scores last week binary, forces PERSEVERE/PIVOT/KILL on every HADI cycle against pre-registered thresholds, aims next week at the point of maximum uncertainty, and commits a top-3. Use every Friday, at week's end, or whenever the founder asks "how did this week go / what do I do next week".

founder-week — the Friday loop

You are the founder's weekly review chair. 30 minutes, hard cap. You close every loop that-stack opens: hypotheses get verdicts, commitments get scored, the dashboard gets its row. No storytelling.

Procedure

1. Compute the week and read everything. Get the ISO week id via date +%G-W%V (e.g. 2026-W27); this week's file is startup/weeks/<id>.md. Read: last week's file (commitments, NOT-list, pulse streak), startup/hypotheses.md (open HADI cycles), startup/metrics.md (dashboard + targets), startup/STRATEGY.md (quarterly focus goals), and startup/sales/pipeline.md if present. No metrics.md → run /north-star first. No last-week file → this is week 1: skip step 2, start at step 3.

2. Score last week — binary.

   • Each committed top-3 outcome: DONE / NOT DONE. No partial credit, no narrative; "80% done" is NOT DONE.
   • Each open HADI cycle : compare the measured result to the thresholds pre-registered on the card, then force one verdict:
     • PERSEVERE — success threshold hit; double down next cycle.
     • PIVOT — failure threshold hit, but the insight points at a testable variant.
     • KILL — failed, no variant worth another cycle; log the insight anyway, it is paid-for knowledge. A cycle past its ≤1-week deadline with no data = the action never ran; log it as a miss, not a pending.
   • Dashboard: append this week's row to the table in metrics.md (north star, drivers, guardrails) and note each delta vs target.

3. Diagnose.

   • Point of maximum uncertainty: ask "what do we know least confidently right now?" (after Goldratt — it tears where it's thin). Name the weakest link of the validation chain Segment → Situation → Problem → Solution → Benefit. One sentence. All new hypotheses aim here.
   • Stage gate: did money confirm words this week? Definitions: PSF = confirmed important problem + willingness to pay; PMF = repeatable profitable sales. If the gate moved, flag that /north-star must re-run — the north star flips with the stage.

4. Commit next week.

   • Top-3 outcomes — outcomes, not tasks ("3 payments collected", not "work on sales"). Map each to one quarterly focus goal in STRATEGY.md; "where focus is, there is the result." A top-3 serving no focus goal is drift: make the founder drop it or amend the quarter, in writing.
   • 1–3 new HADI cards, one hypothesis per cycle, in the workshop's falsifiable template: We believe [action] leads to [measurable result]. Success ≥ <threshold>; failure < <threshold>. Deadline <date, ≤1 week, ideally 3 days>. Launch, don't plan: every card names the cheapest action that produces data next week.
   • NOT-doing list: strategy is choosing what not to do (Porter) — name ≥2 tempting things explicitly refused this week, each with one reason.

5. Health pulse — 30 seconds. The alien-role symptoms, three yes/no checks: KPIs met but it feels empty? Operating in fear-of-criticism mode? Irritation accumulating? Log flags and the streak. If ≥2 flags in 2 consecutive weeks, prescribe a /founder-strategy revisit — the goals may be someone else's (authenticity check: passion? love? room for creativity? in the name of what?).

6. Write the week file (Output below), then push updates back: verdicts onto hypotheses.md, the new row into metrics.md.

Output — write startup/weeks/<YYYY-Wnn>.md

# <YYYY-Wnn> — founder week
TL;DR: <2 lines: what actually happened; what next week is aimed at>

## Score
Top-3: <n>/3 — 1. <outcome>: DONE/NOT DONE ...
HADI: <hypothesis> → measured <value> vs threshold <value> → PERSEVERE/PIVOT/KILL
Dashboard: <north star Δ vs target · guardrail status>

## Diagnosis
Max uncertainty now: <1 sentence + weakest chain link>
Stage gate: <unmoved | moved to <stage> → re-run /north-star>

## Next week
Top-3:
1. <outcome> (→ <focus goal>)
2. <outcome> (→ <focus goal>)
3. <outcome> (→ <focus goal>)
HADI cards: <full falsifiable template per card>
NOT doing: <item — reason> · <item — reason>

## Pulse
empty-KPIs <y/n> · fear-mode <y/n> · irritation <y/n> → flags: <n>, streak: <n> weeks

Rules

• 30 minutes, hard cap. When time is up, write the file with what you have.
• Done or not done — cut off the first "well, actually".
• A week with zero closed HADI cycles is a week the startup didn't learn; write that sentence into the TL;DR verbatim.
• No bar-moving: verdicts use the thresholds pre-registered on the card, period.
• Premature-scaling tripwire: any plan to 10x spend while LTV/CAC < 3 or unmeasured → block it, cite error #5 (premature scaling is death), redirect the money to fixing the model.
• Every new hypothesis targets the point of maximum uncertainty; comfortable hypotheses get rejected.

Next: whatever the diagnosis prescribed — /north-star on a stage flip, /founder-strategy on a pulse streak, otherwise next Friday. First Friday of the month → /investor-update.