# Ceo Weekly Review

> Weekly CEO review — 10 questions, dashboard, self-deception check, next week focus

- Skill: `codealive-ai/ceo-weekly-review` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add codealive-ai/ceo-weekly-review`
- Raw SKILL.md: https://api.skillmd.com/api/skills/codealive-ai/ceo-weekly-review/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: CodeAlive-AI (https://skillmd.com/u/codealive-ai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/codealive-ai/ceo-weekly-review

---


# CEO Weekly Review

Runs every Sunday evening (cron) or manually on the founder's command.

## Process

### Step 0: System Throughput (Critical Chain — DO THIS FIRST)

**How many validated beliefs changed this week?**
- Hypotheses resolved (keep/kill/reframe): X
- ICP assumptions updated: X
- Decisions on pricing/positioning/message: X
- Target: >= 1 meaningful change/week

**Constraint check:**
- Read `MEMORY.md -> System Constraint`
- CEO time on constraint-work (conversations, discovery, sales) vs non-constraint: X/Y hours
- Target: >35% on constraint
- Has the constraint shifted? If yes -> update MEMORY.md

**Focus discipline (from `memory/focus-state.json` + `memory/focus-history.jsonl`):**
- GTM minimum completed: X/Y days this week
- Waivers: N (list reasons — are they legit or avoidance?)
- Streak: N days
- Most common MVNA type completed (research/outreach/strategy)
- If compliance < 3/5 days → diagnose friction: "GTM X/5 дней. Что мешало? (friction, RSD, energy, priorities)" — don't shame

**WIP check:**
- Active hypotheses: X (limit: 3). Active experiments: X (limit: 2).
- Relay race: are there hypotheses with enough evidence for a decision right now?

### Step 1: Gather Data

1. Read `MEMORY.md` — current wedge, stage, hypotheses, constraint
2. Read `memory/hypotheses.json` — experiment status
3. Read daily logs for this week (`memory/YYYY-MM-DD.md` x 7)
4. If PostHog is available — query key metrics for the week

### Step 2: CEO Weekly Dashboard (Bible Section H.1)

Build the dashboard across 9 categories. PostHog for 2,3,5,6,9. The rest — ask the founder.

**Header:** Current stage | Wedge | System constraint | Biggest risk this week

**1. Learning (ASK the founder):**
- Qualified conversations this week (target: 8-12)
- Repeated pain rate (same problem/trigger recurring?)
- What changed our beliefs?

**2. Product (PostHog):**
- TTFV: median timeSincePrevious for First Value Moment
- Activation: funnel_next_stage_available -> first_value_delivered stage
- Funnel: signup -> activated -> return -> regular -> power (compare to last week)

**3. Usage (PostHog):**
- Retained active: Weekly Active milestone count
- Search/chat weekly active users
- Deep workflow usage: deep_workflow_started

**4. Pipeline (ASK the founder):**
- New pipeline in wedge
- Win rate trend
- Sales cycle

**5. Retention (PostHog + manual):**
- Healthy accounts: regular_usage+ stage
- Expansion pull: team_invitation_sent events

**6. B2C -> B2B (PostHog — KEY):**
- Target signups: Account Created + qualified filters
- Active companies: unique organizationId
- Companies with 2+ users
- Same-company emergence rate

**7. Execution (ASK the founder):**
- Experiment cycle time, belief changes this week

**8. Economics (ASK the founder):**
- Burn/runway

**9. PMF signals (MIX):**
- Wedge repeatability (last wins look alike?)
- Organic pull (sharing events, inbound, referrals)

**Attribution (PostHog):**
- Top sources by $initial_utm_source for activated users

### Step 3: 10 Weekly Questions (Bible Section P)

Go through each question and provide a brief answer based on gathered data:

1. What did we learn this week that actually changed our beliefs?
2. Which segment showed the strongest pain, fastest value, and repeatability?
3. Did I (the founder) personally talk to enough customers, buyers, and lost prospects?
4. Which metric moved, and what decision follows from it?
5. Is the B2C layer creating enterprise signal or just activity?
6. Are recent wins becoming more similar or more bespoke?
7. Is time-to-first-value decreasing?
8. Are healthy accounts asking for more on their own?
9. What are we calling progress because it feels good?
10. If we were starting the company today, would we choose the same wedge?

### Step 4: Self-Deception Check (Bible Section P)

Check 10 self-deception signals:

1. Only warm intros convert?
2. Traffic/signups grow, but target-cohort activation doesn't?
3. Calling pilot revenue "ARR"?
4. Every early customer wants a different product?
5. Buyers love demos but don't use the product?
6. Annual contracts look OK, but engagement is dead?
7. Hiring because learning feels slow?
8. Founder can't explain the last 5 deals?
9. B2C grows, but doesn't create accounts or pipeline?
10. Team is very busy, but beliefs and priorities don't change?

If 3+ signals fire — flag hard.

### Step 5: Adversarial Self-Review

Before final recommendations — challenge your own analysis:

1. **Consistency:** do my dashboard conclusions contradict my answers to the 10 questions?
2. **Proxy skepticism:** am I assessing real progress toward PMF or convenient metrics?
3. **Constraint alignment:** do my next-week recommendations exploit the constraint (market learning speed)?
4. **Inversion:** what would have to happen next week for [YOUR_PRODUCT] to get WORSE? Do my bets protect against that?

If issues are found — fix recommendations BEFORE sending.

### Step 6: Next Week

Formulate:
- **Top 3 bets** for next week (tied to constraint)
- **Top 3 kills** (what to stop — Jobs's subtraction)
- **Top 3 asks** (what is needed from the founder or externally)

For each bet: reversibility (two-way / one-way door) and kill signal.

### Step 7: Save

Log the result to `memory/weekly/YYYY-WW.md`.
Update `MEMORY.md` if any of these changed: stage, wedge, key learnings.

## Chat Output Format

Don't send everything in one message. Split into 3-4 short ones:
1. Dashboard summary (5-7 lines)
2. Key answers + self-deception flags (5-7 lines)
3. Next week focus (3-5 lines)

Follow output preferences from USER.md (language, format, platform constraints).

## Tone

Like an honest Chief of Staff: facts -> implications -> actions. Don't sugar-coat.