nerissa-self-audit

# nerissa-self-audit · 9-Lens Startup Self-Audit A field-tested framework for auditing your own startup honestly. Built from real cases where founders (and AI) repeatedly fell into the trap of analyzing competitors while never applying the same rigor to themselves. Every lens corresponds to a documented failure pattern this skill prevents. ## When to Trigger **✅ Trigger when**: - User explicitly says: "audit our current state", "self-audit", "我们到底是什么", "诊断我们自己", "review our positioning", "where are we really" - You detect a gap between user's stated vision and operational reality (e.g. landing copy says one thing, customer pipeline says another) - Before any major strategic decision (pivot, fundraise, large hire, vertical lock-in) - When the user says "we should do X" but you suspect the underlying problem hasn't been diagnosed **❌ Do NOT trigger for**: - Single-feature decisions (use decision-lens for those) - Competitor research (use a competitive-research skill for those) - Casual brainstorming without commitment to act on findings - Execution tasks (writing code, drafting emails, filing tickets) ## Core Principle > **Apply to yourself the same scrutiny you apply to competitors.** Most teams happily run rigorous SWOT / Porter / JTBD on competitors but never on themselves. This skill closes that gap. The 9 lenses are tools you probably already know individually — the discipline is using them together, on yourself, with the same honesty you'd use to analyze a competitor you're skeptical of. ## The 9 Lenses ### Lens 1 · First Principles (foundational) > *Aristotle · popularized by Elon Musk* **The question**: Strip away every assumption, every analogy, every inherited judgment. What are the irreducible atomic facts about your startup right now? **Apply by**: 1. List every "obvious" assumption your team operates on (e.g. "our customers are solo founders", "we need to raise $10M") 2. Question each: is this really true? What's the evidence? 3. Extract 5-8 atomic facts that can't be further decomposed 4. Re-derive your real identity from those facts (not from marketing copy) **Catches**: vision-vs-reality misalignment, inherited assumptions, identity confusion. ### Lens 2 · Jobs-to-be-Done (JTBD) > *Clayton Christensen · Harvard Business School* **The question**: What job are your customers actually hiring you to do? What did they hire before you existed? **Apply by**: 1. For each existing customer, ask: what concrete outcome did they hire you for? 2. Identify their previous "solution" (manual work / a competitor / doing nothing) 3. Your real competitor = whatever they hired before, not the company you think is your competitor **Catches**: wrong competitor framing, feature-vs-outcome confusion. ### Lens 3 · 5 Whys > *Taiichi Ohno · Toyota Production System* **The question**: For any visible symptom (sales stuck, churn high, no signups), ask "why" 5 times to find the root cause. **Apply by**: 1. State the symptom 2. Ask Q1 "why" 3. Take the answer, ask Q2 "why" 4. Continue 5 times 5. The 5th answer is usually the real problem you need to solve **Catches**: treating symptoms as causes, premature solutioning. ### Lens 4 · OKR (Objectives & Key Results) > *Andy Grove (Intel) · popularized by John Doerr / Google* **The question**: What single ambitious objective should the next quarter prove or disprove? What 3-5 measurable results would tell you it worked? **Apply by**: 1. Write 1 ambitious **Objective** (qualitative · directional) 2. Write 3-5 **Key Results** (quantitative · verifiable) 3. KR0 should always be the most foundational ("sign 1 customer", "get 1 piece of real data") — not the most aspirational **Catches**: vague strategy, unmeasurable progress, premature optimization. ### Lens 5 · SWOT > *Albert Humphrey · Stanford Research Institute · 1960s* **The question**: What internal Strengths and Weaknesses do we have? What external Opportunities and Threats? **Apply by**: 1. List **Strengths** (internal advantages · including unconventional ones) 2. List **Weaknesses** (internal gaps · including embarrassing ones) 3. List **Opportunities** (external trends · only credible ones, not wishful) 4. List **Threats** (external risks · including ones you've been avoiding facing) 5. Critically: which Weakness blocks which Opportunity? That's your top problem. **Catches**: blind spots, opportunity-blocker mismatches. ### Lens 6 · Porter Five Forces > *Michael Porter · Harvard Business School · 1979* **The question**: What 5 competitive forces shape our industry profitability — same-industry rivalry / potential entrants / substitutes / buyer power / supplier power? **Apply by**: 1. Same-industry rivalry: high/medium/low? 2. Potential entrants (giants, capital-rich teams): how blocked are they? 3. Substitutes (incumbent processes, manual work): what stops customers from staying with status quo? 4. Buyer power (can customers squeeze you on price?): high/medium/low? 5. Supplier power (do you depend on a critical input — API / talent / data?): high/medium/low? **Catches**: missing the real competitive threat, underestimating substitutes. ### Lens 7 · BCG Growth-Share Matrix > *Bruce Henderson · Boston Consulting Group · 1970* **The question**: For each of our existing product features / business lines, which are Stars / Cash Cows / Question Marks / Dogs? **Apply by**: 1. List all current features / business lines 2. Score each on (a) growth potential and (b) current market share / traction 3. Place in 2x2: - **Stars** (high growth + high share): protect and invest - **Cash Cows** (low growth + high share): milk for resources - **Question Marks** (high growth + low share): invest selectively - **Dogs** (low growth + low share): kill 4. Identify what % of your team's time/resources is on Dogs **Catches**: resource misallocation, feature creep, sunk-cost decisions. ### Lens 8 · Blue Ocean ERRC Grid > *W. Chan Kim + Renée Mauborgne · INSEAD · 2005* **The question**: To create a new uncontested market space, what should you Eliminate / Reduce / Raise / Create? **Apply by**: 1. **Eliminate**: which industry-standard features are you currently doing that you should kill? 2. **Reduce**: which features should be present but at much lower investment than industry standard? 3. **Raise**: which features should be present at much higher investment than industry standard? 4. **Create**: what should you add that no one else in the industry offers? **Catches**: feature parity trap, "we must have what they have" anchoring. ### Lens 9 · Inversion > *Carl Jacobi (mathematician) · popularized by Charlie Munger* **The question**: Don't ask "how do we succeed?" Ask "how would we certainly fail?" — then avoid each failure mode. **Apply by**: 1. Imagine your startup definitely dies in 12 months. What killed it? 2. List 5-10 specific death scenarios (not vague "we ran out of money") 3. For each scenario, ask: are we doing the thing that causes it? If yes, stop immediately. 4. The remaining list is your real risk register. **Catches**: optimism bias, hidden self-destructive behaviors, ignored warning signs. ## The Synthesis Step (Critical) After running all 9 lenses **independently**, force-rank: > "Which 1-2 conclusions did ALL OR MOST lenses point to?" If 3+ lenses point to the same conclusion, that's not a coincidence — it's the real diagnosis. Don't pick conclusions you like; pick conclusions that converge across multiple lenses. ## Output Format When invoked, the skill produces: 1. **9 lens applications** (one short section per lens) 2. **Convergence map** (which conclusions appeared across multiple lenses) 3. **Top 1-3 real diagnoses** (the convergent conclusions) 4. **5 real questions only you can answer** (decisions blocking progress) 5. **Anti-flattery flags** (places where you/Claude might be making yourself look better than reality) ## Required Discipline | Rule | What it prevents | |---|---| | Run all 9 lenses, not just the comfortable ones | Cherry-picking framework to confirm pre-existing belief | | Each lens is run **independently** before synthesis | Letting your favorite conclusion bias each lens | | Convergence across lenses = signal; single-lens insights = lower confidence | Treating one striking framework output as the whole truth | | For each lens, cite evidence (data / quote / code path) | Using lens as decoration over vibes | | Apply same rigor as you would to a competitor's audit | Self-favorable analysis | | First-person voice: "we are" not "they are" | Maintaining ownership | ## When to Re-Audit - After any major external shock (competitor funding, customer churn, regulatory change) - After any major internal change (pivot, key hire, product launch) - Quarterly · scheduled · whether or not anything visible changed - When the team starts using vague defensive language ("things are tough", "it's complicated") ## How to Use This Skill When invoked, Claude will: 1. Confirm what specific aspect of the startup to audit (full company / a product line / a strategic decision) 2. Walk through each of the 9 lenses, citing evidence specific to your situation 3. Produce the convergence map showing which conclusions appeared across multiple lenses 4. Surface the top 1-3 real diagnoses 5. List 5 real questions only you can answer 6. Flag any anti-flattery violations encountered ## Examples Inside This Skill See `examples/` for 5 worked examples (fully anonymized): - `01-vision-vs-reality-misalignment.md` — When your landing page says "for solo founders" but every signed customer is a mid-market B2B firm - `02-resource-misallocation-via-BCG.md` — When 80% of team time is spent on features that score as "Dogs" - `03-wrong-competitor-via-JTBD.md` — When you think you compete with VC-backed AI startups, but your customers are actually choosing you over "doing nothing" - `04-5-whys-finds-the-real-blocker.md` — Why we're stuck on no signed customers · 5 layers down, the root is unmade pricing decisions - `05-inversion-catches-the-quiet-death.md` — Listing the 5 ways the startup would certainly die · catching the one already in progress ## Linked Resources - `references/9-lens-worksheet.md` — Print-and-fill template for a 2-hour audit session - `references/convergence-map-template.md` — How to visualize where lenses agree - `references/anti-flattery-checklist.md` — Before publishing audit conclusions, run this ## Relationship to Other Skills - **`decision-lens`** — Use when you need to pressure-test ONE specific decision through multiple expert roles. This skill (self-audit) is for the WHOLE STARTUP STATE. - **`nerissa-competitive-research`** — Use to research OTHER companies with the same rigor. This skill turns that rigor inward. The three skills form a triangle: - decision-lens: one decision, multiple roles - nerissa-competitive-research: one company (external), one team - nerissa-self-audit: your whole company (internal), 9 lenses ## A Note on Provenance This framework was extracted from a real early-stage startup that repeatedly found itself analyzing competitors with rigor while never applying the same lenses to itself. After the third "vision vs reality" surprise, the team formalized the 9-lens self-audit. Every lens was empirically chosen because it caught at least one mistake the team had made.