code-ai-installer 3.0.0 → 4.0.1-a

package/domains/product/locales/en/.agents/skills/north-star-metric/SKILL.md

@@ -1,301 +1,317 @@
----
-name: north-star-metric
-description: Selecting the NSM for B2B SaaS — balancing leading/lagging, alignment with customer value and business outcome
----
-# North Star Metric (NSM)
-
-> **Category:** Strategy  ·  **Slug:** `north-star-metric`
-
-## When to Use
-
-- When defining product strategy — a single metric of focus.
-- When aligning the team: decision-making through the lens of "does this lift the NSM?".
-- For board / investor reporting — single metric story.
-- When reviewing an existing metric — still meaningful?
-
-## Input
-
-| Field | Required | Description |
-|-------|:--------:|-------------|
-| Product vision | ✅ | The north star must reflect the vision |
-| Business model | ✅ | Subscription / usage / hybrid |
-| Current metric stack | ✅ | MRR, DAU, retention, etc. |
-| Customer value definition | ✅ | What we consider value for the customer |
-| Adversarial camp (Full A) | ⬚ | alpha (Customer-first) / beta (Business-first) |
-
-## Data Sources
-
-1. Current product analytics (Amplitude, Mixpanel, custom).
-2. Customer interviews → what "valuable" means.
-3. Business metrics — NRR, churn.
-4. Industry benchmarks (SaaS Capital, OpenView reports).
-
-### Related Skills
-
-| Skill | What we take | When to invoke |
-|-------|-------------|----------------|
-| `product-vision` | Vision → NSM alignment | Before choosing NSM |
-| `aarrr-metrics` | Framework for decomposition | For input metrics |
-| `saas-metrics` | Business metric layer | For lagging alignment |
-| `okr-framework` | OKR KRs → often tied to NSM | After NSM |
-
-## Criteria for a Good NSM
-
-1. **Reflects customer value** — when NSM grows, customers receive value.
-2. **Leading indicator of revenue** — moves before MRR.
-3. **Actionable** — the team can influence it.
-4. **Singular** — not composite, not aggregate.
-5. **Measurable** — numerically trackable.
-6. **Resistant to gaming** — hard to inflate without real value.
-
-## Protocol
-
-### Step 0 — Role Framing (Full A)
-
-- **Customer-Champion (alpha):** emphasis on user-value metric
-- **Business-Champion (beta):** emphasis on business-value metric
-
-### Step 1 — Business Model Analysis
-
-- **Subscription SaaS:** NSM often related to weekly/monthly active teams, usage depth
-- **Usage-based:** raw consumption (API calls, workloads)
-- **Transaction-based:** successful transactions
-- **Platform:** marketplace size × match rate
-
-### Step 2 — Generate Candidates
-
-Minimum 5 candidates. Each — through the formula:
-
-**[Action verb]** × **[Unit with value]** × **[Frequency/Cadence]** × **[Quality]**
-
-B2B SaaS examples:
-- "Weekly Active Teams with ≥5 seats" (team platform)
-- "Monthly reports generated per paying account" (analytics tool)
-- "Weekly successful deployments per engineering team" (DevOps tool)
-- "Onboarded workspaces completing setup within 7 days" (product-led growth)
-
-### Step 3 — Evaluation
-
-| Metric | Customer Value | Leads Revenue | Actionable | Singular | Measurable | Anti-game | Total |
-|--------|:--------------:|:-------------:|:----------:|:--------:|:----------:|:---------:|:-----:|
-| [cand] | 5 | 4 | 4 | 5 | 5 | 3 | 26 |
-
-Rank. Top candidate = NSM. Second → guardrail or input metric.
-
-### Step 4 — Decomposition (Metric Tree)
-
-NSM → input metrics → leading indicators.
-
-```
-NSM: Weekly Active Teams ≥5 seats
-  ├── Input: New teams activating weekly
-  │     ├── Signup → activation rate
-  │     └── Time-to-5-seats (onboarding speed)
-  ├── Input: Existing teams retaining
-  │     ├── W/W retention rate
-  │     └── Seats-per-team growth
-  └── Guardrails:
-        ├── Churn rate (capped)
-        ├── NPS (no drop)
-        └── Support ticket volume (no spike)
-```
-
-### Step 5 — Guardrails
-
-What **MUST NOT break** in the pursuit of NSM:
-- Churn rate
-- NPS / CSAT
-- Gross margin
-- Support ticket volume
-- p95 latency / error rate
-
-Each guardrail = numeric threshold.
-
-### Step 6 — Baseline + Target
-
-- Current value
-- 90-day target
-- 1-year aspirational
-
-### Step 7 — Anti-Gaming Review
-
-Team brainstorm: "How would we game the NSM?". If it's easy — enhance the definition or change the metric.
-
-Example: "Weekly Active Teams" can be gamed by counting churned teams that were active for the last time → clarification: "with activity in last 7 days".
-
-## Validation (Quality Gate)
-
-- [ ] Minimum 5 candidates reviewed
-- [ ] Evaluation across 6 criteria
-- [ ] Metric tree decomposition (NSM → inputs)
-- [ ] Guardrails with numeric thresholds
-- [ ] Baseline + target
-- [ ] Anti-gaming review conducted
-- [ ] Business model alignment verified
-- [ ] Vision alignment verified
-
-## Handoff
-
-The result is input for:
-- **`okr-framework`** — NSM = basis for several KRs
-- **`aarrr-metrics`** — NSM fits into AARRR as the key stage
-- **Data Analyst** — instrumentation requirements
-- **Mediator** (Full A) — alpha and beta NSM for synthesis
-
-Format: NSM doc (one-pager) with metric tree. Via `$handoff`.
-
-## Anti-patterns
-
-| Error | Why it's bad | How to do it right |
-|-------|-------------|-------------------|
-| Revenue as NSM | Lagging, not actionable | Use leading value metric |
-| Vanity metric | Looks good, doesn't predict | Customer value + revenue proxy |
-| Composite metric | Weighted sum is not singular | Pick ONE |
-| No guardrails | Accept churn growing with NSM | Explicit guardrails |
-| Change every quarter | Team can't align | Stable for ≥ year |
-| Copy other company | Context-specific | Your business model |
-
-## Template
-
-```markdown
-# North Star Metric: [Name]
-
-## Definition
-**[Action]** × **[Unit]** × **[Frequency]** × **[Quality]**
-Precise: [exact definition with inclusion/exclusion rules]
-
-## Rationale
-[Why this metric — customer value + business]
-
-## Metric Tree
-NSM
-├── Input: ...
-├── Input: ...
-└── Guardrails:
-    ├── ...
-
-## Baseline & Targets
-- Current: X
-- 90-day target: Y
-- 1-year aspirational: Z
-
-## Anti-Gaming
-[How could this be gamed? Mitigation]
-
-## Review
-- Quarterly: still meaningful?
-- Annually: re-evaluate
-```
-
-## Worked Example — TeamFlow NSM Selection (post-Mediator synthesis)
-
-**Context:** After Mediator synthesis of visions, the team chooses a unified NSM. Alpha proposed "weekly 1:1s with AI summaries used"; Beta — "ARR from accounts with ≥3 seats on AI tier". Below — full candidate evaluation.
-
-### Business Model Analysis
-
-- **Revenue model:** Subscription, tiered (Core / Team / Enterprise)
-- **Growth lever:** Seat expansion (mid-market, enterprise) + tier upgrade
-- **Primary value delivery:** Per-1:1 workflow savings + per-manager adoption
-
-### 5 Candidate NSMs
-
-| # | Candidate | Formula | Leading/Lagging | Ties to vision? |
-|---|-----------|---------|:---------------:|:---------------:|
-| C1 | Weekly Active Managers using AI summarization (≥ 1 per week) | count(WAM_ai) | Leading | Alpha ✓ |
-| C2 | ARR from AI-tier accounts | $ ARR tagged «AI tier» | Lagging | Beta ✓ |
-| C3 | «Sticky Manager Ratio» — managers running ≥ 4 AI-assisted 1:1s per month | ratio | Leading + quality signal | Alpha ✓ |
-| C4 | «AI-assisted 1:1s completed per week» (total count across base) | count | Leading, usage-based | Alpha ✓ |
-| C5 | Net New AI Tier ARR / Quarter | $ | Lagging | Beta ✓ |
-
-### Evaluation Matrix (6 criteria × 5 candidates, scale 1-5)
-
-| Metric | Customer Value | Leads Revenue | Actionable | Singular | Measurable | Anti-Game | **Total** |
-|--------|:--------------:|:-------------:|:----------:|:--------:|:----------:|:---------:|:---------:|
-| C1 WAM_ai | 4 | 4 | 5 | 5 | 5 | 3 | **26** |
-| C2 ARR AI tier | 2 | 2 (= it IS revenue) | 3 | 5 | 5 | 4 | 21 |
-| C3 Sticky Manager Ratio | 5 | 4 | 5 | 5 | 5 | 4 | **28** |
-| C4 AI 1:1s/week | 4 | 4 | 4 | 5 | 5 | 2 (easy to bulk) | 24 |
-| C5 Net New AI ARR | 2 | 2 (= it IS revenue) | 2 | 5 | 5 | 4 | 20 |
-
-**Winner: C3 «Sticky Manager Ratio»** — balances customer value signal (quality — 4+ 1:1s/month implies real habit) + anti-gaming (can't be bulk-created; requires genuine usage) + clear leading indicator for ARR.
-
-### Chosen NSM — Definition
-
-```markdown
-# North Star Metric: Sticky Manager Ratio (AI Tier)
-
-## Definition
-**Active managers** × **Using AI summarization** × **≥ 4 times per month** × **on ≥ 1 direct report**
-
-**Precise:**
-- **Active manager:** has ≥ 1 direct report in TeamFlow + logged in in last 30 days
-- **Using AI summarization:** completed 1:1 with AI summary generated AND reviewed (not just generated but ignored)
-- **≥ 4 per month:** rolling 30-day window
-- **Exclusions:** internal TeamFlow employees; accounts <30 days old (onboarding); trial accounts
-
-**Formula:**
-```
-Sticky Manager Ratio = Managers meeting criteria / Total active managers in AI-tier accounts
-```
-
-## Rationale
-
-**Why this metric:**
-
-- **Customer value signal:** 4 uses/month ≈ weekly 1:1s = genuine habit, not experimentation. Habit = product embedded in workflow → high retention.
-- **Leading revenue:** Sticky managers → account retention → expansion → NRR. Historical TeamFlow data: features used ≥ 4×/month have 3× LTV.
-- **Actionable:** Team can drive (onboarding, reminders, feature quality).
-- **Singular:** One number, clear dashboard.
-- **Resistant to gaming:** can't «bulk» real usage. Can be inflated via trial seats, so trial accounts excluded.
-
-**Why not alternatives:**
-- C1 WAM_ai: doesn't distinguish casual from habitual usage
-- C2 ARR tier: lagging, doesn't predict expansion before it happens
-- C4 1:1s/week: easily gamed («create 20 fake 1:1s»)
-- C5 Net new ARR: same lagging issue
-
-## Metric Tree
-
-```
-NSM: Sticky Manager Ratio (AI Tier)
-  ├── Input: AI Tier account count (expansion)
-  │     ├── New AI Tier conversions per week
-  │     └── AI Tier churn
-  ├── Input: Active manager activation rate (% new managers reaching ≥ 4 1:1s)
-  │     ├── Time to first AI summary
-  │     └── Onboarding checklist completion
-  ├── Input: Feature quality signal
-  │     ├── Summary approval rate (kept vs edited >50%)
-  │     └── Summary NPS (in-product rating)
-  └── Guardrails:
-        ├── Overall NPS (must not drop > 5 points)
-        ├── Gross churn rate (must not grow >1pp QoQ)
-        ├── Support ticket volume / AI issues (monitor for abuse / dissatisfaction)
-        └── p95 summary generation latency (< 30 sec)
-```
-
-## Baseline & Targets
-
-| Horizon | Target | Rationale |
-|---------|:------:|-----------|
-| Current (Q1 2026) | 0% (feature not live) | Pre-launch |
-| 90 days post-launch (end Q2) | **20%** | Conservative — based on similar AI feature launches in adjacent B2B |
-| End of 2026 | 45% | Category-leading level |
-| End of 2027 (aspirational) | 65% | «Cannot imagine working without it» level |
-
-## Anti-Gaming Analysis
-
-**How could this be gamed?**
-
-- **Bulk fake 1:1 creation:** Mitigation — require AI summary generated AND reviewed (visible edit or approval event)
-- **Single massive account inflating ratio:** Mitigation — weight by account size + report separately
-- **Managers «maintaining streak» with perfunctory 1:1s:** Mitigation — 1:1 duration floor (must be ≥ 10 min scheduled)
-- **Team inflating by marking non-actual 1:1s:** Mitigation — periodic sample audit with customer success conversations
-
-## Review
-- **Weekly:** ratio tracked in exec dashboard
-- **Quarterly:** anti-gaming sample audit (3 accounts reviewed manually)
-- **Annually:** NSM re-evaluated against strategy shift
-```
-
-> **NSM lesson:** Alpha's initial «WAM_ai» was good but too blunt — didn't distinguish trial adoption from habit. Beta's «ARR AI tier» was too lagging. Synthesis («Sticky Manager Ratio») borrowed Alpha's user-centric formulation but applied Beta's rigor about quality threshold (4+ per month = real habit). Result is a metric that **leads** revenue, **measures** customer value, and **resists** easy gaming. This is the kind of refinement that's hard to get on first try — the evaluation matrix forced the conversation.
+---
+name: north-star-metric
+description: Selecting the NSM for B2B SaaS — balancing leading/lagging, alignment with customer value and business outcome
+type: triggered
+domain: product
+owners:
+  - product_strategist
+gates:
+  - PRODUCT_STRATEGIST
+tech: []
+topic: []
+triggers:
+  - "north-star-metric"
+  - "north star"
+  - "NSM"
+  - "ведущая метрика"
+related: []
+budget_lines: 317
+schema_version: 1
+---
+# North Star Metric (NSM)
+
+> **Category:** Strategy  ·  **Slug:** `north-star-metric`
+
+## When to Use
+
+- When defining product strategy — a single metric of focus.
+- When aligning the team: decision-making through the lens of "does this lift the NSM?".
+- For board / investor reporting — single metric story.
+- When reviewing an existing metric — still meaningful?
+
+## Input
+
+| Field | Required | Description |
+|-------|:--------:|-------------|
+| Product vision | ✅ | The north star must reflect the vision |
+| Business model | ✅ | Subscription / usage / hybrid |
+| Current metric stack | ✅ | MRR, DAU, retention, etc. |
+| Customer value definition | ✅ | What we consider value for the customer |
+| Adversarial camp (Full A) | ⬚ | alpha (Customer-first) / beta (Business-first) |
+
+## Data Sources
+
+1. Current product analytics (Amplitude, Mixpanel, custom).
+2. Customer interviews → what "valuable" means.
+3. Business metrics — NRR, churn.
+4. Industry benchmarks (SaaS Capital, OpenView reports).
+
+### Related Skills
+
+| Skill | What we take | When to invoke |
+|-------|-------------|----------------|
+| `product-vision` | Vision → NSM alignment | Before choosing NSM |
+| `aarrr-metrics` | Framework for decomposition | For input metrics |
+| `saas-metrics` | Business metric layer | For lagging alignment |
+| `okr-framework` | OKR KRs → often tied to NSM | After NSM |
+
+## Criteria for a Good NSM
+
+1. **Reflects customer value** — when NSM grows, customers receive value.
+2. **Leading indicator of revenue** — moves before MRR.
+3. **Actionable** — the team can influence it.
+4. **Singular** — not composite, not aggregate.
+5. **Measurable** — numerically trackable.
+6. **Resistant to gaming** — hard to inflate without real value.
+
+## Protocol
+
+### Step 0 — Role Framing (Full A)
+
+- **Customer-Champion (alpha):** emphasis on user-value metric
+- **Business-Champion (beta):** emphasis on business-value metric
+
+### Step 1 — Business Model Analysis
+
+- **Subscription SaaS:** NSM often related to weekly/monthly active teams, usage depth
+- **Usage-based:** raw consumption (API calls, workloads)
+- **Transaction-based:** successful transactions
+- **Platform:** marketplace size × match rate
+
+### Step 2 — Generate Candidates
+
+Minimum 5 candidates. Each — through the formula:
+
+**[Action verb]** × **[Unit with value]** × **[Frequency/Cadence]** × **[Quality]**
+
+B2B SaaS examples:
+- "Weekly Active Teams with ≥5 seats" (team platform)
+- "Monthly reports generated per paying account" (analytics tool)
+- "Weekly successful deployments per engineering team" (DevOps tool)
+- "Onboarded workspaces completing setup within 7 days" (product-led growth)
+
+### Step 3 — Evaluation
+
+| Metric | Customer Value | Leads Revenue | Actionable | Singular | Measurable | Anti-game | Total |
+|--------|:--------------:|:-------------:|:----------:|:--------:|:----------:|:---------:|:-----:|
+| [cand] | 5 | 4 | 4 | 5 | 5 | 3 | 26 |
+
+Rank. Top candidate = NSM. Second → guardrail or input metric.
+
+### Step 4 — Decomposition (Metric Tree)
+
+NSM → input metrics → leading indicators.
+
+```
+NSM: Weekly Active Teams ≥5 seats
+  ├── Input: New teams activating weekly
+  │     ├── Signup → activation rate
+  │     └── Time-to-5-seats (onboarding speed)
+  ├── Input: Existing teams retaining
+  │     ├── W/W retention rate
+  │     └── Seats-per-team growth
+  └── Guardrails:
+        ├── Churn rate (capped)
+        ├── NPS (no drop)
+        └── Support ticket volume (no spike)
+```
+
+### Step 5 — Guardrails
+
+What **MUST NOT break** in the pursuit of NSM:
+- Churn rate
+- NPS / CSAT
+- Gross margin
+- Support ticket volume
+- p95 latency / error rate
+
+Each guardrail = numeric threshold.
+
+### Step 6 — Baseline + Target
+
+- Current value
+- 90-day target
+- 1-year aspirational
+
+### Step 7 — Anti-Gaming Review
+
+Team brainstorm: "How would we game the NSM?". If it's easy — enhance the definition or change the metric.
+
+Example: "Weekly Active Teams" can be gamed by counting churned teams that were active for the last time → clarification: "with activity in last 7 days".
+
+## Validation (Quality Gate)
+
+- [ ] Minimum 5 candidates reviewed
+- [ ] Evaluation across 6 criteria
+- [ ] Metric tree decomposition (NSM → inputs)
+- [ ] Guardrails with numeric thresholds
+- [ ] Baseline + target
+- [ ] Anti-gaming review conducted
+- [ ] Business model alignment verified
+- [ ] Vision alignment verified
+
+## Handoff
+
+The result is input for:
+- **`okr-framework`** — NSM = basis for several KRs
+- **`aarrr-metrics`** — NSM fits into AARRR as the key stage
+- **Data Analyst** — instrumentation requirements
+- **Mediator** (Full A) — alpha and beta NSM for synthesis
+
+Format: NSM doc (one-pager) with metric tree. Via `$handoff`.
+
+## Anti-patterns
+
+| Error | Why it's bad | How to do it right |
+|-------|-------------|-------------------|
+| Revenue as NSM | Lagging, not actionable | Use leading value metric |
+| Vanity metric | Looks good, doesn't predict | Customer value + revenue proxy |
+| Composite metric | Weighted sum is not singular | Pick ONE |
+| No guardrails | Accept churn growing with NSM | Explicit guardrails |
+| Change every quarter | Team can't align | Stable for ≥ year |
+| Copy other company | Context-specific | Your business model |
+
+## Template
+
+```markdown
+# North Star Metric: [Name]
+
+## Definition
+**[Action]** × **[Unit]** × **[Frequency]** × **[Quality]**
+Precise: [exact definition with inclusion/exclusion rules]
+
+## Rationale
+[Why this metric — customer value + business]
+
+## Metric Tree
+NSM
+├── Input: ...
+├── Input: ...
+└── Guardrails:
+    ├── ...
+
+## Baseline & Targets
+- Current: X
+- 90-day target: Y
+- 1-year aspirational: Z
+
+## Anti-Gaming
+[How could this be gamed? Mitigation]
+
+## Review
+- Quarterly: still meaningful?
+- Annually: re-evaluate
+```
+
+## Worked Example — TeamFlow NSM Selection (post-Mediator synthesis)
+
+**Context:** After Mediator synthesis of visions, the team chooses a unified NSM. Alpha proposed "weekly 1:1s with AI summaries used"; Beta — "ARR from accounts with ≥3 seats on AI tier". Below — full candidate evaluation.
+
+### Business Model Analysis
+
+- **Revenue model:** Subscription, tiered (Core / Team / Enterprise)
+- **Growth lever:** Seat expansion (mid-market, enterprise) + tier upgrade
+- **Primary value delivery:** Per-1:1 workflow savings + per-manager adoption
+
+### 5 Candidate NSMs
+
+| # | Candidate | Formula | Leading/Lagging | Ties to vision? |
+|---|-----------|---------|:---------------:|:---------------:|
+| C1 | Weekly Active Managers using AI summarization (≥ 1 per week) | count(WAM_ai) | Leading | Alpha ✓ |
+| C2 | ARR from AI-tier accounts | $ ARR tagged «AI tier» | Lagging | Beta ✓ |
+| C3 | «Sticky Manager Ratio» — managers running ≥ 4 AI-assisted 1:1s per month | ratio | Leading + quality signal | Alpha ✓ |
+| C4 | «AI-assisted 1:1s completed per week» (total count across base) | count | Leading, usage-based | Alpha ✓ |
+| C5 | Net New AI Tier ARR / Quarter | $ | Lagging | Beta ✓ |
+
+### Evaluation Matrix (6 criteria × 5 candidates, scale 1-5)
+
+| Metric | Customer Value | Leads Revenue | Actionable | Singular | Measurable | Anti-Game | **Total** |
+|--------|:--------------:|:-------------:|:----------:|:--------:|:----------:|:---------:|:---------:|
+| C1 WAM_ai | 4 | 4 | 5 | 5 | 5 | 3 | **26** |
+| C2 ARR AI tier | 2 | 2 (= it IS revenue) | 3 | 5 | 5 | 4 | 21 |
+| C3 Sticky Manager Ratio | 5 | 4 | 5 | 5 | 5 | 4 | **28** |
+| C4 AI 1:1s/week | 4 | 4 | 4 | 5 | 5 | 2 (easy to bulk) | 24 |
+| C5 Net New AI ARR | 2 | 2 (= it IS revenue) | 2 | 5 | 5 | 4 | 20 |
+
+**Winner: C3 «Sticky Manager Ratio»** — balances customer value signal (quality — 4+ 1:1s/month implies real habit) + anti-gaming (can't be bulk-created; requires genuine usage) + clear leading indicator for ARR.
+
+### Chosen NSM — Definition
+
+```markdown
+# North Star Metric: Sticky Manager Ratio (AI Tier)
+
+## Definition
+**Active managers** × **Using AI summarization** × **≥ 4 times per month** × **on ≥ 1 direct report**
+
+**Precise:**
+- **Active manager:** has ≥ 1 direct report in TeamFlow + logged in in last 30 days
+- **Using AI summarization:** completed 1:1 with AI summary generated AND reviewed (not just generated but ignored)
+- **≥ 4 per month:** rolling 30-day window
+- **Exclusions:** internal TeamFlow employees; accounts <30 days old (onboarding); trial accounts
+
+**Formula:**
+```
+Sticky Manager Ratio = Managers meeting criteria / Total active managers in AI-tier accounts
+```
+
+## Rationale
+
+**Why this metric:**
+
+- **Customer value signal:** 4 uses/month ≈ weekly 1:1s = genuine habit, not experimentation. Habit = product embedded in workflow → high retention.
+- **Leading revenue:** Sticky managers → account retention → expansion → NRR. Historical TeamFlow data: features used ≥ 4×/month have 3× LTV.
+- **Actionable:** Team can drive (onboarding, reminders, feature quality).
+- **Singular:** One number, clear dashboard.
+- **Resistant to gaming:** can't «bulk» real usage. Can be inflated via trial seats, so trial accounts excluded.
+
+**Why not alternatives:**
+- C1 WAM_ai: doesn't distinguish casual from habitual usage
+- C2 ARR tier: lagging, doesn't predict expansion before it happens
+- C4 1:1s/week: easily gamed («create 20 fake 1:1s»)
+- C5 Net new ARR: same lagging issue
+
+## Metric Tree
+
+```
+NSM: Sticky Manager Ratio (AI Tier)
+  ├── Input: AI Tier account count (expansion)
+  │     ├── New AI Tier conversions per week
+  │     └── AI Tier churn
+  ├── Input: Active manager activation rate (% new managers reaching ≥ 4 1:1s)
+  │     ├── Time to first AI summary
+  │     └── Onboarding checklist completion
+  ├── Input: Feature quality signal
+  │     ├── Summary approval rate (kept vs edited >50%)
+  │     └── Summary NPS (in-product rating)
+  └── Guardrails:
+        ├── Overall NPS (must not drop > 5 points)
+        ├── Gross churn rate (must not grow >1pp QoQ)
+        ├── Support ticket volume / AI issues (monitor for abuse / dissatisfaction)
+        └── p95 summary generation latency (< 30 sec)
+```
+
+## Baseline & Targets
+
+| Horizon | Target | Rationale |
+|---------|:------:|-----------|
+| Current (Q1 2026) | 0% (feature not live) | Pre-launch |
+| 90 days post-launch (end Q2) | **20%** | Conservative — based on similar AI feature launches in adjacent B2B |
+| End of 2026 | 45% | Category-leading level |
+| End of 2027 (aspirational) | 65% | «Cannot imagine working without it» level |
+
+## Anti-Gaming Analysis
+
+**How could this be gamed?**
+
+- **Bulk fake 1:1 creation:** Mitigation — require AI summary generated AND reviewed (visible edit or approval event)
+- **Single massive account inflating ratio:** Mitigation — weight by account size + report separately
+- **Managers «maintaining streak» with perfunctory 1:1s:** Mitigation — 1:1 duration floor (must be ≥ 10 min scheduled)
+- **Team inflating by marking non-actual 1:1s:** Mitigation — periodic sample audit with customer success conversations
+
+## Review
+- **Weekly:** ratio tracked in exec dashboard
+- **Quarterly:** anti-gaming sample audit (3 accounts reviewed manually)
+- **Annually:** NSM re-evaluated against strategy shift
+```
+
+> **NSM lesson:** Alpha's initial «WAM_ai» was good but too blunt — didn't distinguish trial adoption from habit. Beta's «ARR AI tier» was too lagging. Synthesis («Sticky Manager Ratio») borrowed Alpha's user-centric formulation but applied Beta's rigor about quality threshold (4+ per month = real habit). Result is a metric that **leads** revenue, **measures** customer value, and **resists** easy gaming. This is the kind of refinement that's hard to get on first try — the evaluation matrix forced the conversation.