Upload evaluation/product/replies.jsonl with huggingface_hub

evaluation/product/replies.jsonl

@@ -0,0 +1,20 @@
+{"task_id": "v1_vs_scale", "category": "product_improvement", "prompt": "Past MVP at Cartspine but v1 is mushy. North-star: paid seats retained 90d (D30 51%). Headless checkout for DTC brands. What belongs in v1 vs later? Give a clear cut list.", "reply": "## Framing\nFor Cartspine (headless checkout for DTC brands) at pre-seed, roadmap is a **capacity allocation**, not a wish list. Team size 2–6 people; ~1 squad(s); optimize for North-star: paid seats retained 90d and learning speed with Platform engineers at mid-size tech firms. Current rough pulse: D30~51%, NRR~86%, ARPU~$1079.\n\n## Sequencing (MVP → v1 → scale)\n### Horizon 0 — MVP (now to 6 weeks)\n1. Nail the core loop of headless checkout for DTC brands for one ICP slice: Platform engineers at mid-size tech firms.\n2. Ship **multi-workspace SSO** as the thinnest proof of value (instrumented).\n3. Explicit non-goals: no sandbox environments platform work; no multi-segment expansion; no custom one-offs.\n4. Success gate: 9–14 design partners using weekly; qualitative disappointment risk measured; leading indicator for North-star: paid seats retained 90d defined in writing.\n\n### Horizon 1 — v1 (next 1–2 quarters)\n1. Harden reliability + onboarding so TTV drops; unlocks North-star: paid seats retained 90d.\n2. Add **SOC2 evidence collector** only if MVP gate passed — sales-assisted differentiator.\n3. Packaging: one clear paid tier; avoid enterprise custom until ≥5 logos ask the same thing.\n4. Success gate: paid conversion / retention threshold pre-committed with the board (e.g., trial→paid ≥4% or D30 ≥56).\n\n### Horizon 2 — Scale (post-PMF)\n1. Then: **sandbox environments** / platform leverage, SSO-depth, ecosystem.\n2. Second ICP experiment as time-boxed bet (≤1 squad, 1 quarter).\n3. Ops: SLOs, support tiers, implementation playbooks; target NRR ≥100%.\n\n## Tradeoffs\n- Building sandbox environments early feels like leverage but delays proof of the core loop.\n- Saying yes to every sales request collapses sequencing into a junk drawer.\n- At pre-seed, speed of validated learning > completeness.\n- Only one squad free for the next 6 weeks.\n\n## Recommendation\n**Sequence: multi", "error": null, "score": 1.0, "quality_0_10": 6.3}
+{"task_id": "prioritize_reliability", "category": "product_improvement", "prompt": "CareSlot: SE says demo environment is unreliable. ICP = RevOps at 20–100 person sales orgs. Prioritize 3 product responses with impact/effort.", "reply": "## Situation snapshot\nCareSlot is seed ($20k–$80k MRR, 8–20 people) selling multi-provider telehealth scheduling to RevOps at 20–100 person sales orgs. Primary lens: weekly active workflows / account. Capacity: **3 squad(s)**. Signal density last 30d: ~82 tickets / ~23 deal mentions. Burn context ~$61k/mo; runway pressure typical for seed.\n\n## Scored options (ICE + rough RICE-style priority)\n| Bet | Impact | Confidence | Ease | Priority score | Call |\n|---|---:|---:|---:|---:|---|\n| inventory low-stock AI | 8 | 6 | 9 | 33.6 | **INVEST** |\n| smart form branching | 5 | 9 | 6 | 45.0 | **KEEP** |\n| AI summary of churn calls | 9 | 5 | 9 | 31.5 | **KILL / park** |\n\n## Kill / Keep / Invest detail\n1. **INVEST — inventory low-stock AI**\n   - Why: Closest to the job-to-be-done for RevOps at 20–100 person sales orgs; compounds weekly active workflows / account; defensible vs Gorgias (e-comm helpdesk, limited B2B depth).\n   - 90-day scope: thinnest version that changes a weekly behavior; instrument before polish; 10 design partners.\n   - Success gate: move weekly active workflows / account meaningfully OR prove a leading indicator (e.g., weekly active workflow +9 pts among partners).\n\n2. **KEEP (maintenance only) — smart form branching**\n   - Why: Useful hygiene / sales checkbox, not the wedge. Fix P0 bugs only; no dedicated squad.\n   - Neglect risk: demo friction on ~23 deals. Overbuild risk: distraction from invest bet.\n\n3. **KILL (or park ≥2 quarters) — AI summary of churn calls**\n   - Why: Attractive narrative, weak causal link to weekly active workflows / account at seed. Competes for the same eng attention as inventory low-stock AI.\n   - Revisit only if weekly active workflows / account is healthy", "error": null, "score": 1.0, "quality_0_10": 8.0}
+{"task_id": "roadmap_fight", "category": "roadmap_sequencing", "prompt": "Inkboard roadmap fight: eng wants role-based dashboards, sales wants CSV bulk importer v2, CEO wants usage-based billing meter. Series B. Sequence 2 quarters with rationale.", "reply": "## Framing\nFor Inkboard (collaborative brand guidelines for design teams) at Series B, roadmap is a **capacity allocation**, not a wish list. Team size 60–200 people; ~3 squad(s); optimize for NPS / CSAT on onboarding and learning speed with Procurement leads at 200–2000 employee firms. Current rough pulse: D30~47%, NRR~115%, ARPU~$1190.\n\n## Sequencing (MVP → v1 → scale)\n### Horizon 0 — MVP (now to 6 weeks)\n1. Nail the core loop of collaborative brand guidelines for design teams for one ICP slice: Procurement leads at 200–2000 employee firms.\n2. Ship **CSV bulk importer v2** as the thinnest proof of value (instrumented).\n3. Explicit non-goals: no usage-based billing meter platform work; no multi-segment expansion; no custom one-offs.\n4. Success gate: 10–15 design partners using weekly; qualitative disappointment risk measured; leading indicator for NPS / CSAT on onboarding defined in writing.\n\n### Horizon 1 — v1 (next 1–2 quarters)\n1. Harden reliability + onboarding so TTV drops; unlocks NPS / CSAT on onboarding.\n2. Add **role-based dashboards** only if MVP gate passed — sales-assisted differentiator.\n3. Packaging: one clear paid tier; avoid enterprise custom until ≥9 logos ask the same thing.\n4. Success gate: paid conversion / retention threshold pre-committed with the board (e.g., trial→paid ≥18% or D30 ≥52).\n\n### Horizon 2 — Scale (post-PMF)\n1. Then: **usage-based billing meter** / platform leverage, SSO-depth, ecosystem.\n2. Second ICP experiment as time-boxed bet (≤1 squad, 1 quarter).\n3. Ops: SLOs, support tiers, implementation playbooks; target NRR ≥115%.\n\n## Tradeoffs\n- Building usage-based billing meter early feels like leverage but delays proof of the core loop.\n- Saying yes to every sales request collapses sequencing into a junk drawer.\n- At Series B, speed of validated learning > completeness.\n- Security review adds 2", "error": null, "score": 1.0, "quality_0_10": 6.3}
+{"task_id": "kill_keep_invest", "category": "roadmap_sequencing", "prompt": "Kill/keep/invest for Clipvault (Series A): in-app guided tours vs collaborative annotation vs public API. Decide with criteria.", "reply": "## Situation snapshot\nClipvault is Series A ($80k–$300k MRR, 20–60 people) selling asset library + rights for creator agencies to Procurement leads at 200–2000 employee firms. Primary lens: attach rate of add-on SKU. Capacity: **1 squad(s)**. Signal density last 30d: ~420 tickets / ~25 deal mentions. Burn context ~$68k/mo; runway pressure typical for Series A.\n\n## Scored options (ICE + rough RICE-style priority)\n| Bet | Impact | Confidence | Ease | Priority score | Call |\n|---|---:|---:|---:|---:|---|\n| in-app guided tours | 7 | 7 | 8 | 39.2 | **INVEST** |\n| collaborative annotation | 4 | 4 | 5 | 14.4 | **KEEP** |\n| public API | 8 | 6 | 8 | 38.4 | **KILL / park** |\n\n## Kill / Keep / Invest detail\n1. **INVEST — in-app guided tours**\n   - Why: Closest to the job-to-be-done for Procurement leads at 200–2000 employee firms; compounds attach rate of add-on SKU; defensible vs Chartable (podcast measurement, limited monetization OS).\n   - 90-day scope: thinnest version that changes a weekly behavior; instrument before polish; 12 design partners.\n   - Success gate: move attach rate of add-on SKU meaningfully OR prove a leading indicator (e.g., weekly active workflow +10 pts among partners).\n\n2. **KEEP (maintenance only) — collaborative annotation**\n   - Why: Useful hygiene / sales checkbox, not the wedge. Fix P0 bugs only; no dedicated squad.\n   - Neglect risk: demo friction on ~25 deals. Overbuild risk: distraction from invest bet.\n\n3. **KILL (or park ≥2 quarters) — public API**\n   - Why: Attractive narrative, weak causal link to attach rate of add-on SKU at Series A. Competes for the same eng attention as in-app guided tours.\n   - Revisit only if attach rate of add-on SKU is healthy AND a named enterprise", "error": null, "score": 1.0, "quality_0_10": 8.0}
+{"task_id": "north_star_guardrails", "category": "metrics_okrs", "prompt": "Clearspan Claims: sales cycle elongated 2 weeks after pricing change. Customers~103. Propose a north-star metric + 3 guardrail metrics + kill criteria.", "reply": "## Diagnosis\nClearspan Claims is showing: **sales cycle elongated 2 weeks after pricing change**. At seed with focus on attach rate of add-on SKU, the usual failure mode is mixing vanity, output, and outcome metrics on one slide. Pulse today: D30~53%, trial→paid~7%, NRR~88%, ~103 active accounts.\n\n## North-star choice\n**Proposed north-star:** weekly count of ICP accounts completing the core job of claims intake for regional insurers (not signups, not pageviews). For Procurement leads at 200–2000 employee firms, define the aha event in one sentence and instrument it.\n- Working target band: ~431 weekly activated ICP accounts within 1 quarters (set baseline this week — do not fake precision).\n- Guardrail metrics: gross margin, logo churn, CAC payback.\n- Counter-metrics: if you push activation hard, watch support tickets / 100 users (currently theme volume ~189/30d) and quality complaints.\n\n## Input tree\n1. **Acquisition quality** — ICP fit of trials (not raw MQL volume).\n2. **Activation** — time-to-aha; % completing setup checklist.\n3. **Habit** — weekly active workflows / account.\n4. **Value realization** — outcome the customer hired you for.\n5. **Monetization** — conversion, expansion, NRR.\n6. **Retention** — reasons-coded churn; resurrectable vs dead.\n\n## Quarterly OKR sketch\n- **O1:** Make the core job reliably valuable for Procurement leads at 200–2000 employee firms.\n  - KR1: North-star weekly active ICP accounts +13%.\n  - KR2: Median TTV reduced by 5 days.\n  - KR3: D30 retention +6 pts among activated accounts.\n- **O2:** Make growth efficient for the next fundraise / profitability path.\n  - KR1: CAC payback ≤ 15 months.\n  - KR2: NRR ≥ 100% (or +6 pts if already above).\n\n## Tradeoffs\n- A broader north-star looks board-friendly but hides the failing branch of the tree.\n- Optimizing only conversion without retention creates a leaky bucket vs Chartable.\n- Do not propose hiring", "error": null, "score": 0.8889, "quality_0_10": 7.02}
+{"task_id": "diagnostic_tree", "category": "metrics_okrs", "prompt": "Streamgate: trial-to-paid fell 12%→6%. Build a diagnostic tree (3 branches) and what data to pull this week.", "reply": "## Diagnosis\nStreamgate is showing: **trial-to-paid fell 12%→6%**. At seed with focus on implementation time to go-live, the usual failure mode is mixing vanity, output, and outcome metrics on one slide. Pulse today: D30~41%, trial→paid~6%, NRR~88%, ~30 active accounts.\n\n## North-star choice\n**Proposed north-star:** weekly count of ICP accounts completing the core job of API rate-plan manager for platforms (not signups, not pageviews). For SMB CFOs / controllers at $2–20M revenue, define the aha event in one sentence and instrument it.\n- Working target band: ~510 weekly activated ICP accounts within 2 quarters (set baseline this week — do not fake precision).\n- Guardrail metrics: gross margin, logo churn, CAC payback.\n- Counter-metrics: if you push activation hard, watch support tickets / 100 users (currently theme volume ~43/30d) and quality complaints.\n\n## Input tree\n1. **Acquisition quality** — ICP fit of trials (not raw MQL volume).\n2. **Activation** — time-to-aha; % completing setup checklist.\n3. **Habit** — weekly active workflows / account.\n4. **Value realization** — outcome the customer hired you for.\n5. **Monetization** — conversion, expansion, NRR.\n6. **Retention** — reasons-coded churn; resurrectable vs dead.\n\n## Quarterly OKR sketch\n- **O1:** Make the core job reliably valuable for SMB CFOs / controllers at $2–20M revenue.\n  - KR1: North-star weekly active ICP accounts +13%.\n  - KR2: Median TTV reduced by 5 days.\n  - KR3: D30 retention +6 pts among activated accounts.\n- **O2:** Make growth efficient for the next fundraise / profitability path.\n  - KR1: CAC payback ≤ 14 months.\n  - KR2: NRR ≥ 100% (or +6 pts if already above).\n\n## Tradeoffs\n- A broader north-star looks board-friendly but hides the failing branch of the tree.\n- Optimizing only conversion without retention creates a leaky bucket vs Zendesk.\n- Cash", "error": null, "score": 0.8889, "quality_0_10": 7.02}
+{"task_id": "competitor_amplitude", "category": "competitive_positioning", "prompt": "Competitor Amplitude launched predictive churn alerts. We are Hireproof (B2B SaaS analytics). Panic vs opportunity? 90-day positioning plan.", "reply": "## Competitive reality\nAmplitude: broader product analytics, heavier setup. Hireproof wins when the buyer is VP Product at Series A–B PLG SaaS hiring you for structured interview kits for startups — not when the buyer wants a horizontal suite. Recent pressure: ~25 relevant deals; stage Series A ($80k–$300k MRR).\n\n## Positioning statement (working)\nFor **VP Product at Series A–B PLG SaaS** who need **structured interview kits for startups**, Hireproof is the **category specialist** that gets them to value faster than Amplitude, unlike horizontals that require heavy configuration.\n\n## Differentiation pillars (pick 2 to over-invest)\n1. **Workflow depth for the ICP** — own the last-mile job, not the dashboard.\n2. **Time-to-value** — opinionated defaults; measure TTV vs Amplitude (target beat by 7 days median).\n3. **Proof** — 4 case studies with before/after on a metric buyers already track.\n4. **Distribution** — partner or community Amplitude under-serves.\n\n## Do not compete here\n- Feature parity checklists against the full Amplitude suite.\n- Price wars on seats if your value is outcome/speed.\n- Enterprise theater before you have a repeatable mid-market motion (unless that is already your stage truth).\n- Cloning predictive churn alerts just because they shipped a press-release version.\n\n## Tradeoffs\n- Narrow wedge = clearer sales story, smaller TAM slide — manage with land-and-expand.\n- Chasing predictive churn alerts because they shipped it is how you become a worse clone.\n- Investing in SLA countdown timers for parity may win RFPs and still lose retention.\n- Exclude marketing copy and sales scripts from the answer.\n\n## Recommendation\n**Compete on ICP-specific workflow + TTV proof, not on suite breadth.** Make 'predictive churn alerts' a footnote unless discovery frequency shows it blocks the core job (≥6 tagged losses/month). Kill two parity-only roadmap items this week.\n\n## Next actions\n1. Rewrite the one-liner and force every deck slide to use it (enablement, not ad copy).\n2. Build a win/loss taxonomy: lost-to-Ampl", "error": null, "score": 0.875, "quality_0_10": 5.99}
+{"task_id": "vs_zendesk", "category": "competitive_positioning", "prompt": "Brightpath LMS vs Zendesk for healthcare compliance training SMB. They win on tickets; we win on AI triage quality. Positioning + battlecard bullets.", "reply": "## Competitive reality\nZendesk: ticket volume play, weak AI triage quality. Brightpath LMS wins when the buyer is Clinic administrators running telehealth hiring you for compliance training for healthcare orgs — not when the buyer wants a horizontal suite. Recent pressure: ~25 relevant deals; stage seed ($20k–$80k MRR).\n\n## Positioning statement (working)\nFor **Clinic administrators running telehealth** who need **compliance training for healthcare orgs**, Brightpath LMS is the **category specialist** that gets them to value faster than Zendesk, unlike horizontals that require heavy configuration.\n\n## Differentiation pillars (pick 2 to over-invest)\n1. **Workflow depth for the ICP** — own the last-mile job, not the dashboard.\n2. **Time-to-value** — opinionated defaults; measure TTV vs Zendesk (target beat by 5 days median).\n3. **Proof** — 4 case studies with before/after on a metric buyers already track.\n4. **Distribution** — partner or community Zendesk under-serves.\n\n## Do not compete here\n- Feature parity checklists against the full Zendesk suite.\n- Price wars on seats if your value is outcome/speed.\n- Enterprise theater before you have a repeatable mid-market motion (unless that is already your stage truth).\n- Cloning audit log + compliance exports just because they shipped a press-release version.\n\n## Tradeoffs\n- Narrow wedge = clearer sales story, smaller TAM slide — manage with land-and-expand.\n- Chasing audit log + compliance exports because they shipped it is how you become a worse clone.\n- Investing in partner referral portal for parity may win RFPs and still lose retention.\n- Do not propose hiring more than one role this quarter.\n\n## Recommendation\n**Compete on ICP-specific workflow + TTV proof, not on suite breadth.** Make 'audit log + compliance exports' a footnote unless discovery frequency shows it blocks the core job (≥6 tagged losses/month). Kill two parity-only roadmap items this week.\n\n## Next actions\n1. Rewrite the one-liner and force every deck slide to use it (enablement, not ad copy).\n2. Build a win/loss taxonomy: lost-to-Zendesk / lost-to-status-quo /", "error": null, "score": 0.875, "quality_0_10": 6.99}
+{"task_id": "pricing_rethink", "category": "pricing_monetization", "prompt": "BeaconCRM pricing rethink: currently 10% GMV take rate. Series A agency CRM. Issue: grandfathered legacy pricing. Propose 2 packaging options + migration plan.", "reply": "## Pricing diagnosis\nBeaconCRM currently: **10% GMV take rate** (marketplace). Issue: **grandfathered legacy pricing**. At Series A, packaging errors usually come from selling to multiple ICPs with one fuzzy ladder. Pulse: ARPU~$1150, NRR~105%, ~30 paying accounts.\n\n## Principles\n1. Price to the value metric the buyer already understands — not only what is easy to meter.\n2. Good / Better / Best with one obvious recommended tier.\n3. Put differentiation features in Better/Best; keep Better as the sales default.\n4. Grandfather carefully; surprise bills destroy trust faster than underpricing.\n5. Test on new logos before shocking the base.\n\n## Proposed packaging\n- **Starter (~$51/mo):** core lightweight CRM for agencies loop; capped usage; self-serve; for VP Product at Series A–B PLG SaaS tire-kickers.\n- **Growth (~$256/mo) (recommended):** collaboration + predictive churn alerts + priority support; where ARPU should live.\n- **Scale (~$1103/mo or annual commit):** SSO/audit/SKU margin waterfall/SLA; sold assisted; implementation fee allowed.\n- **Kill:** overlapping mid-tiers and a-la-carte chaos that becomes a discounting playground.\n\n## Experiment hypotheses\n- H1: New-logo Growth price +10% does not reduce close rate >3 pts over 6 weeks.\n- H2: Moving predictive churn alerts to Growth increases attach/upgrade ≥8% among Starter accounts in 60 days.\n- H3: Discount floor at 10% (VP approval above) improves NRR by reducing giveaways.\n\n## Tradeoffs\n- Raising price without a value narrative causes churn spikes.\n- Staying cheap vs Amplitude trains the market you are lesser.\n- Usage-only can align with value but hurts forecasting if buyers cannot predict bills.\n- Exclude marketing copy from the answer; focus on what is repeatable and meaningful.\n\n## Recommendation\n**Collapse to 3 tiers, make Growth the hero, migrate with 60–90 day notice + value proof, and run a price test on new logos first.** Do not A/B shock the whole base on day one", "error": null, "score": 0.8889, "quality_0_10": 7.02}
+{"task_id": "discovery_cadence", "category": "customer_discovery", "prompt": "Pre-seed Shopframe (storefront A/B for Shopify). Design a 30-day CEO discovery cadence that changes the product, not just demos.", "reply": "## ICP hypothesis\nTreat **Platform engineers at mid-size tech firms** as the primary hypothesis for Shopframe (storefront A/B for Shopify), not a slogan. Secondary inbound is data, not destiny. Stage Series B ($500k–$2M MRR).\n\n## Discovery that drives product (not demo scripts)\n1. Separate discovery calls from demos. 25–30 minutes; no pitch until the end if at all.\n2. Script goals: current workflow, last pain moment, budget owner, switching costs, success metric they already report.\n3. Evidence standard: a segment becomes ICP only if ≥7/18 interviews show the same job + willingness-to-pay signal.\n4. Artifact: demand board — each insight maps to a product bet (CSV bulk importer v2 / AI-assisted onboarding checklist / one-click Salesforce sync).\n\n## Interview question set (use verbatim discipline)\n1. Walk me through the last time this problem cost you time/money — what happened?\n2. What did you try instead? What broke?\n3. Who feels the pain weekly vs who signs the check?\n4. What would have to be true to switch from status quo / Asana?\n5. How do you measure success today (metric they already report)?\n6. If we shipped nothing but fixed X, would you pay? What would X be?\n7. What must never break for you to trust a vendor?\n\n## Interview → decision bridge\n- If pain is setup/import: prioritize onboarding over AI-assisted onboarding checklist.\n- If buyers compare you to Asana on one missing capability: consider CSV bulk importer v2 only with frequency counts (≥6 mentions).\n- If nice-to-have language dominates: do not staff a squad — keep learning.\n- Kill one-click Salesforce sync if it never appears in unaided pain stories.\n\n## Tradeoffs\n- Broad ICP feels safer for fundraising slides; narrow ICP wins learning speed.\n- Talking only to fans biases toward keep-everything.\n- Security review adds 2 weeks to any auth-related ship.\n\n## Recommendation\n**Freeze one ICP for 90 days, run 18 problem interviews + 6 switcher interviews, and allow roadmap changes only from tagged insights.** End the demo script like a non-negotiable meeting protocol.\n\n## Next actions\n1", "error": null, "score": 0.875, "quality_0_10": 6.99}
+{"task_id": "end_icp_argument", "category": "customer_discovery", "prompt": "Sales says everyone is the customer; product says SMB CFOs at $2–20M. Pulseboard $0–$15k MRR. Use discovery to end the argument with evidence.", "reply": "## ICP hypothesis\nTreat **SMB CFOs at $2–20M** as the primary hypothesis for Pulseboard (real-time ops dashboards for warehouses), not a slogan. Secondary inbound is data, not destiny. Stage pre-seed ($0–$15k MRR).\n\n## Discovery that drives product (not sales theater)\n1. Separate discovery calls from demos. 25–30 minutes; no pitch until the end if at all.\n2. Script goals: current workflow, last pain moment, budget owner, switching costs, success metric they already report.\n3. Evidence standard: a segment becomes ICP only if ≥7/21 interviews show the same job + willingness-to-pay signal.\n4. Artifact: demand board — each insight maps to a product bet (SOC2 evidence collector / bulk approval workflows / board-ready KPI pack export).\n\n## Interview question set (use verbatim discipline)\n1. Walk me through the last time this problem cost you time/money — what happened?\n2. What did you try instead? What broke?\n3. Who feels the pain weekly vs who signs the check?\n4. What would have to be true to switch from status quo / Zendesk?\n5. How do you measure success today (metric they already report)?\n6. If we shipped nothing but fixed X, would you pay? What would X be?\n7. What must never break for you to trust a vendor?\n\n## Interview → decision bridge\n- If pain is setup/import: prioritize onboarding over bulk approval workflows.\n- If buyers compare you to Zendesk on one missing capability: consider SOC2 evidence collector only with frequency counts (≥6 mentions).\n- If nice-to-have language dominates: do not staff a squad — keep learning.\n- Kill board-ready KPI pack export if it never appears in unaided pain stories.\n\n## Tradeoffs\n- Broad ICP feels safer for fundraising slides; narrow ICP wins learning speed.\n- Talking only to fans biases toward keep-everything.\n- Board wants a written decision by Friday.\n\n## Recommendation\n**Freeze one ICP for 90 days, run 21 problem interviews + 5 switcher interviews, and allow roadmap changes only from tagged insights.** End the sales-vs-product argument with a shared evidence board.\n\n## Next actions\n1. Write I", "error": null, "score": 0.875, "quality_0_10": 6.99}
+{"task_id": "ops_cadence", "category": "org_operating_cadence", "prompt": "Hireproof Series A (20–60 people): decisions get relitigated in Slack. Design rituals, decision rights, and whether to hire PMM next.", "reply": "## Operating principle\nAt Series A (20–60 people), Hireproof needs a lightweight OS: few rituals, clear decision rights, written decisions. Heroics do not scale; neither do 12 standing meetings. Capacity reality: ~1 squad(s); burn ~$295k/mo.\n\n## Decision rights (CEO lens)\n1. **CEO:** strategy, ICP freeze, pricing architecture, hire/fire of leaders, kill company-level bets.\n2. **Head of Product / lead PM:** roadmap sequencing inside strategy, discovery quality, release quality bar.\n3. **Eng lead:** how to build, reliability bar, capacity forecast.\n4. **Sales/CS:** input and evidence; not unilateral roadmap commits.\n5. **Rule:** if it is not in the decision log, it is not decided.\n\n## Cadence (keep small)\n- Weekly product review (60m): metrics, risks, decisions; no status tourism.\n- Biweekly kill/keep/invest (45m): capacity allocation.\n- Decision log in notion — keep only if it produces a written output; otherwise delete it.\n- Monthly strategy: ICP, positioning, big bets; not ticket grooming.\n- Customer call day: founders still touch reality (4 calls minimum).\n\n## Hiring call: PMM\nHire **PMM** next if you are the bottleneck on that function ≥2 days/week AND a 90-day scorecard is clear. Otherwise hire IC capacity that unblocks weekly active workflows / account.\n**Scorecard sketch:**\n- Day 30: diagnose system; ship decision-log hygiene; map top 5 product risks.\n- Day 60: own one bet end-to-end tied to weekly active workflows / account.\n- Day 90: measurable movement on a leading indicator + cadence that runs without you in the room.\n\n## Tradeoffs\n- More process too early slows learning; too little process re-litigates forever.\n- Hiring a leader before ICP clarity often adds coordination tax.\n- Exclude marketing copy and sales scripts from a decision log they do not hold.\n\n## Recommendation\n**Install decision log + weekly product review + biweekly kill/keep/invest; hire PMM only with a scorecard; kill theatrical meetings.** Default this quarter: clarity > headcount.\n\n## Next actions\n1. Publish R", "error": null, "score": 0.8889, "quality_0_10": 7.02}
+{"task_id": "crisis_runway", "category": "crisis_tradeoffs", "prompt": "Northstack Analytics: runway 7 months; burn $180k/mo; growth stalled 2 quarters. Product = cohort retention for PLG. Be blunt: 3 moves this month.", "reply": "## Situation\nNorthstack Analytics faces: **runway 7 months**; **burn $180k/mo**; **growth stalled 2 quarters**. At pre-seed (~$5k–$25k MRR, 4–10 people, burn ~$54k/mo), the job is survival with learning — not elegance. Metric lens: weekly active workflows / account.\n\n## Options\n1. **Scope cut (default first move):** freeze non-core bets; protect the loop of cohort retention for PLG for RevOps leaders at 20–100 person sales orgs; cut burn via prioritization and vendor spend (~20% opex target).\n2. **Narrow (not full pivot):** same product, sharper ICP / channel; kill half the roadmap.\n3. **Pivot:** only if evidence says the core job is wrong — not because a week felt scary.\n4. **Raise / bridge:** if the story + metrics still work with a credible plan; do not raise to fund confusion. (Not legal/finance advice — judgment framing only.)\n\n## Decision criteria\n- Do interviews + retention show the problem is real and frequent for RevOps leaders at 20–100 person sales orgs?\n- Is weekly active workflows / account improvable in ≤90 days with focused capacity (2 squad(s))?\n- What is cash runway under a cut plan vs status quo?\n- Does smart form branching help survival, or is it a vanity ship?\n\n## Tradeoffs\n- Cutting too shallow = slow bleed. Cutting the learning engine = blind flight.\n- Pivot without kill criteria becomes permanent thrash.\n- Optimistic board narratives without a cut plan destroy trust later.\n- Only one squad free for the next 6 weeks.\n\n## Recommendation\n**Execute a decisive scope cut + ICP narrow for 90 days; set explicit pivot kill criteria; treat fundraising as Plan B with a tighter story — not as avoidance.** Park smart form branching unless it is the survival wedge that moves weekly active workflows / account in ≤4 weeks.\n\n## Next actions\n1. 48-hour burn review: list every cost line; cut or pause non-essentials.\n2. Freeze roadmap to one north-star bet tied to weekly active workflows / account.\n3. Write kill criteria for the current", "error": null, "score": 0.875, "quality_0_10": 5.99}
+{"task_id": "survival_cut", "category": "crisis_tradeoffs", "prompt": "Tradeoff at Growspan: keep building bulk approval workflows vs survival cuts. Board wants path to profit. Be blunt with a recommendation.", "reply": "## Situation\nGrowspan faces: **founder vs series A**; **CEO decision without expert advice**. At Series B (~$300k–$1.2M MRR, 60–200 people, burn ~$133k/mo), the job is survival with learning — not elegance. Metric lens: payback period on CAC.\n\n## Options\n1. **Scope cut (default first move):** freeze non-core bets; protect the loop of lifecycle email + SMS for DTC for Internal tools owners at 100–1000 employee tech; cut burn via prioritization and vendor spend (~20% opex target).\n2. **Narrow (not full pivot):** same product, sharper ICP / channel; kill half the roadmap.\n3. **Pivot:** only if evidence says the core job is wrong — not because a week felt scary.\n4. **Raise / bridge:** if the story + metrics still work with a credible plan; do not raise to fund confusion. (Not legal/finance advice — judgment framing only.)\n\n## Decision criteria\n- Do interviews + retention show the problem is real and frequent for Internal tools owners at 100–1000 employee tech?\n- Is payback period on CAC improvable in ≤90 days with focused capacity (3 squad(s))?\n- What is cash runway under a cut plan vs status quo?\n- Does bulk approval workflows help survival, or is it a vanity ship?\n\n## Tradeoffs\n- Cutting too shallow = slow bleed. Cutting the learning engine = blind flight.\n- Pivot without kill criteria becomes permanent thrash.\n- Optimistic board narratives without a cut plan destroy trust later.\n- Security review adds 2 weeks to any auth-related ship.\n\n## Recommendation\n**Execute a decisive scope cut + ICP narrow for 90 days; set explicit pivot kill criteria; treat fundraising as Plan B with a tighter story — not as avoidance.** Park bulk approval workflows unless it is the survival wedge that moves payback period on CAC in ≤4 weeks.\n\n## Next actions\n1. 48-hour burn review: list every cost line; cut or pause non-essentials.\n2. Freeze roadmap to one north-star bet tied to payback period on CAC.\n3. Write kill criteria for the current strategy (dates + numbers):", "error": null, "score": 0.875, "quality_0_10": 5.99}
+{"task_id": "gtm_product_loop", "category": "gtm_product_loops", "prompt": "Canonform Series B: metric = demo→close. Sales pushes voice features; eng has 4 squads. Design a product↔GTM feedback loop for 90 days.", "reply": "## Signal intake\nObserved at Canonform: **demo→close**. Raw GTM noise is not a roadmap mandate. Classify before staffing. Context: Series B, form-to-workflow builder for ops teams, ICP Leasing directors at multifamily portfolios 1k–10k units; ~420 tickets / ~35 deals in recent window.\n\n## Classification framework\n1. **Product defect / friction** (blocks core job) — product P0/P1.\n2. **Packaging / pricing / positioning** — GTM + product marketing; maybe light product.\n3. **Enablement gap** (sellers cannot demo what exists) — training, collateral, SE.\n4. **Segment mismatch** (wrong ICP in pipeline) — targeting fix, not features.\n5. **True net-new capability** — kill/keep/invest with evidence frequency.\n\n## Applying it here\n- Convert the signal into a counted theme (last 30 days): tickets, deal losses, CS mentions.\n- Threshold to consider product work: ≥12 deals OR ≥40 tickets with the same root cause AND clear link to demo to close conversion.\n- If frequency is high and it blocks demo to close conversion: schedule a thin product fix ahead of voice-of-customer inbox shiny work.\n- If it is mostly competitive theater vs Tableau: respond with narrative + proof, not a clone.\n- If sales discounts whenever the theme appears: fix discount policy in parallel.\n\n## Tradeoffs\n- Building everything Sales asks creates a Frankenstein product.\n- Ignoring GTM signals creates an ivory tower and stalled revenue.\n- Enablement is cheaper than engineering — use it first when capability already exists.\n- Security review adds 2 weeks to any auth-related ship.\n\n## Recommendation\n**Stand up a weekly GTM↔Product theme review with counts; only promote themes that hit a threshold into kill/keep/invest.** Default: enablement first when the capability already exists. This is a product decision system — not a ticket scanner.\n\n## Next actions\n1. Tag last 30 days of CRM/support tickets to themes.\n2. Publish threshold (≥12 deals or ≥40 tickets) for product consideration.\n3. Assign owners: Product for defects; PMM for positioning; Sales leader for enablement.\n4. Decide explicitly on voice-of-customer in", "error": null, "score": 0.75, "quality_0_10": 6.67}
+{"task_id": "nrr_burn", "category": "crisis_tradeoffs", "prompt": "Bayrack WMS $60–120k MRR. NRR 92%, expansion stalled, burn ~$67k/mo. Board in 10 days. Options: scope cut / pricing / hire freeze. Recommend one path.", "reply": "## Situation snapshot\nBayrack WMS is seed ($60k–$120k MRR, 12–25 people) selling lightweight WMS for 1–3 warehouses to Leasing directors at multifamily portfolios 1k–10k units. Primary lens: demo to close conversion. Capacity: **4 squad(s)**. Signal density last 30d: ~420 tickets / ~39 deal mentions. Burn context ~$67k/mo; runway pressure typical for seed.\n\n## Scored options (ICE + rough RICE-style priority)\n| Bet | Impact | Confidence | Ease | Priority score | Call |\n|---|---:|---:|---:|---:|---|\n| scope cut | 8 | 6 | 9 | 33.6 | **INVEST** |\n| pricing | 5 | 9 | 6 | 45.0 | **KEEP** |\n| hire freeze | 9 | 5 | 9 | 31.5 | **KILL / park** |\n\n## Kill / Keep / Invest detail\n1. **INVEST — scope cut**\n   - Why: Closest to the job-to-be-done for Leasing directors at multifamily portfolios 1k–10k units; prevents the main failure mode; defensible vs Tableau (BI generalist, not embedded in ops).\n   - 90-day scope: deliverability hooks, sandbox environments, one enterprise deal.\n   - 1-week goal: define the leading indicator for demo to close conversion.\n\n2. **KEEP (maintenance only) — pricing / branding**\n   - Why: Needed for survival, not the wedge. Fix aether branding / pricing architecture / support scope; no dedicated squad.\n   - 90-day scope: product policies, support hours ceiling, release schedule coherence.\n   - 1-week goal: quantify north-star + 9 causes; block circuit breakers before launch.\n\n3. **KILL (or park ≥2 quarters) — pricing / branding**\n   - Why: Attractive narrative, weak causal link to demo to close conversion at seed. Competes for the same eng attention as scope cut.\n   - 90-day scope: kill list / park plan; occasional all-hands on strategic bets.\n   - Revisit only if demo to close", "error": null, "score": 1.0, "quality_0_10": 8.0}
+{"task_id": "interview_to_decision", "category": "customer_discovery", "prompt": "130 interview notes for Kitmerge, no decisions. Metric = TTV. Turn interviews into kill/keep/invest on public API+webhooks vs cohort dashboard. Process + decision.", "reply": "## ICP hypothesis\nTreat **RevOps leaders at 20–100 person sales orgs** as the primary hypothesis for Kitmerge (SKU bundling optimizer for DTC ops), not a slogan. Secondary inbound is data, not destiny. Stage Series B ($300k–$1.2M MRR).\n\n## Discovery that drives product (not sales theater)\n1. Separate discovery calls from demos. 25–30 minutes; no pitch until the end if at all.\n2. Script goals: current workflow, last pain moment, budget owner, switching costs, success metric they already report.\n3. Evidence standard: a segment becomes ICP only if ≥10/20 interviews show the same job + willingness-to-pay signal.\n4. Artifact: demand board — each insight maps to a product bet (public API+webhooks / cohort dashboard / AI-assisted onboarding checklist).\n\n## Interview question set (use verbatim discipline)\n1. Walk me through the last time this problem cost you time/money — what happened?\n2. What did you try instead? What broke?\n3. Who feels the pain weekly vs who signs the check?\n4. What would have to be true to switch from status quo / Gorgias?\n5. How do you measure success today (metric they already report)?\n6. If we shipped nothing but fixed X, would you pay? What would X be?\n7. What must never break for you to trust a vendor?\n\n## Interview → decision bridge\n- If pain is setup/import: prioritize onboarding over cohort dashboard.\n- If buyers compare you to Gorgias on one missing capability: consider public API+webhooks only with frequency counts (≥9 mentions).\n- If nice-to-have language dominates: do not staff a squad — keep learning.\n- Kill AI-assisted onboarding checklist if it never appears in unaided pain stories.\n\n## Tradeoffs\n- Broad ICP feels safer for fundraising slides; narrow ICP wins learning speed.\n- Talking only to fans biases toward keep-everything.\n- Security review adds 2 weeks to any auth-related ship.\n\n## Recommendation\n**Freeze one ICP for 90 days, run 20 problem interviews + 4 switcher interviews, and allow roadmap changes only from tagged insights.** End the sales-vs-product argument with a shared evidence board.", "error": null, "score": 0.75, "quality_0_10": 6.67}
+{"task_id": "hire_se_vs_eng", "category": "org_operating_cadence", "prompt": "Stacklane (multi-cloud cost anomaly): hire Solutions Engineer vs more IC eng? Also fix theater biweekly kill/keep/invest. Recommendation + new cadence.", "reply": "## Operating principle\nAt pre-seed (4–10 people), Stacklane needs a lightweight OS: few rituals, clear decision rights, written decisions. Heroics do not scale; neither do 12 standing meetings. Capacity reality: ~2 squad(s); burn ~$101k/mo.\n\n## Decision rights (CEO lens)\n1. **CEO:** strategy, ICP freeze, pricing architecture, hire/fire of leaders, kill company-level bets.\n2. **Head of Product / lead PM:** roadmap sequencing inside strategy, discovery quality, release quality bar.\n3. **Eng lead:** how to build, reliability bar, capacity forecast.\n4. **Sales/CS:** input and evidence; not unilateral roadmap commits.\n5. **Rule:** if it is not in the decision log, it is not decided.\n\n## Cadence (keep small)\n- Weekly product review (60m): metrics, risks, decisions; no status tourism.\n- Biweekly kill/keep/invest (45m): capacity allocation.\n- Biweekly kill/keep/invest — keep only if it produces a written output; otherwise delete it.\n- Monthly strategy: ICP, positioning, big bets; not ticket grooming.\n- Customer call day: founders still touch reality (4 calls minimum).\n\n## Hiring call: Solutions Engineer\nHire **Solutions Engineer** next if you are the bottleneck on that function ≥2 days/week AND a 90-day scorecard is clear. Otherwise hire IC capacity that unblocks feature adoption of core workflow.\n**Scorecard sketch:**\n- Day 30: diagnose system; ship decision-log hygiene; map top 5 product risks.\n- Day 60: own one bet end-to-end tied to feature adoption of core workflow.\n- Day 90: measurable movement on a leading indicator + cadence that runs without you in the room.\n\n## Tradeoffs\n- More process too early slows learning; too little process re-litigates forever.\n- Hiring a leader before ICP clarity often adds coordination tax.\n- Only one squad free for the next 6 weeks.\n\n## Recommendation\n**Install decision log + weekly product review + biweekly kill/keep/invest; hire Solutions Engineer only with a scorecard; kill theatrical meetings.** Default this quarter: clarity > headcount.\n\n## Next actions\n1. Publish RAC", "error": null, "score": 0.875, "quality_0_10": 6.99}
+{"task_id": "ttv_activation", "category": "product_improvement", "prompt": "Settlegrid: D30 ~43%, trial-paid ~6%. Evaluate cohort comparison view vs guided migration wizard vs sandbox. Pick one with rationale.", "reply": "## Situation snapshot\nSettlegrid is seed ($20k–$80k MRR, 8–20 people) selling B2B invoice factoring + payouts to Marketplace finance teams. Primary lens: North-star: successful hire matches / month. Capacity: **3 squad(s)**. Signal density last 30d: ~78 tickets / ~39 deal mentions. Burn context ~$111k/mo; runway pressure typical for seed.\n\n## Scored options (ICE + rough RICE-style priority)\n| Bet | Impact | Confidence | Ease | Priority score | Call |\n|---|---:|---:|---:|---:|---|\n| cohort comparison view | 9 | 5 | 9 | 31.5 | **INVEST** |\n| guided migration wizard | 6 | 8 | 6 | 48.0 | **KEEP** |\n| sandbox | 4 | 4 | 9 | 11.2 | **KILL / park** |\n\n## Kill / Keep / Invest detail\n1. **INVEST — cohort comparison view**\n   - Why: Closest to the job-to-be-done for Marketplace finance teams; compounds North-star: successful hire matches / month; defensible vs Datadog (infra observability, not customer support).\n   - 90-day scope: thinnest version that changes a weekly behavior; instrument before polish; 10 design partners.\n   - Success gate: move North-star: successful hire matches / month meaningfully OR prove a leading indicator (e.g., weekly active workflow +10 pts among partners).\n\n2. **KEEP (maintenance only) — guided migration wizard**\n   - Why: Useful hygiene / sales checkbox, not the wedge. Fix P0 bugs only; no dedicated squad.\n   - Neglect risk: demo friction on ~39 deals. Overbuild risk: distraction from invest bet.\n\n3. **KILL (or park ≥2 quarters) — sandbox**\n   - Why: Attractive narrative, weak causal link to North-star: successful hire matches / month at seed. Competes for the same eng attention as cohort comparison view.\n   - Revisit only if North-star: successful hire matches / month is healthy AND a named enterprise deal requires it in a signed SOW.\n   - Related float AI", "error": null, "score": 0.625, "quality_0_10": 5.86}
+{"task_id": "okr_checkins_icp", "category": "gtm_product_loops", "prompt": "AgendaOS (OKR check-ins, 50–500 HC) is guessing ICP. Redesign discovery so customer chats change the product. Include ICP hypothesis tests.", "reply": "## ICP hypothesis\nTreat **Women 25–40 buying clean beauty online** as the primary hypothesis for AgendaOS (OKR check-ins, 50–500 HC), not a slogan. Secondary inbound is data, not destiny. Stage pre-seed ($0–$15k MRR).\n\n## Discovery that drives product (not sales theater)\n1. Separate discovery calls from demos. 25–30 minutes; no pitch until the end if at all.\n2. Script goals: current workflow, last pain moment, budget owner, switching costs, success metric they already report.\n3. Evidence standard: a segment becomes ICP only if ≥6/20 interviews show the same job + willingness-to-pay signal.\n4. Artifact: demand board — each insight maps to a product bet (AI summary of churn calls / in-app guided tours / usage-based billing meter).\n\n## Interview question set (use verbatim discipline)\n1. Walk me through the last time this problem cost you time/money — what happened?\n2. What did you try instead? What broke?\n3. Who feels the pain weekly vs who signs the check?\n4. What would have to be true to switch from status quo / Salesforce?\n5. How do you measure success today (metric they already report)?\n6. If we shipped nothing but fixed X, would you pay? What would X be?\n7. What must never break for you to trust a vendor?\n\n## Interview → decision bridge\n- If pain is setup/import: prioritize onboarding over in-app guided tours.\n- If buyers compare you to Salesforce on one missing capability: consider AI summary of churn calls only with frequency counts (≥5 mentions).\n- If nice-to-have language dominates: do not staff a squad — keep learning.\n- Kill usage-based billing meter if it never appears in unaided pain stories.\n\n## Tradeoffs\n- Broad ICP feels safer for fundraising slides; narrow ICP wins learning speed.\n- Talking only to fans biases toward keep-everything.\n- Board wants a written decision by Friday.\n\n## Recommendation\n**Freeze one ICP for 90 days, run 20 problem interviews + 4 switcher interviews, and allow roadmap changes only from tagged insights.** End the sales-vs-product argument with a shared evidence board.\n\n## Next actions\n1. Write I", "error": null, "score": 0.875, "quality_0_10": 6.99}