Competitor Deep Research Sop

Competitor Deep Research (SOP v1.3.1)

You guide pain-first, evidence-backed competitor research that produces actionable outputs: not only analysis, but Tier-1 reachable contacts, outreach angles, and ongoing monitoring.

This skill is the agent-executable layer on top of the full written standard:

• Full text: templates/competitor-ux-research-framework-v1.3.1.md (930 lines — channels, matrices, templates, A-tier lite mode).

If another skill in the workspace is named competitive-analysis, distinguish them:

| Skill | Primary lens | |-------|----------------| | competitive-analysis | Positioning, capability grid, growth/SEO signals, strategic brief | | competitor-deep-research-sop (this) | User pain × seed users × 12-layer evidence × M1×M2 matrix × Tier-1 list |

Pairing: run this for “who hurts & where”; run competitive-intel-mining (if present) to operationalize community mining; run competitive-analysis for product/GTM framing.

---

When to Use

• Single competitor deep dive (S-tier style) with reproducible methodology.
• Need M1 × M2 pain matrix, S1–S3 seed signals, and Tier-1 outreach list.
• Must prove channel coverage (12 layers) with sub-channel hits (v1.3.1).
• Building monitoring (RSS, alerts) after the research spike.

When Not to Use

• Quick positioning memo only → use competitive-analysis or a short brief.
• Only “mine GitHub issues for leads” → use competitive-intel-mining-style narrow workflow.

---

Non-Negotiable Principles

1. No official-story-only: force user-side evidence across 12 layers (see full doc §3).
2. Every pain item: tag M1 (who hurts) × M2 (pain type) when evidence supports it.
3. Sources: web links or identifiers; no fabricated metrics (DAU, revenue).
4. Outputs must ship: pain report structure + Tier-1 list + ≥2 outreach hooks + monitoring sketch.

---

Concept Quick Reference (memorize direction; details in full doc)

14 dimensions

| Bucket | IDs | Role | |--------|-----|------| | Dual axis | M1 users · M2 pain types | Build 6×10 heat-style mapping | | Themes | M3–M9 | Evidence path, business model, trade-offs, onboarding, hype gap, churn/migration, feature gaps | | Signals | S1 anger · S2 alt-seeking · S3 self-rescue | Feed Tier-1 prioritization (S3 first) |

12 layers (L1–L12)

Code hosts → forums → long-form/newsletters → social → review aggregators → search → app stores → Q&A → hiring → finance/disclosures → product-native communities → awesome lists. v1.3.1: each layer lists sub-channels; coverage and hit rules are in full doc §3 + §3.x hit rules.

---

Execution Phases (map to full doc §5)

Advance phase-by-phase; do not skip Phase C–E if the user asked for a full S-tier deliverable.

Phase A — Charter (§5 stage 1)

Output a scope block:

## Charter
- Competitor(s):
- Tier intent: S full | A lite | custom
- Decision this informs:
- Owner / date:
- Output paths (adapt to user repo): report file, monitoring file, matrix source

Phase B — Product snapshot (§5 stage 2)

Deliver at top of report:

• One-line positioning + short description
• GitHub stats if OSS (stars, contributors, rough age, maintainer handles)
• Domain / entry points list
• Audience types (map toward M1)
• Who they sell to (commercial target ≠ generic audience)

Phase C — Evidence sweep — 12 layers (§5 stage 3)

• Work layer-by-layer; record sub-channel hits (v1.3.1 granularity).
• Minimum discipline from SOP: strong L1/L3/L6/L11 coverage when applicable; never skip L1 + L3 + L11 without explicit rationale (see full doc §3 rules).
• Maintain appendix evidence list: URL, quote stub, date, proposed M1×M2 tags.

Phase D — Pain mapping (§5 stage 4)

1. M1×M2 matrix: aggregate counts or severity per cell (🔴/🟠/🟡/empty); call out top 3 hot cells.
2. M4–M9: fill theme sections (business model, trade-offs, onboarding, hype gap, flows, feature backlog signals).
3. S1–S3: classify leads; elevate S3 (forks, PRs, workarounds) to Tier-1 priority.
4. Orphan bucket: evidence that fits no cell → park in §orphan; if ≥3 similar orphans, flag as candidate new dimension (do not rewrite framework ad hoc).

Phase E — Comparison grid (§5 stage 5)

If multi-competitor: fill comparison table — at minimum business model + self-host / deployment posture + dimensions user chose; add “differentiation insight” paragraph per table.

Phase F — Seed pipeline (§5 stage 6)

• Tier-1 list: P0 5–10 with handles + why (signal type).
• ≥2 outreach angles (technical pain vs onboarding pain vs self-rescue).

Phase G — Playbook for “your product” (§5 stage 7)

• 5 insights, each containing an explicit “avoid X’s trap” style lesson where possible.
• 4-step seed motion (week 1 / week 2 / month 1 / ongoing).
• Monitoring: at least one hook each for L1-class and L6-class signals (RSS, alerts, saved searches — see full doc §8).

Phase H — Wrap-up (§5 stage 8)

• Short insights file (bullet memory).
• Update competitive matrix / facts if this workspace keeps them.

Phase I — External framework check (§5 stage 9, quarterly optional)

Optional: Mom Test, JTBD, Kano, PMF survey — one paragraph each on gaps.

---

Report Skeleton (must match §6 in full doc)

Use this outline for the main deliverable:

1. Snapshot (positioning, GitHub, URLs, audiences)
2. TL;DR + M5 trade-offs (≥3)
3. Method — channel coverage table (L1–L12 with sub-channel hits)
4. Pain — M1×M2 matrix + top-3 cells deep dives
5. Themes — M4–M9 sections
6. Signals — S1/S2/S3 + Tier-1 list
7. Playbook — implications + monitoring + next steps
8. Appendices — evidence chain, templates used

Evidence citation style (per full doc):

> "quote"
> — [source](url) · reactions if any
> **Tags**: `M1 …` × `M2 …` · S1/S2/S3 …

---

A-Tier Lite Mode

If user selects A-tier (cross-layer watch, less depth):

• Follow §9 in templates/competitor-ux-research-framework-v1.3.1.md: reduced dimension set, drift tracking table, shorter report shell.
• Still keep L1/L3/L6/L11 signal coverage unless explicitly scoped down.

---

Quality Gates (before calling research “done”)

• [ ] Every 🔴/🟠 pain has M1×M2 or explicit “uncategorized” with orphan note
• [ ] 12-layer table shows which sub-channels hit (or “N/A” with reason)
• [ ] Tier-1 list has ≥5 contacts or explicit “insufficient public identity — next steps”
• [ ] Monitoring section has actionable subscriptions/alerts, not vague “watch Twitter”

---

Related Skills / Assets

• Full methodology: templates/competitor-ux-research-framework-v1.3.1.md
• Pair with seed tooling in same repo (e.g. scrapers, CSV enrichment) when moving from list → email.

---

Maintainer Notes

• Product names: user may say Nexu / internal codenames — keep output product-agnostic unless scope says otherwise.
• Language: this SKILL.md is English; use SKILL.zh-CN.md for Chinese-first execution.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Celina-create
• Source: Celina-create/X-Studio
• License: MIT

Install and usage instructions live in the source repository linked above.