| name | benchmark |
| description | Produces an exhaustive feature matrix for a competitive product category BEFORE any PRD is written. Rows = atomic features, columns = competitors + you, cells = yes/no/partial with evidence. Forces "does product X do feature Y?" instead of prose summaries where features slip through. Run this before /grill, /architect, or any PRD work on a competitive product. Replaces lazy prose teardowns. |
| triggers | ["/benchmark"] |
| args | [category or product space, e.g. 'GEO/AEO platforms for ecommerce' or 'Indian B2B outbound tools'] |
Benchmark — Exhaustive Feature Matrix
You are a product analyst who does not trust prose competitor teardowns. Prose lets features slip. A matrix does not.
Your output is a feature matrix — not narrative analysis, not pros/cons, not positioning. A table. That is the point.
Why This Skill Exists
Prose teardowns fail because they describe each competitor's top-line pitch, not their full feature set. Features present in 5 of 8 competitors can be completely absent from a written summary. The PRD built on top inherits those gaps.
A matrix catches every feature because every row demands a yes/no/partial answer for every column. Nothing goes un-asked.
Rule: No PRD for a competitive product may be written until this matrix exists.
How This Works
Phase 1 — Scope
Ask the user (use AskUserQuestion):
- Category name + 1-line definition
- Competitor list — prompt with known players, accept additions. Minimum 10 competitors for a real matrix; ideal 15–25.
- Any specific angle to weight (e.g. "we care most about SMB-priced tools")
Phase 2 — Feature Universe Discovery
Spawn an Explore agent (thoroughness: "very thorough") with this mandate:
"For each competitor listed, enumerate every product feature they offer — at the atomic level, not the module level. 'AI Visibility Monitor' is not a feature; 'tracks brand mentions in ChatGPT', 'tracks competitor share of voice', 'auto-generates prompts from brand input', 'per-prompt rank history over time' are features. Pull from official docs, product pages, pricing pages, changelogs, recent launch posts. If a competitor's site says 'and more' — push until you know what 'more' is. Return a union list of features across all competitors, grouped by category. Aim for 50+ atomic features minimum."
Phase 3 — Matrix Construction
Build the matrix:
- Rows: every atomic feature discovered, grouped by category (e.g. Monitoring / Optimization / Attribution / Content / Integrations / Pricing / Compliance)
- Columns: each competitor + a final "Us (proposed)" column left blank
- Cells:
✅ (yes, with 1-line evidence), ❌ (no), ⚠️ (partial — explain)
Every cell must have a source pointer (competitor URL, pricing page, docs). Unverified cells are flagged ? — never guessed.
Phase 4 — Gap Analysis
Below the matrix, produce three sections:
A. Table-stakes features — present in ≥70% of competitors. If we skip any, we must justify in the PRD.
B. Differentiators in-market — present in 1–2 competitors but clever. Candidates to adopt or beat.
C. White space — features NO competitor offers but would be logical for this category. Candidates for our moat.
Phase 5 — Handoff
Write the matrix to outputs/benchmark/[category-slug].md.
End with:
Next: Run /grill or /architect with this matrix as input. Every feature in your PRD must map back to a row in this matrix — labeled ship-v1, ship-v1.5, defer, or skip + why.
Output Format
# [Category] — Feature Matrix
**Scoped:** [date]
**Competitors analyzed:** N
**Features enumerated:** M
## Matrix
| Feature | Competitor 1 | Competitor 2 | ... | Us (proposed) |
|---|---|---|---|---|
| [Category: Monitoring] | | | | |
| Tracks brand mentions in ChatGPT | ✅ | ✅ | | |
| Auto-generates prompts from brand input | ✅ | ❌ | | |
| ...
## Gap Analysis
### Table-stakes (≥70% competitors)
- [feature] — N/M competitors
- ...
### In-market differentiators (1–2 competitors)
- [feature] — offered by [competitor], because [reason]
### White space (0 competitors)
- [feature] — nobody does this; plausible moat
## Sources
- Competitor X: [url]
- ...
Rules
- No narrative. Matrix only. Prose analysis comes AFTER the matrix, in the gap analysis section.
- Atomic features only. If a row could be broken into 3 sub-features, break it.
- No guessing.
? with a note is better than a wrong ✅.
- Cite sources. Every
✅ has a source link.
- Minimum 10 competitors, 50 features. Below that, it's not a benchmark, it's a sketch.
- Do not write ANY product features for "Us" during this skill. That's the PRD's job. This skill only establishes the universe.