competitor-research

Map a product's competitive landscape — direct + adjacent competitors, their positioning, naming conventions, and white-space. Persists findings to a per-product competitors file so it compounds over

SKILL.md

GitHub

competitor-research

When you activate

  • Before naming a product (this is phase 1 of name-product — it seeds themes and the collision baseline).
  • Before positioning, pricing, or GTM.
  • Whenever the user asks "who else does this", "competitors", "landscape", "who are we up against".

What you produce

  1. A competitor table — direct + adjacent + incumbents.
  2. A naming-conventions read — what patterns the category uses (to fit or to break).
  3. The white-space — positioning angles nobody owns.
  4. A persisted file: products/<slug>/competitors.md (create/update, dedupe) so the picture compounds across sessions.

Protocol

  1. Frame. One line: what the product does · its category · its ICP. (Pull from the product config / positioning brief if it exists.)
  2. Find, three rings (WebSearch, a few angles each):
    • Direct — same job-to-be-done for the same user.
    • Adjacent — overlapping capability or audience (these still cause name collisions).
    • Incumbents — the big, generic players everyone compares against.
  3. For each competitor, capture: name · URL · one-line what-they-do · positioning angle · naming pattern (coined? descriptive? founder-name? compound? suffix?).
  4. Synthesize:
    • Naming conventions in the space — and whether to conform (legibility) or contrast (distinctiveness).
    • Positioning white-space — angles/wording nobody owns.
    • The collision baseline — the exact names/roots a new name must avoid (hand this to name-clearance).
  5. Persist. Write/merge products/<slug>/competitors.md with a dated section. Dedupe against prior entries; note what's new.
  6. Machine-readable twin (added for market-xray, 2026-08-16). Also write/refresh products/<slug>/research/competitors.csv — columns name,url,type,notes (type: direct/adjacent) — the same deduped set. market-xray's stage 2 reads it to build targets.json after the operator prunes; standalone runs of this skill keep it current too.

Output format

Markdown: a competitor table, then three short lists — naming patterns, white-space, names/roots to avoid. Keep it scannable. Cite URLs.

Notes

  • This is a living asset — each run appends and refines, so coverage improves over time.
  • Run it before generating names: it both inspires themes and defines what must be cleared.