customer-pain-mining

Customer Pain Mining

Competitors at the early stage aren't a threat — they're a proxy for the customer. Their unhappy users have already done your custdev calls. This skill harvests them.

The goal is verbatim wording, not paraphrase. Founders summarize and lose the gold. Pull exact phrases — they become product copy.

When this skill applies

• A founder asks what users dislike about competitors or the category
• Preparing custdev calls — you want talking points in the customer's own language
• Writing landing-page copy or positioning — you need real pain wording
• Hunting white-space features competitors aren't shipping
• After competitor-discovery — going deeper on the named players

What you need

1. Competitor list — 1 to 5 named competitors (run competitor-discovery first if there isn't one)
2. Niche / category — one line for context (e.g. "AI writing for students")
3. Use case for the output — affects source weighting:
   • Custdev prep → Reddit (long-form verbatim wins)
   • Ad copy → Exa for short blog-post pull-quotes
   • Product strategy → Exa for structured "Pros / Cons" review pages

Tools

Anysite MCP:

• mcp__claude_ai_Anysite__execute(source="reddit", category="search", endpoint="search_posts", params={...}) — Reddit posts by query (anonymous, viral, long-form).
• mcp__claude_ai_Anysite__execute(source="linkedin", category="search", endpoint="search_posts", params={"keywords": "<pain-phrase>", "sort": "relevance", "count": 15}) — LinkedIn posts BY professionals describing the pain. Critical: search by the PAIN, not the competitor name (see Step 5).
• mcp__claude_ai_Anysite__query_cache(cache_key, conditions, ...) — filter cached results (e.g. vote_count > 50).
• mcp__claude_ai_Anysite__get_page(cache_key, offset) — paginate.
• mcp__claude_ai_Anysite__execute(source="youtube", category="video", endpoint="video_comments", params={"video": "<id>", "count": 50}) — comments under a competitor-review YouTube video; people who watched a 15-min review have committed to evaluating the product. Filter like_count >= 5. Watch for affiliate astroturf (5+ short identical-tone praises from different accounts).
• mcp__claude_ai_Anysite__execute(source="twitter", category="search", endpoint="search_posts", params={"query": "<niche+pain>", "min_likes": 5, "count": 20, "language": "English"}) — viral short-form pain quotes. Set min_likes ≥ 50 for breakthrough quotes; otherwise filter aggressively for relevance because the query catches handle namesakes (e.g. "Bright" matches users named "Bright").