Synthetic buyer-persona builder — 4–6 personas with skepticism profile

A Claude Code skill that researches a target company (homepage, product pages, pricing, "who it's for", reviews on G2 / Capterra / TrustRadius, "[company] vs" comparison queries, job posts) and builds 4–6 synthetic buyer personas. Each persona is a specific individual — name, title, company profile, demographics, situation paragraph, ranked pain points, buying trigger, ranked decision criteria, skepticism profile (trust level, research style, key objections), language they use (how they describe the problem, what they search for, red-flag words, trust signals), and evaluation behavior (first visit, deep eval, deal breakers). Output is saved as personas.json (machine-readable), personas.md (prose), and segments.md (summary table), so sibling skills — landing-page CRO, copywriting, popup CRO, A/B test setup — can load personas directly.

The "language they use" and "skepticism profile" dimensions are the load-bearing differentiators. A traditional persona says "Marketing Manager, 30–45, cares about ROI"; this skill says "describes the problem as 'we need to consolidate our toolchain', red-flag words are 'revolutionary' and 'AI-powered', deal-breakers are SOC2 missing and no public changelog." That is directly actionable for copywriters, ad teams, and salespeople — and feeds downstream skills as JSON, not as a slide. Diversity rule is enforced: at least one technical buyer, one business buyer, one skeptical profile, one junior/researcher. Output is a reusable client asset, not a one-off workshop deliverable.

1. Phase 1 — Company research. WebFetch homepage + product / solutions / pricing / "who it's for" pages. WebSearch for "[company] customers", "[company] case studies", "[company] reviews", "[company] vs", "[company] jobs". Extract: problem solved, pricing / ACV signals, industries served, company sizes, roles named in case studies, GTM motion (self-serve / sales-led / hybrid).
2. Phase 2 — Identify 4–6 distinct segments. For each: segment name, role / titles, company profile (size, stage, industry, tech stack), core pain, buying trigger, decision criteria ranked, sophistication, alternatives considered, size estimate (primary / secondary / emerging). Apply diversity rules: ≥1 technical buyer, ≥1 business buyer, ≥1 skeptical profile, ≥1 junior / researcher.
3. Phase 3 — Build synthetic personas as JSON. One persona per segment (1–2 for the largest). Each persona populates the full schema: id, name (specific, not "Marketing Manager"), segment, title, company, demographics, situation (one paragraph), pain_points (3+), buying_trigger, decision_criteria_ranked, skepticism_profile, technical_sophistication, language, evaluation_behavior.
4. Phase 4 — Save artifacts. Write personas.json, personas.md, segments.md. Sibling skills load personas.json directly — keep the schema stable.
5. Output the summary table and coverage check. Segment overview table + 2–3-sentence summary per persona + coverage check (did we hit one technical, one business, one skeptical, one researcher?) + next-step suggestions (e.g. "run the landing-page audit against persona P3's skepticism profile").
6. Re-run on quarterly cadence. Personas drift as the target company's GTM shifts. Refresh quarterly or when material product / pricing changes.

Verbatim persona schema fragment from the skill (the two most load-bearing fields):

{
 "skepticism_profile": {
 "trust_level": "Low — has been burned by vendor promises before",
 "research_style": "Deep dive. Reads docs, checks GitHub issues, asks peers in Slack communities",
 "key_objections": [
 "Will this actually scale or will we outgrow it in a year?",
 "What's the real implementation cost beyond the license?",
 "How good is the support when things break at 2am?"
 ]
 },
 "language": {
 "describes_problem_as": "We need to consolidate our toolchain and reduce operational overhead",
 "searches_for": [
 "engineering productivity platform",
 "developer tools consolidation",
 "[competitor] alternative enterprise"
 ],
 "red_flag_words": ["revolutionary", "AI-powered", "seamless"],
 "trust_signals": ["SOC2 badge", "customer logos in their industry", "transparent pricing", "public changelog"]
 }
}

Install (verbatim — keep distributor path runnable):

npx gooseworks install --claude
# alternates: --cursor, --codex, --all

Trigger phrasing (verbatim from skill):

Build buyer personas for [company URL]
Generate synthetic personas based on [company]'s actual customers

• Generic-archetype trap. "Marketing Manager" alone fails the skill's specificity check. Each persona needs name, budget authority, recent buying trigger, and concrete deal breakers — otherwise downstream skills produce generic copy.
• Missing skepticism profile. Skepticism is the most-skipped and most-valuable layer; every persona needs an explicit "what would make them NOT buy." Without it, the copy review skill has nothing to push back on.
• Real customer names in JSON. The skill produces synthetic personas — don't paste real customer data into personas.json. Synthesize from public review data, don't transcribe.
• One-and-done. Personas drift; an annual workshop pretends they don't. Refresh quarterly or when product / pricing changes materially.