Creator Diligence Audience Report

Creator Diligence Report — Audience-Verified (premium)

The premium tier of the diligence dossier: everything the cheap creator-diligence-report gives PLUS a verified third-party audience report — real audience demographics, geography, interests, audience brand-affinity, and a fake-follower / credibility score. For procurement, legal, and brand-side teams who need to prove a creator's audience is real and on-brand before a contract, not just vibe it.

This is the PAID tier: it pulls a full audience report via get_audience_report (25 credits, or $0 on a cache hit within 30 days — but still billed 25). That is ~10× the cheap profile-only report, so it must be a deliberate, confirmed choice — never a silent upgrade.

Read first: ${CLAUDE_PLUGIN_ROOT}/shared/conventions.md (tool schemas, credit prices, the nine conventions). This skill honors the Freshness Gate (${CLAUDE_PLUGIN_ROOT}/shared/freshness-gate.md), Refusal Recovery (${CLAUDE_PLUGIN_ROOT}/shared/refusal-recovery.md), and the credit-estimate discipline (${CLAUDE_PLUGIN_ROOT}/shared/credit-modes.md).

Profile-only, no paid audience pull → creator-diligence-report. A live quoted fee to evaluate → fair-price-brief. Competitor-relationship screening only → conflict-check. A whole roster of audience reports → roster-microsite-builder.

Inputs to collect

• Creator identifier (required) — handle, profile URL, Creatorland ID, or email. Map to one of the five identifier types. Also capture the platform (instagram / tiktok / youtube) and, if known, the stable source_user_id (it guarantees a cache hit and avoids a re-pull).
• Engagement context (optional) — the brand/campaign this is for; powers the on-brand "fit" framing.
• Competitor brands to screen (optional) — ask once.

Flow

1. Cost preflight (MANDATORY — this is a paid pull). Run check_audience_coverage { creators: [{ platform, handle|source_user_id }] } (free) first. State the result to the human and get an explicit go-ahead:

   This pulls a full audience report — premium, 25 credits (~$0.63), or free if cached within 30 days. Coverage shows: <cached (free) | fresh ≈25 credits>. Continue?

   Do NOT pull until the human confirms. If they decline, fall back to the free profile-only creator-diligence-report and say so explicitly.

2. Resolve the creator — get_creator_profile { identifier: { type, ... } } (1 credit). Capture identity, platforms, follower count/tier, audience geo, data freshness, and brand affiliations. Halt cleanly if the identifier does not resolve.

3. Pull the audience report — get_audience_report { platform, source_user_id | handle } (cache-first). Capture as_of {profile_updated} for the freshness stamp and the coverage block. If the report cannot be built (no audience data), say so and fall back to the profile-only dossier.

4. Compose the one-pager (no tool calls). Fixed section order; every section present even when a platform doesn't supply it (labeled, never blank). Lead with audience authenticity.

Deliverable

# Audience-Verified Diligence — @<handle>
_Creatorland Data · <date> · audience report as of <profile_updated>_
**Data freshness: <FRESH | AGING | STALE — re-verify before contracting>**
**Audience coverage: <Instagram: full | TikTok/YouTube: demographics + geo only>**

## 1. Audience authenticity  ← lead with this
**~<X>% suspicious/fake followers** (bot/fake-flagged) · **~<Y>% mass followers** (real but low-value — accounts that follow thousands) · credibility score <0–1> · class <normal/…>
Audience-type split: real <%> / suspicious <%> / mass-followers <%> / influencers <%>.
**Verdict vs floor:** 🟢 real audience / 🔴 below credibility floor — <one line>.
> Credibility & fake-follower % are **vendor-derived** (the audience-data
> provider's score), surfaced here — not a Creatorland computation.
> **fake-follower % counts suspicious/bot accounts only; mass-follower % is the
> separate figure — real people who over-follow (low-value, not fake). Never
> blend the two into one "fake" number.**
> Instagram-only today; for TikTok/YouTube this section reads "not available".

## 2. Identity & reach
Handle / platform(s) / followers & tier / verified gender & age / location /
real engagement rate (provider-measured).

## 3. Audience fit
Top geographies with shares (US / target-market weight called out) · age &
gender split · top interests · top audience brand-affinities (IG).
<If engagement context given: one line on on-brand fit.>

## 4. Brand history & conflicts
Paid-partnership brands (from the report's commercial posts, IG) + Creatorland
corpus affiliations. <If competitors named: per-brand verdict — "no overlap
found IN OUR CORPUS" / "⚠ affiliation with <brand>".>
Note: affiliations are corpus-derived + report-derived — not an exhaustive
deal history; absence ≠ no relationship.

## 5. Market position (vertical band context)
`query_market_intelligence` rate band for the creator's inferred vertical —
p25 $<x> · median $<y> · p75 $<z>. The creator's <tier> sits <framing>.
> <provenance line verbatim> · recency window: <window>
Corpus-level band for the vertical/tier — NOT this creator's rate history.

## 6. Provenance & freshness
Audience data **as of <profile_updated>**; deal data derived from the
Creatorland corpus. Credibility score is vendor-derived (labeled). PII excluded
by design — no contact info in this report.

---
**Provenance:** Creatorland Data MCP · 1 profile + 1 audience report
(<cached / fresh>) <+ 1 rate benchmark> · <date>
**Credits used:** ~<N> (1 profile + 25 audience report <+ 5 rate>)

5. Offer the artifacts. Default deliverable is the clean markdown one-pager. Then offer: (a) the report's whitelabeled PDF (the pdf_url the tool returned, when present), and/or (b) a .docx / .pdf render of the full one-pager for procurement filing.

Honesty rules

• Cost is confirmed, never silent. The 25-credit pull happens only after the human says yes to the preflight; declining falls back to the free profile-only diligence.
• Lead with authenticity. Section 1 is first and carries the verdict — surface fake-follower % (suspicious/bot) and mass-follower % (real but low-value) as two distinct figures, never one blended "fake" number.
• Credibility is vendor-derived — labeled. Never present the score as a Creatorland computation (it is the audience-data provider's).
• Instagram-first honesty. For TikTok/YouTube, the credibility and audience brand-affinity sections read "not available for this platform today" — never fabricated.
• Band ≠ their rate. Section 5 framing is fixed.
• PII invariant (convention 7). No contact info appears, regardless of source — including the report's own contacts field (which the MCP already strips). Scrub bio snippets.
• Stale leads. Freshness + as_of are in the header, not buried.

Credit footprint

~26–31 credits per creator: 1 profile + 25 audience report (or 1 + free on a cache hit, billed 25) + optional 5 rate benchmark. Always run the free check_audience_coverage preflight first; a cached creator is the same 25 but $0 vendor cost. State the estimate before pulling.