Wrap Report Skeleton

Wrap Report Skeleton

Campaigns end in a wrap deck; agencies are paid in decks. This skill takes the finished campaign's creator roster + what was actually paid, refreshes each creator's data, frames spend against the market band, and scaffolds the wrap deck — with honest, labeled placeholders where the customer's own performance metrics slot in (the MCP has no campaign-performance data, and the skeleton never pretends otherwise). For agency account leads and brand-side managers.

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

Inputs to collect

• Campaign roster (required) — the creators who ran, in any format (handles, a sheet, a prior shortlist from this session). Map each row to one identifier type.
• What was paid (required for the benchmark framing) — per-creator fees or a total + rough split; "undisclosed" rows simply skip the per-creator framing line.
• Campaign vertical + deal type — infer from context if a brief or shortlist exists in-session; otherwise ask once.
• Customer's performance numbers (optional, can come later) — if supplied now, they fill the placeholder slots; if not, slots stay labeled.
• Credit mode — default thorough; thrifty on the standard triggers.

Flow

1. Re-profile the roster — for each roster creator: get_creator_profile { "identifier": { "type": "social_handle", "platform": "<platform>", "handle": "<handle>" } } (exactly one identifier type per call, whichever the roster provides). Capture CURRENT follower count, audience geo, freshness, and brand affiliations (pro) — the wrap deck shows where each creator is NOW vs. campaign start, and flags any new competitor affiliations acquired since. 1 credit × N. Credit estimate fires here when N + benchmark calls exceed ~30 credits (a 25-creator roster does): state the estimate and offer thrifty before fanning out. Thrifty: profile only the top-spend creators (those covering ~80% of budget) and say how many were skipped.

2. Freshness Gate — classify each re-profiled creator. In a wrap context stale means "current-state claims (follower count, affiliations) may have drifted" — flag those rows rather than excluding (the creator already ran; exclusion makes no sense post-campaign).

3. Benchmark the spend — one query_market_intelligence per distinct vertical/deal-type in the campaign (usually one): { "mode": "rate", "vertical": "<campaign vertical>", "deal_type": "<if known>", "creator_tier": "<the ran creators' tier, if they share one>" } — wrapped in Refusal Recovery (thorough: full ladder; thrifty: max 2 rungs), clearance level disclosed. When the creators who ran share a tier, pass creator_tier (emerging <1k / nano 1k-10k / micro 10k-100k / mid 100k-500k / macro 500k-1M / mega 1M+) so the spend is benchmarked against same-size creators; when tiers are mixed keep the vertical-wide band as the fallback and say so (a too-thin tier auto-broadens, disclosed). This powers "you paid $X vs market median $Y" — corpus band vs actual spend, the wrap deck's most defensible slide. 5 credits per call incl. ladder retries. If a quoted per-creator fee should be positioned precisely, add "quoted_rate": <USD> on a dedicated call (thorough mode only, and only when the user asks for per-fee positioning).

4. Scaffold the deck (no tool calls) — assemble the skeleton below. Performance sections are placeholder slots unless the customer supplied numbers; never fill them from the corpus, memory, or the web.

Deliverable

A wrap-deck skeleton in markdown (slide-per-section, paste-into-deck):

# Campaign Wrap — <campaign/brand name>
_<flight dates> · <N> creators · scaffolded with Creatorland Data, <date>_

## Slide: Campaign summary
❏ Objectives (from your brief) · ❏ flight dates · ❏ total spend: $<total>

## Slide: The roster, then and now
| Creator | Platform | Followers (current) | Audience geo (top) | Freshness | New affiliations since campaign |
|---|---|---|---|---|---|
| @<handle> | <p> | <n> | <geo %> | fresh/aging/⚠ stale | none found in corpus / ⚠ <brand> |
<one row per creator; stale rows carry the drift caveat inline>

## Slide: Spend vs market
Market band for the **<vertical>** vertical<scope note if broadened>:
p25 $<x> · median $<y> · p75 $<z>.
Your average per-creator spend: $<avg> — <below/at/above> the corpus median.
<per-creator framing lines where fees were supplied>
> <provenance line verbatim from the tool> · recency window: <window>
This compares your spend to the corpus-level market band for the
vertical/tier — it is not any creator's personal rate history.

## Slide: Performance                       [placeholder — your data]
❏ Impressions · ❏ engagement · ❏ CTR/conversions · ❏ EMV — supply from your
analytics tools. The Creatorland Data MCP has no campaign-performance data;
these slots are deliberately left labeled rather than estimated.

## Slide: Learnings & next campaign
❏ What worked (yours) · suggested data follow-ups: re-run lookalikes off the
top performer (`search_creators` lookalike mode) for the next flight.

---
Data freshness: <N>/<M> creators synced within the last sync window; <K>
flagged — current-state figures may have drifted.
Benchmark basis: <clearance level; if broadened: "a privacy feature of the
data source, not missing data">.
Provenance: Creatorland Data MCP · <M> profiles + <R> rate benchmark(s) · <date>.
Credits used this run: ~<N> (<M> profiles ×1 + <R> market-intel ×5).

Honesty rules

• The skeleton never fakes performance. No corpus source for campaign results exists — placeholder slots stay labeled, never estimated.
• Spend-vs-band framing is corpus-level. "Market median for the vertical", never "what these creators usually charge".
• Drift is disclosed, not hidden. Stale rows are flagged inline (they can't be excluded post-campaign).
• New-affiliation flags are corpus-derived — "none found in our corpus", never "no new deals".
• Disclose every Refusal Recovery broaden at the clearance level.
• PII invariant (convention 7). No contact info in the deck, regardless of source.

Credit footprint

thorough: ~(N + 5–10) credits for an N-creator roster (N profiles + 1–2 rate calls; e.g. 10 creators ≈ 15–20) · thrifty: ~(0.6N + 5) credits (top-spend profiles only + 1 rate call, max 2 ladder rungs). Estimate stated up front whenever the plan exceeds ~30 credits.