Vertical Briefing

Quarterly Vertical Briefing

The recurring leadership artifact: a one-to-two-page "state of creator pricing in , " memo built from corpus aggregates — deal volume, value bands, deal-type mix, recency signals — every number carrying its provenance. Written for a CMO/VP audience that will forward it without you in the room, so it must be defensible by construction. This is a point-in-time portrait; if the user wants change-over-time ("what's moved since last quarter"), that's vertical-forecast-brief.

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) and Refusal Recovery (${CLAUDE_PLUGIN_ROOT}/shared/refusal-recovery.md). No creator lists are produced, so the Freshness Gate does not fire; recency comes from the market-intel responses themselves.

Inputs to collect

• Vertical (required) — e.g. "Beauty". If the user runs this for their brand, infer from the brand's category; confirm once, never re-ask on recurring runs.
• Period label (default: current quarter) — cosmetic header label; also sets active_since to the quarter start for the recency-focused calls.
• Sub-categories of interest (optional, max 3) — e.g. "skincare, fragrance". Each adds calls; warn on cost in thorough mode.
• Audience (default: CMO/leadership) — shapes tone, not data.
• Mode — default thorough; thrifty on request.

If this is a recurring run (the user has run it before this session or names a cadence), reuse all prior inputs without asking; offer to set up a schedule via the harness's scheduling facility if one exists.

Flow

All market-intel calls are wrapped in Refusal Recovery (market-mode floor: 5 brands / 25 deals; rate-mode floor: 10 brands / 50 deals). 5 credits each, including each ladder retry.

1. Vertical landscape — query_market_intelligence { mode: "market", vertical: <vertical> } → deal counts/distributions, deal-type mix, company-type spread.

2. Recency slice — query_market_intelligence { mode: "market", vertical: <vertical>, active_since: "<ISO quarter start>" } → how active the vertical has been this period. If this slice refuses, ladder per the protocol (here active_since is the narrowest constraint — widen the window one step, e.g. quarter → trailing 6 months, and disclose) and label the section with the actual window that cleared.

3. Pricing bands — query_market_intelligence { mode: "rate", vertical: <vertical> } → p25/median/p75 for the vertical. In thorough mode, add one { mode: "rate", vertical: <vertical>, deal_type: <dominant type from step 1> } for the headline deal type. Optionally add creator_tier (emerging <1k / nano 1k-10k / micro 10k-100k / mid 100k-500k / macro 500k-1M / mega 1M+) to get a tier-specific band; omit it (the default) for the vertical-wide band. A tier too thin for the privacy floor auto-broadens (disclosed) to the all-tier band.

4. Sub-category cuts (thorough only) — for each requested sub-category: { mode: "market", vertical: <vertical>, sub_category: <sub> }. Expect thin slices; on refusal take at most 2 ladder rungs then report the sub-category as "below the privacy floor — reported at vertical level" rather than walking to numbers that no longer describe the sub-category.

5. Compose the memo (no tool calls). Pre-run estimate fires before step 1 if sub-categories push the plan over ~30 credits.

Thrifty: steps 1 + 3 only (2 calls, ~10 credits), no sub-categories, max 2 ladder rungs anywhere.

Deliverable

# State of creator pricing — <Vertical>, <Period>
*Prepared <date> · source: Creatorland deal corpus*

## Headlines (3 bullets)
- <volume headline with number>
- <pricing headline: median + band>
- <mix/recency headline>

## Market activity
<Deal volume, company-type spread, deal-type mix — each stat as a
citation-ready block: stat + the provenance line the tool returned +
recency window (convention 1).>

## Pricing
Vertical band: p25 $<x> · median $<y> · p75 $<z>
<Headline deal type band if pulled.>
These are corpus-level market bands for the vertical — planning context,
not any individual creator's rate.

## This period's activity
<active_since slice — with the actual window that cleared the floor.>

## Sub-category notes            [thorough, if requested]
<Per sub-category: numbers at the level that cleared, or the honest
"below privacy floor" note.>

## What this means for <brand/team>
<2–3 implications. Label every forward-looking sentence as interpretation,
not data.>

---
**Benchmark basis:** <clearance level per section; floor-disclosure note
where any query was broadened — "a privacy feature of the data source,
not missing data">
**Provenance:** <provenance lines from each response, verbatim>
· recency windows as noted per section
Credits used this run: ~N (breakdown: <calls>×5)

Honesty rules

• Portrait, not trend. There is no historical time-series tool; this memo describes the corpus now. Never write "up/down since last quarter" unless a prior snapshot exists — in which case route to vertical-forecast-brief.
• Bands are vertical-level market bands — never attach them to a specific creator (convention 2).
• Sub-category refusals are findings, not failures: "skincare is below the privacy floor" is itself reportable; never substitute vertical numbers under a sub-category heading without the label.
• Interpretation is labeled. The "what this means" section is the only place opinions live, and it says so.
• Provenance survives the forward. Never strip footers for prettiness — the memo travels without you (convention 1).
• No contact info (convention 7).

Credit footprint

thorough: ~20–30 credits (4 base calls + up to 3 sub-category calls, ×5; ladder retries add 5 each) · thrifty: ~10 credits (2 calls, capped ladder)