Verify every audience, attendance and reach number in sponsor material against reconciled source counts before drafting or sending. Use when the user drafts a sponsor proposal, deck or partner claim with audience numbers.
Installs into .claude/skills of the current project.
Sponsors
SSkills DirectorySkills Directory
Is that skill safe to install?
SSkills DirectorySkills Directory
Know which skills are safe — weekly.
Are you the author of Sponsor Audience Claims?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/lionelsimai-sponsor-audience-claims)
---
name: sponsor-audience-claims
description: "Verify every audience, attendance and reach number in sponsor material against reconciled source counts before drafting or sending. Use when the user drafts a sponsor proposal, deck or partner claim with audience numbers."
category: Sales & Marketing
license: MIT
version: 1.0.0
---
# Sponsor Audience Claims Check
## Pin the counting unit
- Name the counting unit beside every audience number before you use it, choosing one of invite row, registration, unique person, check in or consented opt in.
- Treat an invite row as an invitation record and never as a registration, an attendee or a reachable person.
- Count unique people once across events when a number spans more than one event, and say plainly when a number is a row count instead of a people count.
- Separate people who were merely touched from people who consented to sponsor updates, and use only the consented count for any sponsor facing reach claim.
## Reconcile to the verified base
- Reconcile each number against the latest verified extraction or source export on record before you draft, revise or analyse sponsor facing material.
- Replace any number that exceeds its reconciled source count with the source count, or drop the number when no source supports it.
- Derive expected attendance from the verified show rate in the source data rather than an assumed fill rate, and show the arithmetic in the working notes.
- Recompute derived counts such as repeat attendees and segment sizes from the underlying rows whenever a draft states them, because a wrong counting unit inflates every figure built on it.
- Rerun the whole check on a revised version instead of trusting the pass you ran on an earlier draft.
## Check the promise
- Strike any promise to hand a sponsor the full registration or attendance list, and offer the consented opt in list plus an aggregate audience profile instead.
- Check currency, price tier and partner conversion terms against the current price book and deal record on the same pass, so a corrected audience claim does not sit beside a stale commercial claim.
- Hold the material as not ready when a number has no source, and tell the user which number is ungrounded rather than rounding it into place.
## Present the checked result
- Present each corrected number with its counting unit and its source beside it, so the user approves the exact claim that will be sent.
- Report each correction plainly by naming the claimed figure, the verified figure and the unit error that caused the gap.
- Keep the corrected figures in the delivered document itself rather than only in the covering message.
## When to use
Use this skill when the task matches its description above and you need a repeatable, checked workflow rather than an improvised answer.
## Guardrails and quality checks
- Never fabricate facts, numbers, clients or results. Use only sources the user provides or verifies.
- Keep drafts as drafts until a human approves any external send, payment or commitment.
- Check the final output against the workflow above before delivering it, and state anything you could not verify.
## Output format and example
Deliver the result in the structure the workflow defines. Example request: "Use the Sponsor Audience Claims skill for my situation: [describe the business, audience and goal]." Example output: a completed, checked deliverable following every step above, with open questions listed at the end rather than guessed.