Copy Compliance Scrub

---
name: copy-compliance-scrub
description: Scrub Meta ad copy and creative text for policy violations before launch — personal attributes, implied characteristics, prohibited claims, and misleading promises. Use when asked to check ad compliance, review copy before launch, fix a rejected ad, work out why ads keep getting rejected, make claims policy-safe, avoid account restrictions, or someone says 'Meta rejected my ad again'. For writing the copy in the first place, see headline-and-primary-text. For claims and proof discipline, see ten-angles-copy.
metadata:
  version: 1.0.0
---

# Copy compliance scrub

Meta reviews every ad automatically, and the single most common rejection is the
**Privacy Violations and Personal Attributes** policy: copy that asserts *or
implies* something about the person reading it. "Struggling with debt?" implies
their financial status — that is enough. Rejections stack toward account
restriction, which costs far more than the ad that triggered it.

This scrubs before launch, and diagnoses after a rejection. It is a policy *check*,
not a policy *guarantee* — Meta's standards move, and the current text at
transparency.meta.com/policies/ad-standards is the authority.

## Before you start

| Input | Why |
| --- | --- |
| The full copy set — overlay, headline, primary text | Violations hide in any field |
| What the product is | Restricted categories depend on it |
| The landing page | Meta reviews the destination too |
| Rejection notice, if this is a diagnosis | The cited policy narrows the search |

## Step 1 — Scrub for personal attributes (the #1 rejection)

Copy must not assert **or imply** the reader's: race, ethnicity, religion, beliefs,
age, sexual orientation, gender identity, disability, physical or mental health
(including medical conditions), vulnerable financial status, voting status, union
membership, criminal record, or name.

The "you" test — implication counts:

| Violates (implies their attribute) | Safe (describes the product/outcome) |
| --- | --- |
| "Struggling with debt?" | "Tools for managing money better" |
| "Living with diabetes?" | "Meal plans designed around blood sugar" |
| "Are you over 50 and single?" | "A community for meaningful connection" |
| "Feeling depressed?" | "Support that meets you where you are" |

Rewrite pattern: **move the sentence's subject off the reader and onto the product
or the outcome.**

## Step 2 — Scrub claims

- Every number, rating, and testimonial is real and supplied — misleading or
  exaggerated claims are a top-three rejection trigger.
- No promised outcomes stated as certain ("You WILL get hired").
- Health/appearance products: no copy engineered to create negative
  self-perception — Meta prohibits it for diet and weight-loss promotion
  explicitly.
- Before/after framing on body-related products is high-risk; describe the
  program, not their body.

## Step 3 — Check the category

If the product touches money, health, housing, employment, politics, dating,
alcohol, or similar sensitive areas, it likely sits in a **restricted category**
with extra rules or declarations. Check the current category list at
transparency.meta.com before assuming standard rules apply — this skill does not
enumerate them because they change.

## Step 4 — On a rejection: fix, don't just resubmit

1. Read which policy the rejection cites.
2. Fix every instance across all fields *and* the landing page — resubmitting
   unchanged copy trains the account's risk score the wrong way.
3. Appeal only when you genuinely believe the automated call is wrong.
4. Repeated rejections on one theme → stop the theme, not just the ad.

## Output

```
# Compliance scrub: <ad set>

| Field | Text | Finding | Rewrite |
| overlay | "..." | personal attribute (financial) | "..." |
| headline | "..." | clean | — |
...

Category check: <standard / restricted — which, and why>
Landing page: <consistent with copy claims? y/n>
Verdict: LAUNCH / FIX FIRST — <one line>
Note: checked against policy as of <date>; authority is transparency.meta.com
```

## When it breaks

| What you see | What it means | The fix |
| --- | --- | --- |
| Rejected: personal attributes | Copy implies who the reader is | Subject off the reader, onto the product |
| Rejected again after rewording | Synonym swap, same implication | Change the sentence's subject, not its vocabulary |
| Approved ad rejected weeks later | Re-review or policy change | Re-scrub against the current standards, not the launch-day ones |
| Rejections across the whole account | Risk score elevated by repeat offences | Pause the offending theme; fix everything before new submissions |
| Ad fine, still restricted | The landing page violates | Scrub the page's claims the same way |
| "Competitors run this exact copy" | Their risk, or their pending rejection | Their account is not the policy standard |

Never fix a rejection by resubmitting the same copy hoping for a different
reviewer. Automated enforcement doesn't have moods, and repeats compound the
account's risk score.

## Rules

- **Implication counts as assertion**, because the policy explicitly covers
  indirect implications — "struggling with…?" fails without naming anything.
- **The subject moves off the reader**, because that one edit resolves most
  personal-attribute violations without losing the message.
- **Unverifiable claims do not ship**, because misleading claims are a top
  rejection trigger and a trust cost beyond the platform.
- **This scrub is not a guarantee**, because standards change and enforcement is
  Meta's — the Transparency Center is the authority, checked fresh for anything
  high-stakes.
- **Repeat rejections stop the theme, not the ad**, because account-level
  restriction costs more than any single creative.

## Related skills

- **headline-and-primary-text** — write the copy this skill scrubs.
- **ten-angles-copy** — the Callout angle is the most scrub-sensitive; its safe
  form names the situation, not the person's attribute.
- **offer-builder** — honest urgency and guarantees, upstream of claim problems.