policy-compliance-gate: can we actually run this?

Launch & scale Cowork-ready

What it does

Tells you whether your ads will actually be allowed to run, before you launch them. It maps your vertical against Meta, Google and TikTok’s prohibited and restricted categories (health, finance, housing, employment, supplements, gambling, crypto and the rest), enforces special-ad-category declarations and their targeting floors, scans every asset and landing page against per-platform prohibited-claim patterns plus your own banned-claims list, and audits the destination URLs live.

Then it renders a verdict per platform lane: POLICY-CLEAR / RESTRICTED / BLOCKED.

It also drafts the appeal when a genuinely compliant asset gets rejected anyway, which happens.

It answers what is ALLOWED, never what is true. Whether a claim is honest is a different question, handled by your verified-claims library and red-team.

When to reach for it

• Before every launch. This is one of the three hard gates in new-client-launch.
• “Why was my ad rejected?”
• You are entering a restricted vertical and need to know the rules before you build creative to them.
• You need to know whether special ad category applies, and what it costs you in targeting.
• An account has been banned or restricted and you need the appeal written.

Where it runs

Cowork ✓ Desktop ✓

Both. It fetches the live destination pages, so it needs web access.

What it needs from you

• The assets: ad copy, image text, video scripts.
• The destination URLs, which it will actually load, because the landing page is where most rejections really originate.
• The vertical and the platforms you intend to run on.
• (It reads your own banned-claims list from your Brain if you have one.)

What you get back

• A verdict per platform, since the same asset can be clear on one and blocked on another.
• The specific offending line or element, quoted, with the policy it hits.
• Special-ad-category determination and the targeting restrictions that follow.
• A ban-risk score for the account overall.
• A drafted appeal where one is warranted.

A worked example

You: “Can we run this on Meta and Google?” (supplement brand)

Meta: RESTRICTED. Ad copy is clear, but the landing page carries “clinically proven to reverse insulin resistance,” which hits the prohibited health-claims pattern. Quoted with its policy reference. Google: BLOCKED on the same line plus a second one in the FAQ. Fix: both lines rewritten to the structure-function form that is permitted. Note: no special ad category applies here.

Tips & gotchas

• The landing page gets you banned more often than the ad does. Most people scan their creative and never load their own page against the policy list. This does.
• BLOCKED has no override in a launch. That gate exists because the alternative is discovering it with a disabled account.
• Clear on Meta does not mean clear on Google. Read the lane you are launching in.
• Allowed is not the same as true. This gate will happily pass a claim that is permitted and unproven. Your claims library is what stops that.

Related skills

A hard gate inside new-client-launch. Pair with red-team for whether the claim persuades, and your Brain’s verified-claims library for whether it is true.