Why Was My Ad Rejected, and How Do I Word Claims Safely?

Your new ad sits at "Disapproved" in the Status column. The policy label says Misrepresentation, yet the copy looks no riskier than last month's version. Rewording at random and resubmitting is tempting, but repeat violations of one policy earn strikes. And the label names a policy, not the sentence that broke it.

Ads are rejected when reviewers judge the ad, its landing page and the business together, and the claim promises more than that context can back. That means an outcome you cannot evidence, a credential or partnership you do not hold, or, on Meta, a statement about the reader's own circumstances. Safe wording claims only what the destination proves, and describes the product, never the person.

How does an ad actually get reviewed?

Google reviews ads with a mix of its own AI and human evaluation. Automated models either block an ad from serving or flag it for trained reviewers when the decision needs the ad and account read in context.

What matters for copywriting is what feeds that decision. Google's page on what happens when you violate its policies lists the inputs:

• ads, accounts, websites and content
• user complaints and consumer reviews
• regulatory warnings and rulings
• other third-party sources

So the unit under review is not the headline. It is the claim the headline makes, checked against the page it sends people to and the business behind both.

"Certified installer" can be well written and still fail, because the certification is nowhere on the site or lapsed last year. This is also why two near-identical ads can get different verdicts: the words matched, the destinations did not.

A disapproved ad cannot serve until you fix the violation and Google reviews it again. The fix means changing whichever part of that chain made the claim untrue. That is often the landing page or the offer rather than the ad text.

Which claims trigger a disapproval?

When an ad is disapproved for its claims, the cause usually falls into one of four patterns.

Claims about who you are. Google's Misrepresentation policy bars misleading statements about your identity, affiliations or qualifications, and omitting material information about them. Its own examples include implying endorsement by an organization or private citizen without their consent. Another is an unlicensed contractor presenting itself as licensed. "Official partner", "approved by", "as featured in" and "licensed" all live here.

Offers you cannot deliver. Google's Unacceptable business practices policy covers three things:

• products or services you do not have or cannot deliver, including without the right licenses or qualifications
• lying about services that could put health or safety at risk
• impersonating another brand to obtain money or personal information

Outcomes the evidence does not support. "Doubles your leads", "clinically proven" and "guaranteed results" are the usual suspects.

A claim can be literally true and still misleading. That happens when it only holds in narrow conditions, or when the overall impression says more than the words. Ads are judged as a whole, so implied claims count too. A photo of a uniformed paramedic beside your first-aid kit says "professionally endorsed" without a word.

Health and efficacy claims carry the highest bar. They need evidence about your specific product and that specific claim, not a study on one of its ingredients.

Claims about the reader. On Meta, the personal attributes rule is the usual culprit. Copy that asserts or implies something about the person seeing it can fail even when the targeting is broad. That includes their health, finances, age or beliefs.

"Struggling with debt?" tells the reader something about themselves. "Debt advice that starts with a free review" describes a service. Check Meta's own policy text for the current list of covered attributes rather than a third party's summary of it.

When does rejection become suspension?

This is where "reword and resubmit" gets risky, because consequences differ sharply by policy.

Under Misrepresentation, Google says violations won't lead to immediate suspension without prior warning. It issues a warning at least seven days before suspending. Repeat violations of a policy earn strikes, up to a maximum of three per policy.

Unacceptable business practices works differently. Google classes these violations as egregious, meaning unlawful or posing significant harm. Accounts are suspended on detection, without prior warning, and reinstated only in compelling circumstances.

An ad that implies a government affiliation is not a copy problem to iterate on. Neither is one that offers a regulated service without the license. Both are account risks.

The working rule: claims about credentials, partnerships, or health and safety get verified before they ship. They never get tested in live traffic to see what passes.

How do you word a claim so it passes?

Write every claim so the landing page proves it within one click.

• Put the proof where the click lands. If the ad says "award-winning", the page names the award and the year. If it says "certified", the certificate is listed and current.
• Prefer the checkable claim to the superlative. "Same-day dispatch on orders before 2pm" survives review and gives the buyer something concrete. "Fastest delivery anywhere" does neither.
• Describe the product, not the person. Swap "you who are X" for "people who want X", or lead with what the product does. Move qualification onto the landing page or the lead form, where readers sort themselves.
• State relationships precisely. Use your own name, logo and images, and follow Google's trademark policy for any brand reference. If you mention a brand you are not an official partner of, Google suggests a disclaimer in your ads and on your site.
• Keep the qualifier next to the claim. A footnote on one page does not undo a stronger impression elsewhere. "From $49" is honest when most plans cost more; "$49" alone is not.

A compliant claim is only half an ad. If stripping out the risky lines leaves you short of angles, the question becomes where good ad ideas come from when you've tried everything.

What should you do with a rejected ad?

Start from the policy label, then audit the claim against the destination. Ask what in the ad, the page or the offer would make a reasonable reader believe something untrue. Fix that at the source, then resubmit once.

If you believe the decision is wrong, appeal with evidence rather than argument. For a disputed partnership, Google suggests contracts, written agreements, a credible news article or a public statement from the brand.

What you should not do is cycle near-identical variants through review. Each repeat of the same violation counts against you.

Then make the fix durable. Keep an approved-claims list: each claim, the evidence behind it, who owns it, and when someone last checked it was still true. The safest new copy is built from claims you have already proven. That is why writing new ads from what already works beats starting from a blank page.

That list only helps if whoever drafts the next ad actually reads it. AI tools drafting variants at volume read nothing they are not given. In SproutMe Knowledge, brand guidelines and positioning sit in each client's workspace. The claim rules you write into them become context the agent drafts from, not something re-pasted into every prompt.

Conclusion

A rejection is rarely about one word. Reviewers read the ad, the page and the business together, so a claim fails when that context cannot carry it. It fails hardest when it touches credentials, partnerships or health, where the penalty can be the account rather than the ad.

The fix sits upstream of the copy: decide what you can prove, put the proof on the page, and write about the product rather than the person reading it.

See how brand guidelines, positioning and ICP definitions are held per workspace, so every agent drafting your ads starts from the same documented context.