Meta & Facebook Ads
4 min read
Healthy CTR, terrible CPA. The instinct is to blame the ad, so you rewrite the hook, test 5 new variants, and nothing improves. The ad was doing its job. The page broke the promise it made.
The free Claude skill landing-page-message-match restates what your ad promised, then scores the page against it on 5 checks and tells you which side is cheaper to fix.
This post covers the 5 checks, what the score thresholds mean, and why the cheaper fix is sometimes rewriting the ad.
Key takeaways
The page gets scored out of 10 across 5 checks, each rated absent, present, or prominent.
Under 5 out of 10 means the pair is burning clicks, so don't scale it no matter how good the ad metrics look.
Every break comes with 2 fixes, 1 on the page and 1 on the ad, plus a recommendation on which side to change.
Matching is scored per pair, so if 5 ads point at 1 page, that's 5 separate audits.
The 5 checks
Check | The question it asks |
|---|---|
Headline echo | Does the page's first headline restate the ad's promise, not the brand slogan? |
Offer consistency | Same discount, bundle, and guarantee, with no checkout surprises |
Proof carry through | Is the ad's claim substantiated above the fold? |
Visual continuity | Does the hero match the creative world of the ad? |
Audience continuity | Does the page speak to the objection the angle targeted? |
Each scores 0, 1, or 2, giving a total out of 10. At 8 or above you ship. Between 5 and 7 you fix the listed breaks first. Under 5 the pair is burning clicks and no amount of bid tuning will rescue it.
Audience continuity is the one that catches sophisticated teams. A mechanism ad that explains why the product works, landing on a lifestyle page full of mood photography, is a genuine break even though both assets are individually good.
Sometimes the ad is the cheaper fix
This is the part that separates the skill from a generic landing page audit. For every break it gives you both options: the fix on the page and the fix in the ad.
If your ad over promised, rewriting the ad is faster and cheaper than rebuilding a page to support a claim your product cannot really carry. And the skill will not recommend adding claims to a page just to satisfy an overreaching ad. In that argument, the ad loses.
When the page genuinely cannot change, on a marketplace listing or under compliance constraints, it hands that constraint forward as a production instruction: write the ad to the page that exists.
When the fix is the ad, make the ad
Rewriting to match the page means new creative, not just new copy. Vibelets generates on-brand product ads from your store in minutes, no designer or brief required.
Match is scored per pair, not per page
If 5 ads point at 1 landing page, that is 5 separate audits, because each ad makes a different promise of the same page. A page can score 9 against your mechanism ad and 4 against your price ad without changing a pixel.
This is why a single page audit misses the problem. The page is not good or bad in isolation. It is aligned or misaligned with a specific promise, and the moment you add a new angle you have created a new pair that has never been checked.
Mobile is the default assumption
The audit assumes mobile rendering, because that is where the clicks land. Anything that only works on desktop gets called out.
This matters more than it sounds. Proof that sits above the fold on a laptop can be 3 scrolls down on a phone, which means the proof carry through check quietly fails for most of your actual traffic while looking fine to whoever built the page.
How to set it up in Claude
Download the skill file from the skill's Google Drive folder, free, no email gate, then add it to Claude by uploading it as a skill under Settings or attaching it to a project or chat. Anthropic's Help Center guide to skills covers the upload path.
Give it the ad, meaning the copy plus a description of the creative, and the landing page URL or pasted content. Run it before launching any new angle, and any time CTR looks healthy while CPA does not.
Conclusion
The click is a promise. The page either keeps it or refunds it. Score the pair, fix the cheaper side, and stop rewriting ads that were never the problem.
Score your ad to page match out of 10
Free Claude skill, yours in 1 click.
Frequently asked questions
Why is my CTR high but my CPA bad?
Usually a broken promise between ad and page. The ad sells one thing, the page greets the click with another, so qualified clicks bounce. Scoring the message match finds the break faster than testing new hooks.
What is a good message match score?
8 out of 10 or above is ship ready. Between 5 and 7, fix the listed breaks first. Under 5 the pair is burning clicks and should not be scaled regardless of how strong the ad metrics look.
Should I fix the page or the ad?
Whichever is cheaper, and the skill recommends a side for each break. If the ad over promised, rewriting the ad usually beats rebuilding a page to support a claim the product cannot carry.
Do I need a separate audit for each ad?
Yes. Message match is scored per ad and page pair, so 5 ads pointing at 1 page means 5 audits. Different angles make different promises of the same page.
Does the audit assume mobile or desktop?
Mobile by default, since that is where most paid social clicks land. Anything that only holds up on desktop, such as proof that sits above the fold on a laptop but 3 scrolls down on a phone, gets flagged.
Sources
Make ads that sell
your Products.
Generate on-brand product ads in minutes, no designer or brief required.
Trusted by 400+ growing brands