Why did Meta reject an ad that follows the policy?
Because Meta did not only review your ad. It reviewed the page the ad points to, and the two did not match.
This is the single most misdiagnosed rejection in paid social. The disapproval notice names the ad, so the instinct is to rewrite the headline, swap the image, resubmit, and get rejected again. The creative was never the problem. Meta's ad review documentation states the review process may include an ad's associated landing page or other destinations, and its ad destination policy requires that the products and services promoted in an ad match those promoted on the landing page.
A compliant ad pointed at a page that promises something different is still a rejected ad. Fix the page, not the headline.
What Meta actually treats as a mismatch
A mismatch is any gap between what the ad promises and what someone lands on. It does not require bad intent, and most of the ones we clean up for clients are honest drift rather than anything deceptive. A page gets rebuilt, an offer gets renamed, a promo ends on the site but not in the ad set, and suddenly a campaign that ran fine for months starts getting knocked back.
These are the ones we see over and over:
- The offer in the ad never appears on the page. The ad says "free consultation," the link goes to a generic homepage with a contact form and no mention of a consultation. To a reviewer, the promise simply is not there.
- The price or discount disagrees. Ad says 30% off, page says 20%, or the ad quotes a price the page only honors on an annual plan. Numbers are the easiest thing in the world for an automated review to compare.
- "Free trial" that is not free at the door. The ad implies no-risk, the page opens with a credit card field. This one is a reliable rejection.
- Product naming drift. The ad promotes "Starter Plan," the page calls it "Basic." A human would shrug. A matching check will not.
- The destination fights the visitor. Full-screen interstitials, pop-ups that cover the offer, or pages that trap the back button. Meta treats a destination people cannot navigate away from as non-functional, and that is its own violation regardless of your copy.
The pattern underneath all five: the reviewer only sees what loads, cold, with no context. They do not know your funnel. They do not know the offer is on step two. If the promise is not visible on arrival, it does not count.
How to clear a rejection that has already happened
Work in this order. It matters, because editing a rejected ad resubmits it for review immediately, and every failed attempt makes the next one slower.
1. Write down the ad's literal promises. Every concrete claim: offer, price, discount, product name, trial terms, guarantee. That list is your checklist.
2. Open the real destination, cold. Incognito window, on a phone, no cookies, no logged-in session. Use the exact URL attached to the ad, parameters and all, not the page you think you linked. We have found more than one rejection that traced back to a stale UTM-laden URL pointing at a retired page.
3. Diff the promise against what loads. Every claim from step one should appear on that page, in the same words, visible without dismissing a pop-up first. Whatever is missing or renamed is your rejection.
4. Fix the destination. Not the creative. The creative is usually fine.
5. Confirm the fix is actually live before you resubmit. Deploys cache. CDNs lag. Resubmitting against a page that has not propagated gets you an identical rejection and burns a review cycle. Hard-refresh the live URL and check it yourself.
6. Then resubmit or request review. Meta says most ad reviews complete within about 24 hours and account-level restriction reviews within about 48, though both stretch during busy periods.
One warning worth taking seriously: do not appeal the same ad repeatedly hoping for a different reviewer. Stacked failed appeals can escalate into account-level restrictions and slower reviews on everything else you run. Fix it once and submit once.
The pre-flight check we run before any launch
Cleaning up rejections is reactive, and reactive is expensive when a client's launch is sitting idle. So we moved the check earlier.
Before anything gets submitted, we run an automated comparison between the queued ad copy and the live destination. In n8n it is a short workflow: pull the ad copy from the queue, fetch the rendered landing page HTML, extract the concrete claims from the ad (offer, price, product name, trial terms), and flag any of them that do not appear on the page. Anything flagged goes back to the strategist as a draft comment instead of going to Meta as a submission.
It is not clever. It takes seconds per ad. And it catches the boring failures — the promo that ended on the site but not in the ad set, the renamed plan, the homepage link that should have been a landing page — while they are still drafts.
The thing worth internalizing: the ad and the page are one unit now, so they need one owner. Most agencies have a creative team writing ads and a web team owning pages, and nobody is accountable for the seam between them. That seam is where rejections live. If your ad creative process and your landing page work never meet before submission, you will keep paying for it in stalled launches.
What the check actually compares
You do not need language models to do this well, and we would caution against making the first version clever. Start with exact, boring string matching on the handful of things that actually get ads rejected:
- Numbers. Every percentage, dollar figure, and duration in the ad copy. Pull them with a regex and confirm each one appears in the page text. Numbers cause more rejections than adjectives do, and they are trivial to compare.
- Offer nouns. "Free consultation," "free trial," "demo," "audit," "quote." If the ad names a thing the visitor is supposed to get, that phrase should be findable on the page.
- Plan and product names. Keep a per-client list of the names that are allowed to appear, and flag when the ad and page disagree. This catches the rename-drift that nobody remembers to propagate.
- The rendered page, not the raw HTML. If the client's site is a single-page app, fetching the URL returns a shell with none of the copy in it. Render it, or you will generate false alarms on every page and the team will start ignoring the check, which is worse than not having one.
Two failure modes to design around. First, fetch the page the way a reviewer would, without cookies or geo-personalization, or you will validate a version of the page that only your office sees. Second, flag, do not block. A check that hard-stops launches on a false positive gets switched off within a week. Ours writes a comment on the draft and lets a human decide.
The honest limitation: this catches literal mismatches, which is the large majority of them. It will not catch a page that is technically accurate and still misleading in spirit. That still needs a person.
What if the page is fine and it still gets rejected?
Then widen the review. A few things that are not strictly "landing page" but get caught by the same net:
- Check the whole path, not just the first page. If the destination immediately redirects, or the offer lives one click deeper, the reviewer may never reach it.
- Check the page it is associated with. For ecommerce especially, Meta may look at the Facebook Page, the ad, and the destination together when it is detecting misleading practices. Inconsistent branding across those three reads as a signal.
- Rule out the account, not the ad. If several unrelated ads start failing at once, you are probably looking at an account-level restriction rather than a creative problem. That is a different fix, handled in Account Quality, and it is worth reading your attribution and account health signals before you touch another ad.
- Confirm your tracking is not the tell. Broken or mismatched conversion setup will not get an ad rejected on its own, but a destination that fires nothing and matches nothing is worth fixing anyway while you are in there. Our Conversions API setup guide covers it.
The short version
Meta reviews the ad and its destination as one unit. If the ad promises something the page does not visibly deliver — an offer, a price, a plan name, a free trial — the ad gets rejected no matter how clean the creative is. Diagnose by opening the real URL cold on a phone and diffing it against the ad's literal claims. Fix the page first. Verify it is live. Submit once.
And then automate the check, so the next one never reaches a reviewer at all.
Running paid social and tired of launches stalling in review? We audit the seam between your ads and your landing pages, then automate the pre-flight check so mismatches get caught in draft. Get a Free Automation Audit and we will show you where your funnel is leaking review cycles.