The notice says your ad doesn't comply with our advertising policies. It links to a page containing roughly forty policies. It offers a button marked Request review. Then it stops talking.
This feels personal. It is not. It is closer to a parking ticket than a verdict on your character, and like a parking ticket, it refers to a specific written rule that somebody can quote at you if you ask.
The rules are public. All three of the platforms you are likely using publish them: Meta's Advertising Standards, TikTok's Advertising Policies, Google Ads policies. They are long, dull, organised by category, and searchable. Almost nobody selling their first product reads them, which is reasonable, because they read like insurance documents. But the thing that rejected your ad was matching it against those documents, mostly automatically, mostly in seconds. You can do the same match yourself, with better context than the machine had.
The rejection reason is a label, and sometimes it is the wrong label
When an ad is rejected, you usually get a category name. Personal attributes. Unacceptable business practices. Circumventing systems. Destination not working. Insufficient original content. Misrepresentation.
That label is a pointer to a clause. Find the clause, read the actual sentences, and the rejection shrinks into something workable.
The honest caveat is that the label is sometimes wrong. Review at this scale is done by classifiers trained on patterns, and a classifier can flag your ad under one heading when the thing it disliked sits under another. I have seen people rewrite an entire ad to satisfy a policy that was never the issue. So read the named clause, but also read your ad against the two or three clauses either side of it, because the machine is guessing at a name for what it noticed.
The clauses that catch people selling their first thing
There is a short list of rules that new advertisers break constantly, and they are not the rules about guns and gambling.
The first is personal attributes. Meta's version says ads must not assert or imply that you know something private about the person seeing them: race, ethnicity, religion, age, sexual orientation, gender identity, disability, medical or physical condition, financial status, criminal history, and several more. The logic is not squeamishness about the subject matter. It is that the person scrolling should not feel identified. "Struggling with acne?" implies you know they have acne. "Skincare for acne-prone skin" describes the product instead. Same product, same audience, different grammar. Second person plus a condition is the shape that gets caught, and it is the shape almost every copywriting guide teaches, which is why so many first ads trip on it.
The second is unrealistic outcomes and unreliable claims. Meta prohibits ads that set unrealistic expectations, and health and appearance content gets extra scrutiny. Google has an entire section on unreliable claims covering get-rich-quick offers, guaranteed results and cures. Numbers attached to money are the fastest way in. An ad that names a figure per month, or promises a result by a date, is asking to be rejected. If it slips through, you have a larger problem than a rejection, which is that you have made a written promise to strangers. I wrote separately about what works online and what is a scam, and the shared thread is that the claims which get ads pulled are usually the claims that were not true.
The third is before-and-after imagery. Both Meta and TikTok prohibit it in the health, fitness and appearance space, including split screens, weight numbers, and images that zoom in on an isolated body part. This one surprises people because before-and-after is the most persuasive asset they own. It is prohibited precisely because it is the most persuasive asset they own.
The fourth is restricted categories, which are not banned but need permission or age gating. Financial products and services, cryptocurrency, supplements, alcohol, dating, anything prescription-adjacent. Meta requires certification or written authorisation for parts of this. TikTok's prohibited industries list is longer than Meta's, and it changes, so check it against the country you are targeting rather than against what someone told you in March. If your product sits in one of these categories, the rejection is not about your creative and rewriting the headline will do nothing at all.
The fifth is circumventing systems, and this is the one that turns a small problem into a large one. Making eleven microscopic edits and resubmitting each time is a pattern the system recognises. So is misspelling a word to dodge a filter, sending the reviewer to one page and buyers to another, or opening a second ad account after a restriction. The other clauses are about your ad. This one is about your account.
The landing page is half the ad
Here is the part people miss, because it is not on the ad at all.
Reviewers load your link. Google publishes destination requirements as a policy in its own right, and the rejection reasons under it are specific: destination not working, destination mismatch, insufficient original content, unexpected pop-ups or interstitials, automatic downloads. Meta's landing page rules are shorter, but the idea is identical. The page must load, must work on a phone, must be about the thing the ad advertised, and must not do anything surprising when it opens.
Google's insufficient original content clause names pages whose primary purpose is sending people somewhere else. That is a description of a link-in-bio page, so I should say plainly that Relay can be part of this problem. If you are running Google Search ads to relayme.bio/yourname, you are pointing a policy that dislikes redirectors at a page that redirects. We have no custom domains, and we do not process payments, so the URL is ours and the checkout happens on somebody else's site. For Meta and TikTok traffic, that arrangement is usually fine, because those platforms care mainly that the page loads, matches the ad, and behaves. For Google Search, it is a real risk, and if that is your main channel, the ad should go to a product page with actual product content on it.
What a link page is good for is the messy middle. You are selling on TikTok Shop, taking DMs, and you also have a guide on Gumroad, and you want one destination that survives all three. If you are still working out whether you need a shop at all, TikTok Shop vs Shopify vs just a link covers the trade-offs. The relevant bit for ad review is that a page with your photo, a bio that says what you sell, and clearly labelled links reads as a real page to a human reviewer, where a bare stack of six unlabelled URLs does not.
On Relay, that page is free with no card and no time limit: unlimited links on every plan, your own photo and bio, social icons, five themes, headings and dividers to group things, a QR code and a story-sized share image. Tap and view counts are included, which matters more than it sounds during an ad test, because when the platform reports 68 clicks you can see how many people actually arrived. If those two numbers are far apart, the problem is not your copy. Pro is $30 a year or $4 a month plus VAT and adds all 47 themes, all 8 fonts, custom colours, a background image, per-link images, scheduled links, inline players, email capture and removal of our badge. Data is held in the European Union, fonts are self-hosted, and there is no advertising and no third-party analytics on your page, which also means no surprise scripts for a reviewer's browser to choke on. If you want the version where we list what we lose on, it is on the comparison page.
What requesting review actually does
It sends the ad back through a mix of automated and human checks. It is free. It usually resolves within a day, sometimes faster, occasionally not for a week with no explanation for the difference.
Appealing once with the ad unchanged is legitimate if you have read the clause and believe the classifier misread you. That happens often enough to be worth a click. Text baked into an image gets misparsed; a photo of a smoothie gets read as a supplement; a course about bookkeeping gets read as a financial service.
What is not legitimate is the loop: appeal, refused, change one word, resubmit, refused, change one word. That is the pattern the circumventing systems clause exists for, and it moves you from a rejected ad to a flagged account. If two appeals fail, the platform means it, and the fix is a different ad rather than a more determined version of the same one.
A single rejection appears to carry no lasting penalty on any of the three platforms, though none of them publish enough for me to say that with certainty. A run of rejections does carry weight. Accounts get restricted on patterns, and a pattern is easier to trigger from a new account with no history than from one that has spent quietly for six months.
The order I would check things in
Read the stated reason, then open the policy page it links to rather than the summary in the notice. Read your primary text out loud as though it were addressed to one stranger, and notice whether it claims to know anything about them. Look at the image with the before-and-after rule in mind even if you are nowhere near fitness. Check whether your product sits in a restricted category, because if it does, nothing you do to the creative will help.
Then open your landing page in a private window on a phone, on mobile data, with no saved logins. Time how long it takes to become useful. Check that every link goes where its label says. Check that the page mentions the thing the ad was about, in words, near the top.
Then appeal once, and go and do something else for a day.
One last piece of arithmetic. A rejection on the first day of a small test is irritating but not expensive, because you were not spending while the ad sat in review. If you are working with a small budget, the maths of what $50 buys is worth reading alongside this, since it explains why a two-day delay costs you far less than the variance already baked into a campaign that size. And if the first sale is still ahead of you, the timeline post is the less cheerful companion piece.
The rules are not a mood. They are written down, they are boring, and they mostly say four things in different words: do not pretend to know who is reading, do not promise what you cannot deliver, do not sell restricted things without permission, and make the page at the other end work. We can help with the last one, and only partly. The rest is yours to write.
