How Scam Ads Slip Past Review: Cloaking, Swaps & Weak Spots

Scam ads get approved because ad review judges a snapshot while a scam campaign is a moving target. Networks review the creative and landing page an advertiser submits — mostly with automated scanners, sometimes with human eyes — and then largely stop watching. Scammers exploit that gap with four reliable techniques: cloaking (showing reviewers a clean page while real users see the scam), post-approval swaps (changing the destination content after the ad goes live), redirect-chain switching (re-pointing the click path through trackers the advertiser controls), and disposable accounts that treat bans as a routine cost of doing business. None of this is technically sophisticated. It is patient, high-volume abuse of one structural weakness: approval is a point-in-time decision about artifacts the advertiser controls.

This article walks through each technique, explains why native ad networks get hit especially hard, and shows how to recognize — and document — a scam ad that made it through review.

Ad review is a snapshot. A scam campaign is a movie#

Every major ad platform reviews ads in roughly the same way. An automated pass scans the creative's text and image for banned categories, fetches the declared landing page, and scores the account on risk signals — age, payment method, past violations, similarity to previously banned advertisers. Human reviewers see only the slice that automation flags, or that falls into sensitive categories. That is not laziness; it is arithmetic. Networks process enormous volumes of new creatives every day, and per-ad human review does not scale.

The structural weakness is when review happens: at submission, and sometimes again on edit. Approval is a judgment about what the advertiser showed the network on Tuesday. The campaign then runs for weeks, touching systems the network does not control — the advertiser's landing domain, the advertiser's tracking domain, the advertiser's server-side routing logic. Everything a scammer needs to change can be changed without touching the ad platform, which means nothing triggers a re-review. That one fact explains almost every "how did this get approved?" headline.

Technique 1: cloaking — the reviewer never sees the scam#

Ad cloaking means serving different content to reviewers than to real users. The mechanics are unglamorous: the landing page, or a router sitting in front of it, checks each visitor against lists of known data-center and ad-network IP ranges, then checks geo, device type, user agent, and sometimes behavioral signals. Visitors who look like reviewers get a "white page" — a bland, compliant article or storefront. Visitors who look like the target audience get the money page: the fake celebrity endorsement, the miracle cure, the investment scheme.

Cloaking is sold as a commercial service with monthly pricing, dashboards, and support channels, which tells you how mature the market is. Countermeasures exist — networks re-crawl from residential-looking IPs and randomize their review fingerprints — but it is an arms race in which the defender has to win every fetch and the cloaker only has to win at review time. We cover the detection side in depth in how ad cloaking works and how evidence exposes it.

Technique 2: the post-approval swap#

The simplest bypass needs no software at all: submit a compliant page, wait for approval, then change the page. The landing page lives on the advertiser's own domain. The network approved what was there at review time; it does not diff the page daily. A harmless advertorial about "joint health after 50" becomes an aggressive supplement funnel with fabricated doctor quotes a week later, and the ad platform's records still show the compliant version it approved.

A variant targets the ad side. Some networks allow edits to headlines, thumbnails, or destination URLs on live campaigns with lighter-touch re-review than a fresh submission. Operators probe exactly how much they can change before a full re-review triggers, then stay under that line. The ad that got approved and the ad that runs at scale are technically the same campaign object — and materially different artifacts.

Technique 3: redirect chains that move after approval#

Almost no performance ad points directly at its final page. The click travels through a redirect chain: the network's click handler, then one or more advertiser-controlled tracking domains, then the lander. The network reviews the destination it can see at submission — but the middle of the chain belongs to the advertiser, and a tracker is just a server answering "where should this click go?" That answer can change at any moment, or vary per visitor.

The per-visitor part is the potent bit. A tracker can split traffic by geo, device, carrier, or time of day: a reviewer or monitoring service fetching from a US data center gets the compliant path, while mobile users in the target geo get routed to the scam. The approval was honest, in a narrow sense. Everything after it was theater.

Technique 4: burner accounts and the economics of volume#

All of the above assumes the scammer wants to keep the account. Many don't. Accounts are cheap: fresh signups warmed with a few weeks of legitimate-looking spend, aged accounts bought on gray markets, agency seats resold with someone else's clean history attached. If a scam funnel turns profitable within days and detection-plus-ban takes days, a ban is not a deterrent — it is a line item.

This is why "why doesn't the network just ban them?" misses the shape of the problem. Networks ban them constantly. The countermeasure that actually bites is fingerprinting repeat offenders across accounts — payment instruments, creative assets, landing infrastructure — and networks have invested heavily there. But the contest is adversarial, and the attacker iterates faster than the defender's release cycle.

Why native networks are a soft target#

Native advertising has a structural property that makes scam laundering easier: the legitimate baseline is already gray. Pre-landers and advertorial funnels are normal, accepted mechanics in native — plenty of legitimate affiliate campaigns run story-style pages with dramatic headlines. A scam funnel is not structurally different from a legitimate affiliate funnel; it is the same shape with fabricated content. Review systems that key on structure struggle when structure carries no signal.

Demand resale adds another layer. On parts of the native supply chain, an ad can enter through a demand partner and surface on a network that never had a direct relationship with the advertiser, diluting accountability at each hop. And the vertical mix concentrates temptation: across OpenAdLibrary's index of 725,000+ live native ad creatives (June 2026), health and finance are the two largest verticals at roughly 24,000 classified creatives each — exactly the categories where miracle cures and get-rich schemes prefer to hide. We break down the wider taxonomy in Ad Fraud in Native Advertising.

What review actually catches#

None of this means review is useless. Modern review stacks catch an enormous amount: image classifiers flag celebrity faces and before/after shots, text models flag miracle claims and banned categories, landing scanners follow declared destinations, verification programs demand business documents in sensitive verticals, and payment-risk systems kill accounts before the first impression serves. The obvious, lazy scams die at submission — which is why you mostly see the survivors and wonder how they got through.

What survives is the adversarial tail: operations that invest in cloaking, account hygiene, and swap discipline. That tail is small as a percentage of submissions and large in absolute terms, which is how the same fake endorsement scheme can be banned a thousand times and still be running somewhere today.

How to spot an approved scam ad in the wild#

A few signals separate scam ads from merely aggressive ones:

• A celebrity or public figure endorsing a product they have never mentioned through any official channel.
• The brand name shown on the ad does not match the domain the click lands on — advertiser/destination mismatch is the single most reliable tell.
• The landing domain was registered within the last few weeks.
• Steep discounts on recognizable products from a storefront you have never heard of, usually with a countdown timer.
• The same creative running under several different advertiser names at once.

Verifying most of these requires seeing the click path and the landing page, which is exactly what an independent ad intelligence index is for. OpenAdLibrary records each native ad with its resolved advertiser, redirect trail, and captured landing page — over 1.3 million landing captures as of June 2026 — so you can check what an ad actually resolved to without clicking it yourself. If the ad borrows a brand's identity outright, that is its own playbook: see Copycat Landing Pages and Trademark Infringement in Ads.

Report it with evidence that sticks#

Reports fail when they contain a screenshot and nothing else. Networks act fastest on reports that include the creative, the publisher page where it appeared, a timestamp, the click URL, and the final landing domain — enough to reproduce the violation and match it against internal records. Our guide to reporting a scam ad covers where to file for each network and how to package the evidence. Consumers can also file with the FTC at ReportFraud.ftc.gov, which feeds law-enforcement databases even after the individual ad disappears.

Approval is not an endorsement. It is a snapshot that a motivated adversary controls. Once you internalize that, scam ads stop being mysterious — and their tells become much easier to read.