How to Spot an MFA Site in 60 Seconds: 6 Warning Signs

Made-for-advertising (MFA) sites are designed to pass a quick glance. They look like real publishers, with article layouts, category navigation, and author bylines. But the purpose is different. They exist to collect ad impressions, not readers. The content is filler. The traffic is bought. The "audience" is anyone who clicked a headline somewhere else.

You're funding this ecosystem every time your ad runs on one of these sites.

The good news is that MFA sites have tells. Experienced display advertisers develop an eye for them. Here's a checklist you can run manually in about 60 seconds for any domain in your placement report.

Signal 1: Ad Density

Count the ads on the page relative to the content. On a legitimate publisher, ads exist alongside meaningful content. On an MFA site, ads are the content. They're between every paragraph, in sidebars, autoplay in video slots, and often sticky to the bottom of the screen.

A rough threshold: if more than 30% of the visible screen area is occupied by ad units, you're likely on an MFA site. Industry studies consistently find several times more ad units per page on MFA sites than on legitimate publishers, and all that clutter costs your ad the attention you paid for.

Signal 2: No Real Navigation

Real websites have depth. They have about pages, contact information, categories with meaningful content, authors with bios that link somewhere real. MFA sites are shallow. Navigation links often lead to more aggregated content or are purely decorative.

Try to find: a real About page, a physical address or company name, an author bio that links to a real person. If none of these exist, that's a strong indicator.

Signal 3: Content That Reads Like It Was Written for Search, Not Readers

MFA content is optimized to rank for keywords and capture programmatic traffic. It's not there to inform anyone. It uses the same template over and over: broad question headline, 400–600 words of vague information, a handful of bullet points, a generic conclusion.

Ask: does this article take a position? Does it cite sources? Would a knowledgeable person in this field find it useful? If the answer to all three is no, the content was probably generated for impressions. Fluent writing doesn't clear a site anymore: AI-generated MFA sites read cleanly and still fail this test.

Signal 4: Pagination That Serves No Reader

Some MFA sites split articles across 10+ pages, not because the content warrants it, but because each page load triggers additional ad impressions. If you see "Page 1 of 12" on a 400-word article, you're on a site engineered for ad revenue, not reading.

Signal 5: Traffic Sources That Don't Add Up

Alexa (now retired) and tools like SimilarWeb can give you a rough traffic source breakdown. MFA sites often have unusually high shares of traffic from content recommendation networks (Outbrain, Taboola, MGID). A site that gets 60%+ of its traffic from content recommendation is essentially buying audiences and reselling them to advertisers at a margin.

Legitimate publishers with real audiences don't need to buy 60% of their traffic.

Signal 6: No Real Social Presence

Real publishers have audiences that engage with them off-site. Check the site's social links. Do they have a Twitter/X account with actual followers? A Facebook page with engagement? If a site claims to have millions of monthly readers but has 400 Twitter followers and no post engagement, the audience isn't real in any meaningful sense.

Building Your Exclusion List

As you audit your placement report, keep a running list. Flag every domain that hits three or more of these signals. Add it to your exclusion list and apply it at the account level in Google Ads.

Over time, this list becomes one of your most valuable campaign assets. It travels with you across campaigns and account restructures, and it reflects real spend data rather than someone else's opinion of what's low quality.

For high-spend placements, add performance data to the check. These eight signals combine the on-page tells with bounce rate and conversion history.

The manual process is slow. But the judgment it builds is useful even if you eventually automate it, for example with DisplayGateGuard's free placement exclusion list generator.