How to identify MFA sites when no single signal is a verdict
Part of MFA avoidance
The short answer
To identify an MFA site, sample several recent articles and record observable signals in four areas: page layout, traffic sources, content, and ownership. Treat each signal as a reason to look closer, never as proof. Exclude a site only when strong evidence from more than one area agrees, and keep the screenshots, dates and links behind the call.
Key takeaways
- An observable signal is something a second reviewer could see and confirm on the live site, such as a sponsored link pointing to it or an ad slot that reloads on a timer.
- An MFA signal can have an innocent explanation, so the checklist records both what was seen and what else could explain it.
- Findings from different areas, such as layout plus bought traffic plus shared ownership, carry more weight than several findings from one area.
- Evidence decays as sites change, so each finding needs a date, a URL and a screenshot or archived copy.
What counts as an observable MFA signal?
The trade groups that defined made-for-advertising sites in September 2023, the ANA and the 4A's among them, in a joint definition the ANA published (opens in a new tab), described five characteristics that MFA sites usually show in some combination. They did not publish a test. That leaves reviewers with a practical problem: turning loose traits such as generic content or paid traffic into things a person can check, write down and defend when a publisher asks why it was excluded.
Definition
- Observable MFA signal
- A fact about a site that a second reviewer could see and confirm on the live pages or in public files, such as an ad slot that reloads on a timer, a sponsored link pointing to the site, or an ads.txt owner shared with many thin domains.
This checklist groups signals into four areas: page layout, traffic, content, and ownership with the selling path. The five characteristics, and the reasons MFA inventory looks clean in campaign reports, are covered in our explainer on what MFA sites are. This page is the reviewer's working sheet: what to look at, what each finding can prove, and what to save.
Which page layout signals should a reviewer check, and how?
Layout is the easiest area to observe and the easiest to misread. Measure it on the article pages a visitor would land on, not on the homepage, and on a phone as well as a desktop browser, because some templates load more units on small screens.
Two public standards give reviewers a consistent method. The Coalition for Better Ads (opens in a new tab) measures mobile ad density by adding up the heights of all ads in the main content of a page and dividing by the height of that main content, and its standards flag mobile density higher than 30 percent. Headers, footers and ads below the main content are left out, and a sticky ad counts once. Google's publisher policy (opens in a new tab) does not allow its ads on screens with more ads or paid promotional material than publisher content, and it does not count whitespace, headers or footers as content.
Layout signals to record for each sampled article
- Ad density on a phone was measured with the main content method, and the figure was written down.
- Any screen where ads and paid promotional units outweigh the article was captured in a screenshot.
- Ad slots that reload while the page sits untouched were timed, with the interval noted; our guide to ad refresh explains how refresh is declared.
- Stories split across many pages, each page with its own set of ads, were counted and the page count recorded.
- Video that starts without a click, sticky units and pop-ups were listed by type.
- Recommendation boxes labeled as sponsored were noted, along with where their links lead.
How can a reviewer see whether a site buys its traffic?
Paid traffic is the signal closest to the MFA business model and the hardest to see from outside, because a site's analytics are private. The joint trade group definition (opens in a new tab) says MFA publishers often have little to no organic audience and that buying paid traffic is the primary cost driver of operating an MFA business. A reviewer can still collect public traces of that buying.
Four public traces of bought traffic
- Note sponsored links that lead to the site. Labeled recommendation units on other sites are paid placements; record which domains they point to.
- Keep the landing URL. Campaign tags and click identifiers in the address bar show that the visit was bought.
- Check the advertiser record. Google's Ads Transparency Center lets people search for an advertiser (opens in a new tab); a publisher promoting its own article pages there is buying visits.
- Look at where the click lands. Bought clicks in the arbitrage pattern can land deep inside a slideshow, not on a homepage or a subscription offer.
Buying traffic is not wrong in itself. Respected publishers can also pay to promote stories or grow a newsletter, and the landing page shows the difference. Our guide to traffic arbitrage covers the full tracing method and how to tell audience building from arbitrage.
Which content signals point to a site made for ads?
Content signals answer one question: would anyone visit this page if the visit had not been paid for? Read three to five recent articles from different sections as a reader would, and record what you find rather than an overall impression.
| Signal | How to observe it | Possible innocent explanation |
|---|---|---|
| Same text on many domains | Search a distinctive sentence in quotes and count the other domains that carry it | Licensed wire copy or syndicated columns, which can appear on legitimate news sites |
| No author, date or sourcing | Look for bylines, publication dates and links to sources on each sampled article | Reference and recipe sites sometimes present undated evergreen pages |
| Topics with no common reader | List the subjects of the last 20 headlines and ask whether any one reader would follow them | General interest magazines can also range across subjects |
| Mass produced pages | Look for identical structure, phrasing and length across many articles published on the same day | Formats such as weather, scores or market closes are legitimately repetitive |
What do ads.txt, sellers.json and domain records reveal about ownership?
Ownership signals show whether a site is one of many built from the same kit. They come from public files and records, so they are easy to save as evidence. The ads.txt 1.1 specification (opens in a new tab) and the sellers.json specification (opens in a new tab) define the fields worth recording; our guides to ads.txt and sellers.json explain how each file works.
Ownership and selling path signals
- The site's ads.txt OWNERDOMAIN value was recorded; the spec defines it as the business domain of the entity that owns the site, so one owner behind many thin sites suggests a network.
- Any MANAGERDOMAIN value was recorded; the spec uses it for a primary or exclusive monetization partner that sells the site's inventory.
- Each DIRECT line was matched to the exchange's sellers.json entry, and the seller_type (PUBLISHER, INTERMEDIARY or BOTH) was written down.
- Ad and analytics account IDs in the page source were compared with the other candidate sites on the list.
- The domain's creation date was taken from registration data; ICANN (opens in a new tab) made RDAP the definitive source for generic top-level domain registration data as of January 28, 2025, in place of WHOIS.
- Archived copies of the homepage were checked for a recent change of template or topic.
Read these signals with care. Large media groups legitimately own many domains, a monetization partner can be normal for a small publisher, and a young domain may belong to a new newsroom. An ownership signal becomes weighty when it connects a candidate to other sites that already showed layout or traffic findings.
How should a reviewer weigh signals that point in different directions?
A checklist produces findings; it does not produce a verdict. A weighing rule that holds up when a publisher disputes a call is to count areas, not signals. Three layout findings on one site are really one finding about its template. A heavy layout, plus sponsored links leading to the site, plus an owner shared with sites already excluded, is three independent lines of evidence.
What evidence should a reviewer save for each finding?
Sites change templates, drop sponsored campaigns and rewrite ads.txt files. A finding without saved evidence cannot be checked later by a colleague, a client or the publisher. The Internet Archive's Save Page Now (opens in a new tab) stores a single page at a time, which suits article level evidence.
Evidence file for one reviewed site
- The exact URL of every sampled article, with the date and time of the visit.
- Screenshots of the first screen and of any screen where ads dominate.
- An archived copy of each page that supports a finding.
- A saved copy of the ads.txt file and the matching sellers.json entries on the day of review.
- The innocent explanation the reviewer considered for each finding, and why it was or was not accepted.
- The reviewer's name and the area each finding belongs to.
What does a signal review look like on a real plan?
Fictional example
Three candidate sites for a pet food brief in the US South
A fictional Atlanta pet food maker asks its agency for an English display plan in Georgia and the Carolinas, beside pet care advice and dog training content. The reviewer works through three candidates that looked alike on a spreadsheet.
The first, a regional veterinary advice blog, carries a dense layout on phones and a sidebar unit that reloads after the reader scrolls. Its articles are signed by staff, dated and specific, its ads.txt names its own business as owner, and no sponsored links to it turn up. One layout finding in one area: it stays on the plan with a note.
The second is a pet trivia site whose stories are 20 page slideshows. Sponsored links on two lifestyle sites point to deep slideshow pages with campaign tags in the URL, and its ads.txt OWNERDOMAIN matches four other trivia sites on the same template. With findings in layout, traffic and ownership, the reviewer writes up an exclusion case with screenshots and dates.
The third blocks automated page loads, so the phone review could not be completed. The reviewer records the gap and passes the site to a colleague for a manual visit instead of excluding it.
How does PlacementGraph record the signals behind an MFA call?
Frequently asked questions
How many articles should a reviewer sample per site?
Enough to see the site as a regular reader would: several recent articles from different sections, plus the pages that paid links lead to. Pages built for bought traffic can sit deep in a site, so a review of the homepage and top stories alone can miss them.
Should the review be done on a phone or a desktop?
Both. Some templates load extra units, sticky ads and interstitials only on small screens, while some desktop layouts carry wide side rails. Record the device and browser with each screenshot so a second reviewer can repeat the check on the same setup.
Can a site show no signals and still be made for advertising?
Yes, for a while. A site can switch templates or start buying traffic after it was reviewed, and some signals only appear on pages reached through paid links. That is why every finding carries a date and why a clean review describes the site on that day, not forever.
Is a newly registered domain a reason to exclude a site?
No. A recent registration date fits an MFA network that launches new domains, but it also fits a new newsroom or a rebrand. Treat domain age as a supporting clue that adds weight only when layout, traffic or ownership findings already point the same way.
Sources
- Leading Trade Groups Define “Made For Advertising” Websites (opens in a new tab)ANA (Association of National Advertisers), September 26, 2023
- Mobile Ad Density Higher Than 30% (opens in a new tab)Coalition for Better Ads, last revised February 16, 2018
- More ads or paid promotional material than publisher-content (opens in a new tab)Google Publisher Policies Help
- Advertiser verification (opens in a new tab)Google Advertising Policies Help
- ads.txt Specification Version 1.1 (opens in a new tab)IAB Tech Lab
- sellers.json specification (opens in a new tab)IAB Tech Lab
- ICANN Update: Launching RDAP; Sunsetting WHOIS (opens in a new tab)ICANN, January 27, 2025
- Save Pages in the Wayback Machine (opens in a new tab)Internet Archive Help Center
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.