What is an MFA site network, and how do shared IDs reveal it?
Part of MFA avoidance
The short answer
An MFA site network is a group of websites run by one operator that share templates, ad accounts and analytics tags while each one resells bought traffic as ad impressions. Buyers reveal a network by matching the public IDs in page code, ads.txt and sellers.json, then judging the group's shared behavior before deciding site by site.
Key takeaways
- A network turns one MFA finding into a pattern: one operator, template and set of ad accounts can sit behind several domains that look unrelated.
- AdSense publisher IDs, Google tag IDs, ads.txt account lines and sellers.json entries are public clues that link sites, but each one proves a different thing.
- Shared IDs also appear in legitimate media groups and corporate site families, so a link is a reason to review the group, not a verdict.
- A network finding should be written down with the identifiers, the dates seen and the behavior observed on each site.
What is an MFA site network?
The trade groups that defined made-for-advertising inventory in September 2023, the ANA (opens in a new tab), the 4A's, the WFA and ISBA, described single sites: heavy ad loads, refreshing placements, bought traffic, generic content and what they called poorly designed, templated website designs. A template that works for one arbitrage site can be copied onto the next domain quickly, so the risk a buyer meets on one site can belong to an operator running a family of them.
Definition
- MFA site network
- Two or more websites under common control that share layout, monetization accounts or measurement tags, and that show made-for-advertising traits as a group rather than one site at a time.
Academic work shows how such families can be traced. A paper by Papadogiannakis, Papadopoulos, Markatos and Kourtellis, published on arXiv for The Web Conference 2022 (opens in a new tab), used ad related publisher specific IDs to build a graph of who administers which sites across the top million websites. The authors found that approximately 90% of the websites are associated each with a single publisher, and that small publishers tend to manage less popular websites. The study covered site administration, not MFA, so it offers a method, not a list of bad sites.
Why do MFA operators launch similar sites in batches?
The economics reward copies. The ANA's 2023 definition states that buying paid traffic is the primary cost driver of operating an MFA business, and that covering those costs pushes operators toward aggressive monetization. Once one template, ad stack and traffic routine turn a margin, a new domain costs little more than its registration and a batch of generic articles.
The Web Conference paper describes the same incentive across the wider web: publishers spawn new websites, and older ones change hands with the sole purpose of monetizing user traffic. Batches also spread risk: when a buyer blocks one domain, the others keep earning.
Measurement standards already recognize traffic that moves between sister properties. The Media Rating Council's June 2020 invalid traffic addendum (opens in a new tab) defines affiliate traffic as traffic directed to properties from commonly owned and associated properties, separate from purchased traffic. A network passing visitors between its own sites creates that flow.
Should a whole network be excluded?
Not automatically. Shared identifiers are normal in legitimate businesses. Google's own Analytics account structure guide (opens in a new tab) describes a national insurer whose subsidiaries each have a website, and offers a setup with one Google Analytics 4 property with one (web) data stream for all the websites, corporate and subsidiary.
The decision should rest on what the linked sites do. If the linked sites share the traits the trade groups named, the IDs explain the pattern and support excluding the group. If only one site behaves that way, the link is context, and each sibling is judged on its own pages.
| What the review found | Suggested list action | Reason |
|---|---|---|
| Shared DIRECT seller account and template, MFA traits on the sites opened | Exclude the linked domains together and note the shared IDs | Common control and shared behavior both have evidence |
| Shared analytics tag only, sites differ in layout and ad load | Keep each site in its normal review | A tag can be installed by a vendor working for unrelated clients |
| Declared OWNERDOMAIN for a known publishing group, one weak site | Exclude the weak site, keep its siblings under review | A group can run strong and weak titles side by side |
| Same seller_id, seller_type PUBLISHER, unrelated topics and names | Hold the sites and ask the exchange or publisher to explain | IAB Tech Lab lists this as an irregular pattern worth checking |
How do you document a network finding?
A network decision reaches beyond the site that started it, so the record must stand up to a publisher's challenge. IAB Tech Lab's FAQ for sellers.json and the SupplyChain object (opens in a new tab) (July 2019, updated September 2020) supports a corroborated approach. It says buyers can spot check by asking publishers to confirm when a sellers.json file claims that a specific seller ID represents that publisher directly, and it flags cases where a number of apparently unrelated apps/domains are observed for a given seller ID but the seller type is set to PUBLISHER.
What a network finding record should hold
- The starting domain and the reason it was opened for review.
- Each shared identifier, its type and the date it was seen on each domain.
- The sellers.json name, domain and seller_type for each shared seller account.
- The OWNERDOMAIN or MANAGERDOMAIN values found, or a note that none were present.
- What a reviewer saw on two or three sampled articles per linked site.
- Which domains were excluded, which stay under review and why.
- Who signed off the decision and when the group will be looked at again.
Keep the account IDs in the record, not only the domains: they let the next reviewer recognize the operator on a new domain. The hub on MFA avoidance sets out the calendar and roles that keep such records current.
What does a network review look like on a real brief?
Fictional example
A hearing care clinic group's display plan in Florida
A fictional group of hearing aid and audiology clinics in Tampa, Orlando and Jacksonville asks its agency for English display beside health, retirement living and local news content in Florida. The planner's candidate list holds 60 sites, and one senior health tips site fails the page review: four refreshing slots on each article and a stream of visitors from sponsored link units.
The planner saves that site's AdSense publisher ID, its Google tag ID and its ads.txt DIRECT lines, then searches the rest of the candidate list for the same values. Five more domains carry the same publisher ID and DIRECT seller account, share the first site's footer text and layout, and repeat the same generic advice under different headlines. Two others share only the Google tag ID.
The first site and its five copies go to the exclusion list together, with the shared IDs recorded. The two tag only domains turn out to be a local retirement magazine and its events site, built by the same web developer. Their pages read as original reporting, so they stay under normal review. The final plan carries 31 sites, each with its reason.
How does PlacementGraph handle sites that belong to one network?
Frequently asked questions
Is an MFA site network the same as an ad network?
No. An ad network is a company that sells ad space on behalf of publishers. An MFA site network is a set of websites under common control. The two can overlap when one seller account represents the operator's domains, which is why sellers.json seller_type and name matter in a review.
Can unrelated sites share an analytics ID by mistake?
Yes. A web developer can reuse a tag across client builds, or a theme can ship with an ID left in its code. That is why a shared analytics tag on its own is the weakest link in the table above, and why it needs a second, stronger identifier before a buyer treats sites as one group.
What if a suspected network uses different ad accounts on each domain?
Separate accounts remove the easiest link, not the others. Shared templates, identical footer text and the same sellers.json name can still connect them. If no identifier links the sites, review each one on its own pages and record that no link was found.
Should a network finding be shared across clients?
A finding built on identifiers and observed behavior describes the inventory, not one brief, so it belongs on the agency's shared exclusion list with its record. Each client's tolerance still decides whether a borderline sibling site runs on that client's plan.
Sources
- Leading Trade Groups Define “Made For Advertising” Websites (opens in a new tab)ANA, with the 4A's, WFA and ISBA, September 26, 2023
- Leveraging Google's Publisher-specific IDs to Detect Website Administration (opens in a new tab)Papadogiannakis, Papadopoulos, Markatos and Kourtellis, arXiv, February 10, 2022 (The Web Conference 2022)
- Find your publisher ID (opens in a new tab)Google AdSense Help
- [GA4] Find your Google tag ID (opens in a new tab)Google Analytics Help
- [GA4] Google Analytics account structure (opens in a new tab)Google Analytics 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
- FAQ for sellers.json and SupplyChain Object (July 2019, updated September 2020) (opens in a new tab)IAB Tech Lab
- Invalid Traffic Detection and Filtration Standards Addendum, June 2020 Update (Final) (opens in a new tab)Media Rating Council
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.