MFA FAQ
Below you’ll find a list of frequently asked questions (and answers) about MFA ("Made for Advertising" or "Made for Arbitrage") domains. If you have a question that isn’t answered here, please reach out to a HUMAN representative for assistance. We’re always happy to help!
Why is HUMAN taking the approach of focusing on indicators rather than just using a list of MFA sites?
HUMAN has historically been focused on the ability to clearly determine whether something is a “bot or not,” so it’s in our DNA as a company to be focused on accuracy. As we’ve been participating in various industry discussions around MFA, we’ve seen that the definitions put forward weren’t actually definitions but rather different indicators that can be seen as pointers to an MFA. These aren’t always definitive and might be interpreted or valued differently by different parties. Therefore, our approach is to identify those indicators in order to share them with our clients as insights so that you can make your own business decisions.
How is MFA related to ad fraud or invalid traffic (IVT)?
MFA in and of itself isn’t proof that traffic is fraudulent or invalid. MFA, when properly disclosed, is more of an inventory quality issue, where different buyers and sellers may have different views about the significance of the MFA characteristics and the value of the inventory. MFA may be related to IVT when other factors are involved. Some such factors include deliberate efforts to cloak the ads so that only visitors from specific vectors will see them or so that scrapers that are specifically looking for MFA won’t find them. Others include mismatches between the domain intended for ad delivery and the actual domain on which the ad was served, where there is a material difference in the content between the domains (e.g., one domain leading to a normal news site and the other leading to an MFA-designed version of the site). In these cases, it’s the material difference in the content sold to the advertiser vs. what is expected that causes the MFA to be associated with IVT, not that a site is MFA in and of itself. In more extreme examples, a page might forcibly auto-scroll the user through the page to generate more ad loads or forcibly redirect the user to a new page (with new ad loads) after a certain period of time. Sites are also classified as IVT when standard IVT behaviors such as hidden ads, referral overwriting, or pop-under traffic is occurring.
How did HUMAN choose the indicator thresholds for defining MFA?
Our approach to thresholds is based on the methodology of looking at distributions and also performing manual reviews for determination. Most indicators require the presence of multiple indicators to determine that something is MFA.
Does HUMAN use machine learning (ML) to determine MFA sites?
No. The approach we use is deterministic, based on our indicators in addition to manual reviews.
How does HUMAN avoid flagging heavily monetized but legitimate sites as MFA?
Some sites may trigger a sufficient number of indicators that would otherwise be flagged as MFA, but they may actually be legitimate, non-MFA sites. An example is certain heavily monetized news sites. HUMAN has an internal review process to determine whether such sites should in fact be identified as MFA. Sites that might otherwise be flagged as MFA but that are actually legitimate, non-MFA sites are excluded from being categorized as MFA when they’ve been identified as such. We’re continually iterating on this process to ensure accuracy.
Updated 3 months ago