Understand Traffic Insights
The Insights page of the HUMAN Dashboard displays insights about your ad inventory traffic. These traffic insights are additional data points that HUMAN may use as part of overall invalid traffic (IVT) detection but aren’t sufficient to determine IVT by themselves. This data can, however, provide insights into inventory quality.
These insights are based on signals HUMAN already collects and are meant to help you make informed business decisions. For information about IVT, see IVT Taxonomy.
Note
Traffic insights are currently only available for MediaGuard integrations, not for FraudSensor-only integrations.
The page contains the following insights:
Each insight displays the five risk entities with the highest volume of traffic with inventory quality issues. Within each insight area, you can also:
- Select
to generate and download a CSV file containing data for the first 5,000 entities. - Select View in Explore to review all data and create views, reports, and alerts on the Explore page.
You can learn more about each traffic insight in this article.
Domain Mismatch
Includes data since 1/3/2025
Domain mismatches are instances where the intended targeted domain declared pre-bid was different from the actual delivery domain detected post-bid. For example, the impression’s associated BidRequest object may be configured incorrectly, which can cause delivery on unintended domains. Non-benign mismatches that show material differences between intended domain and actual domain are considered IVT. For further information see FraudSensor Domain Mismatch IVT Classifications.
Note
Domain mismatch data only represents JS traffic on desktop and mobile.
You should address these configuration issues so HUMAN can continue to detect actual fraud when it occurs. Including accurate domain data also helps buyers know where the impression will be served and make informed purchases for ad space. If mismatches persist, ads may be served to unintended locations, leading to lower engagement and wasted ad spend.
We suggest reaching out to any impacted publishers to make sure that their BidRequest objects are populated with the correct domain information.
IP Mismatch
Includes data since 1/3/2025
IP mismatches are instances where the intended targeted IP differs from the actual delivery IP. These are not flagged as IVT unless an entity has a high concentration of mismatches. Mismatches are typically due to configuration errors, such as an impression's associated BidRequest object containing an incorrect or incomplete IP address.
You should address these configuration issues as soon as possible so HUMAN can continue to detect actual fraud when it occurs. Including accurate IP data in the BidRequest object also helps buyers make informed purchases for ad space, as they will know where the impression will be served. If mismatches persist, ads may be served to unintended locations, leading to lower engagement, wasted ad spend, and reduced return.
We suggest reaching out to any impacted publishers to discuss the mismatch and investigate the underlying cause.
Removed Apps
Includes data since 1/3/2025
Removed apps are apps that have been recently removed from the iOS, Android, or Roku app stores but continue to serve ads. Sometimes apps are removed from the store because they violated that store’s policy. If an app has been removed and still has traffic coming from it, you may want to stop serving inventory from that source because that traffic could be low quality or non-compliant. Ultimately, this is a business decision for your organization to make.
Note
Removed Apps only shows apps that HUMAN has seen at least once before they were removed. To identify apps that may have been removed prior to HUMAN detecting them or apps that have never been listed, select View in Explore and run a report with the Last Status and Last Found fields blank.
Not all app removals are due to policy violations, which is one reason why this insight isn’t considered IVT. For example, a person could be using a legacy version of the app, generating legitimate traffic, but the developer may have taken down the app of their own volition. HUMAN can’t determine if an app was removed due to a violation or some other reason, so it’s up to your organization to investigate and decide how to address app removals when they occur.
To determine if you should take action, we recommend considering the app’s IVT rates, the Last Found date, and any other signals your organization uses to determine quality. If these metrics are above your thresholds, you may want to stop serving inventory from that app.
Paid & Social Sourcing
Includes data since 1/3/2025
Paid and social sourcing refers to instances where sourcing for web domains has a high percent of traffic coming from paid or social referrals rather than organic traffic. HUMAN defines paid and social sourced traffic differently:
Note
Paid and social metrics are mutually exclusive. This means that traffic from a domain can either be paid or social, but not both.
- Paid sourcing: Traffic from paid referrals, or traffic generated through paid advertisements or promotional activities. These include paid advertising campaigns like display ads or search engine marketing.
- Social sourcing: Traffic from social media platforms. HUMAN further considers social sourcing as:
- Paid social sourcing: Traffic that comes from paid advertisements on social media platforms. This is specific to paid advertisements or content sharing on social media and is considered separate from paid sourcing.
- Organic social sourcing: Traffic that comes from non-paid, organic interactions on social media platforms. This includes organic traffic from a post going viral.
High rates of paid and social sourcing aren’t indicative of IVT but can indicate low quality traffic. However, not all traffic that’s labeled with a high rate of paid or social traffic is low quality. For example, a content piece about your organization going viral would contribute to a spike in social traffic. HUMAN can’t determine the cause of paid or social traffic, so it’s up to your organization to investigate and decide how to address paid and social sourced traffic when these occur.
If you see an unusual amount of paid or social sourced traffic, you can decide what your organization’s threshold is and take action accordingly. For example, if there are domains that are generating higher rates of paid traffic than your business would like, you can block those domains.
Incentivized Traffic
Includes data since 2/20/2025
Incentivized traffic refers to instances where the relevant traffic comes from a source where the user has been paid with legal tender to generate the traffic. This traffic is typically considered lower quality than traffic generated from genuine interest.
Note
Incentivized Sites insights are only applicable to customers sending their full URL in MediaGuard pre-bid requests.
Incentivized traffic is different from:
- Automated incentivized traffic: This is considered IVT because of the presence of automated browsing. For more information, see IVT Taxonomy.
- Rewarded traffic: This refers to when users are offered non-monetary, in-context rewards for viewing an ad. This traffic is usually declared with clear indicators, so HUMAN doesn’t classify it the same way as incentivized traffic.
Select the Incentivized Sites and Incentivized Apps tabs to view data from each source.
If you see an unusual amount of incentivized traffic, you can decide what your organization’s threshold is and take action accordingly. For example, if there are sites or apps that are generating higher rates of incentivized traffic than your business would like, you can block those sites or apps.
MFA Domains
Includes data since 9/16/2024
MFA ("Made for Advertising" or "Made for Arbitrage") inventory features various characteristics that suggest that a site is designed specifically for excessive monetization by advertising rather than the delivery of content. MFA is challenging to identify because it has no universally accepted definition. However, it does have generally recognized characteristics (e.g., variable ad load when coming to the site from different routes, efforts at cloaking), so we do have some ways to identify these sites.
HUMAN's solution is primarily focused on exposing the key characteristics of MFA rather than using more subjective or arbitrary elements to identify MFA sites, as those elements can produce false positives. We help you identify sites and domains that meet threshold requirements to be considered MFA. For a list of MFA indicators and their definitions, see MFA Indicators.
We evaluate all MFA indicators at the overall site and specific domain levels in order to provide granularity to our MFA classification. This is important as there are cases where only specific domains within a site actually engage in MFA behavior. When we're talking about site-level vs. domain-level MFA detection, the "domain" is the specific root domain or subdomain, while the "site" is the root domain and any/all subdomains taken together. Think of it as the “specific domain” vs. the "whole site." For example, jpost.com, www.jpost.com, m.jpost.com, live.jpost.com, and magazine.jpost.com are all different domains but are all part of the same site—jpost.com.
A substantial portion of sites flagged as MFA based on automated indicators also go through a manual review by HUMAN to ensure a high degree of accuracy of the overall domain list. These manual reviews help us determine whether a website should be deemed MFA. This includes assessing 3+ full URLs per website to determine what indicators are present on a page. When an indicator shows “Manual Review” on the Explore page, it means that there may or may not have been sufficient clarity from automatic indicators alone. “MFA Overridden” indicates that the automatic flagging of indicators would typically result in an MFA designation, but our manual reviews determined that the site or domain wasn’t MFA.
Updated 3 months ago