Share on LinkedIn
Share on XIAS is committed to delivering the advertising industry’s most trusted and transparent media measurement and optimization offerings, and we will continue to take actions towards protecting customers from fraudulent actors.
On March 28, media consultancy Adalytics issued a report containing unsubstantiated and misleading claims about Integral Ad Science and our fraud detection capabilities. That same day, we published a blog post titled “Update: IAS’s Multi-Tiered Approach to Combatting Ad Fraud” to address and clarify these initial inaccuracies. IAS continues to have serious concerns about these claims and the flawed assessment techniques upon which they’re based.
Upon further analysis and due diligence by our technical teams, the following summary aims to correct misleading claims in the report and provide additional context on how IAS detects and classifies invalid traffic.
Overview of Key Terminology
- General Invalid Traffic (GIVT) Ads: Legitimate bots like search or large language model (LLM) ‘bots’ are tracked by the IAB in its “Bot User Agent List,” and IAS leverages its list of known spiders and bots to identify benign General Invalid Traffic (GIVT). This is traffic that failed to meet ad quality criteria filtered through and caught by known IP and bot lists (ex: Google web crawlers). GIVT is always invalid traffic.
- Sophisticated Invalid Traffic (SIVT) Ads: Ad fraud bots, otherwise known as Sophisticated Invalid Traffic, or SIVT, are malicious and designed to go undetected. SIVT and other ad fraud schemes are outlined and defined in the IAS Ad Fraud Glossary blog post. This is traffic that failed to meet proprietary and sophisticated invalid traffic detection criteria. Examples of failed traffic can include: hijacked devices, ad tags, or creative; adware; malware; and misappropriated content.
- Reduced Value Inventory (RVI) Ads: Traffic that IAS flagged as being generated in a way counter to the advertiser’s interests. Possible examples include the use of incentivized browsing and “proxy servers.” RVI impressions are flagged in our systems to allow advertisers to evaluate the performance of these impressions. Some impressions that may be initially flagged as RVI may later be flagged as IVT based on our fraud detection.
Findings on the Bots Described
Adalytics calls out three bots: two known (URLScan.io and HTTP Archive) and one unknown (anonymous). These types of bots have constantly shifting IPs underneath them, meaning the bots will work through new IPs on a consistent basis. IAS fraud detection systems are able to identify and classify both URLScan.io and HTTP Archive.
Addressing the Unsubstantiated and Misleading Claims
Claim 1: IAS treated URLScan.io as valid traffic in 77% of tests.
IAS Response: IAS classified URLScan.io impressions as GIVT, SIVT, or Reduced Value Inventory (RVI). These bots rotate through undeclared and constantly shifting IP addresses. To prevent false positives, our fraud models rely on impression level thresholds to confirm fraud. Once that threshold is met, impressions are blocked and the IPs are incorporated into our fraud solutions.
Claim 2: IAS treated HTTP Archive as valid traffic 16% of the time.
IAS Response: IAS flagged HTTP Archive traffic as GIVT.
Claim 3: IAS publisher tools simultaneously flagged the same impression as both valid and invalid.
IAS Response: External analysis cannot accurately read our backend adjudication. As stated previously, fraud detection by its very nature requires that bad actors or external sources do not have access to public information that would allow them to create new threats which in turn circumnavigate fraud detection technology. That is why we collaborate closely with industry partners and customers to stay ahead of bad actors.
Claim 4: Publishers served ads to known bots even after they were flagged.
IAS Response: IAS provides the capability to block IVT, but publisher partners choose whether to enforce those blocks. As stated previously, only IAS and its customers have access to settings configured for their campaigns — including whether or not they are using IAS’s pre-bid IVT avoidance or have chosen to block IVT real-time prior to creative rendering.
Our Commitment to Trust and Transparency
At IAS, we constantly innovate to create media measurement and optimization products that help maximize digital ad spend, drive efficiency and deliver real results, as well as protect customers from fraudulent actors. We partner with industry bodies and collaborate closely with our advertiser and publisher customers to continuously improve our solutions. As always, we are committed to working in partnership with the industry and other companies to reduce the proliferation of bad actors while preserving the open internet.
You can read more about IAS’s Multi-Tiered Approach to Combating Ad Fraud here.
