The Customer is a fast growing, RTB-enabled, Ad Exchange Network that wanted to move beyond ad delivery, verification and post-impression analysis to provide product differentiation through pre-emptive brand safety and improved transparency for its customers.
RTB ad tech platform with over 20 billion impressions monthly (“Customer”).
The Customer had gained considerable traction and market share since the initial launch of their platform offerings. However, to continue their revenue growth rate, the Customer had to demonstrate meaningful steps to alleviate brand safety concerns across their platform.
The Customer identified several issues related to brand safety that were critical to adding new and larger advertising customers, as well as retaining existing customers, including:
|Ads being placed on inappropriate websites – such as adult, pornographic and malicious or compromised websites. As websites become more and more dynamic, with content changing more frequently and sometimes continuously, ads were increasingly being placed on websites with inappropriate content|
|Page-level visibility deep within sites – lack of visibility to content categories deep within social network, blogging and related sites made it particularly challenging for large online ad campaigns to effectively target the most popular websites and domains.|
|Real-time checks for malicious and compromised websites – because of the constant state of flux of malicious sites (they can change rapidly from “clean” to “,inflected”) ads were being placed on infected sites, unbeknownst to The Customer, while sites that had been previously flagged as malicious and had since cleaned up were not being targeted because of delays in getting updated data on these sites.|
The Customer believed that successfully addressing these critical brand safety issues would provide a significant competitive advantage.
“Our existing platforms didn’t have the URL database support required to perform pre-emptive categorization look-ups at the volume of impressions our business required. Further, we had very limited visibility to compromised or malicious website data or any way to perform real-time validation of a URL being clean or infected at the specific time we were looking to run an ad. These were the features our advertisers were telling us they required.”
– VP Business Development, Customer
Once the Customer made the decision to focus on developing and emphasizing pre-emptive brand safety, they identified the need for a high-performance URL database with real-time content categorization and malicious website detection, as well as the ability to distribute the database across the dozens (and growing) of servers on which the platforms operated. After identifying zvelo as a possible vendor, the Customer undertook a thorough investigation of the following:
|Coverage & Accuracy – an evaluation of the zvelo’s URL database coverage and accuracy for the Alexa global top 1 million list. The Customer also conducted a test of specific adult, pornographic, violence and other websites that had been sources of brand safety issues in the past.|
|Coverage for Social Websites – an evaluation of the inclusion of sub-domain and sub-path URLs for social websites (such as Facebook.com, Twitter, YouTube, etc.) within the database, including the ability to distinguish different content between different sub-domains.|
|Malicious Website Detection – an evaluation of the vendors’ coverage and detection abilities of malicious websites was performed, using a range of internal/external sources for websites identified as hosting spyware, malware or other threats. Additionally, an evaluation was performed to test zvelo’s capabilities at real-time detection when websites went from clean to infected and from infected to clean.|
|Website Revisit and Recategorization – a test was performed to determine how often websites with changing content were revisited and their content categorization updated.|
|zveloDB SDK Distributed Deployment and Performance – the zveloDB SDK was tested for ease of integration and deployment across the servers operated by the Customer, as well as the ability of the zvelo to successfully meet the high URL query volume requirements forecast by the Customer.|
|Category Mapping – the Customer tested zvelo’s ability to create specific category maps for use by RTB and DSP platforms, as well as support for the IAB Tier 1 and Tier 2 category sets.|
The Customer finalized the relationship with zvelo after determining zvelo was able to successfully address each of these requirements.
“We were able to put zvelo through the paces on our requirements and they stood up to the testing very well. Their coverage, accuracy and malicious website detection gave us confidence that we could indeed target a superior brand safety offering with zvelo. Further, the flexibility of the zveloDB SDK to be deployed across the dozens of servers that support our platform systems, combined with the outstanding responsiveness of zvelo from initial contact through testing, integration and launch, convinced me they were the right partner for us.”
– VP of Technology, Customer
Customer’s DSP and RTB Platforms in Action:
The Customer finalized implementation of the zveloDB SDK across their servers, using a “primary” zveloDB SDK to communicate with the zvelo’s cloud network for updates, real-time URL categorizations and zvelo Instant Protection™ (zIP) – a real-time malicious website detection feature. The primary zveloDB SDK also acted as a conduit for providing updates to the distributed databases on other servers.
The Customer’s platforms query the zveloDB SDK using URLs for pre-emptive brand safety verification (e.g. is the URL an appropriate category and is the URL malicious or compromised). The zveloDB SDK processes (on each of the servers) work with the following processes:
|Categorization DB – this includes URLs that the Customer has defined in blacklists and whitelists.|
|API Pipeline – real-time URLs categorized by zvelo’s cloud systems as blockable (porn, adult, violence, hate, etc.) or malicious. The zvelo Instant Protection (zIP) feature immediately updates the local zveloDB SDKs with any such categorizations made by the categorization and malicious website detection systems.|
|zvelo DB – provides access to hundreds of millions of URLs in nearly 200 languages covering over 8 billion web pages.|
URLs not found in any of these databases are delivered to zvelo’s cloud systems for real-time contextual content categorization and malicious detection analysis. In the event a URL is categorized as blockable, inappropriate or malicious, the URL and the associated category values are immediately distributed to the zveloCACHE of the zveloDB SDKs.
zvelo’s ability to harness the collective web activity of all of the URL queries and users provides the basis for the extremely high coverage and dynamic changes of the websites, including social sites, domains, malicious websites and more. Each additional customer and user increases the breadth of websites visited and categorized, thereby further increasing the coverage and malicious website detection for all of the collective users.
The Customer was able to secure new customers and advertisers through aggressive promotion of the pre-emptive brand safety messaging, backed up by successful results in testing and ad campaigns. Additionally, the Customer was able to significantly reduce the time and energy spent on damage control due to ads being placed on inappropriate or malicious websites. The positive brand safety messaging and positioning are supporting the Customer’s continued rapid growth trajectory.
“We have been able to successfully target an opportunity in the market for a differentiated, pre-emptive brand safety message and offering by virtue of integrating the zvelo URL database and zveloNET technologies. We have big growth plans and zvelo is a considerable part of supporting these efforts.”
– CEO, Ad Exchange