Metrics Focus: Verify Your Paranoia

Ad spend should not be a matter of trust

Metrics-Verify Your Paranoia

Few professionals are as willing to trust their business partners as media practitioners. This quality is especially obvious during industry love fests - also known as upfronts - where billions in media spend are distributed in advance, primarily based on mutual trust between the buyers and the publishers. With all the things that can and do go wrong, it seems hard to believe that after the ink dries on the contract, all TV spots will run during the right programs, every ad will appear exactly when and where it belongs, and each message will reach its intended audience. Yet media practitioners take that leap of faith every day.

It feels good to be a nice guy, but consider giving paranoia a chance. As the digital vs. traditional balance of a media plan begins to tip against the latter, so, too, increases the likelihood and extent to which the client's money disappears. The advanced targeting and placement alternatives in online advertising today - from roadblocks and frequency capping to precise geographic, demographic and behavior targeting, and to direct audience acquisition through the use of ad networks and exchanges - all rely on a myriad of correct technical executions by publishers and ad networks. Herein lies the unpleasant side effects of digital media: unprecedented opportunities for human error and abuses of trust.

Maybe your ad makes it to the exact audience you wanted, but after running out of U.S. visitors to whom to broadcast your ads, an unscrupulous publisher starts serving impressions to international audiences. Sure, people in Australia can learn about your product, but, even if it is available in Australia, the Sydney (not New York) budget should cover the funds for its promotion in the Land Down Under. Even worse, a non-transparent network might place your ad with the intended consumer demographic, but that might be on a porn site or somewhere equally unsavory.

Avoiding these professional nightmares is the business proposition of several verification companies in the marketplace. DoubleVerify, MediaTrust, AdXpose, AdPepper, AdSafe and Lucid Media are the leading firms offering services designed to minimize uncertainty in online advertising. They do so by creating a transparent view into every ad's lifecycle and providing regular updates on when an ad appeared alongside prohibited content. 

Metrics-Verify Your Paranoia

How is content determined to be inappropriate in the first place? Currently, the main approach is to periodically crawl and read the prospective content, categorize pages and flag sites based on predetermined ratings. Semantic technology that can read the context of the page in real-time exists, but is not yet built to the necessary scale needed.

Prevention methodologies typically include IAB-compliant crawlers that do not inflate the numbers of unique visitors and tracking pixels that identify where an ad is or will be served. For example, verification company DoubleVerify employs "virtual visitors," advanced bots that analyze the domain and HTML structure of a Web page, visually capturing its advertising placement and creatives. Unlike other crawlers, "virtual visitors" execute all the code on the page (exactly like any human-used Web browser). They come back with screenshots to prove the occurrence. This helps advertisers identify and address publishers or ad networks that incorrectly serve ads - and remedy the situation.

Virtual visitors and crawlers can convincingly imitate real people and thus uncover errors in publishers' geo-targeting (by crawlers originating from different locations) and sometimes even demo-targeting (by altering the crawler's cookie structure).

The ideal way is, of course, to block the ad before it even appears in a problematic environment. Real-time blocking technology already exists and is offered by Lucid Media, AdPepper and AdSafe. However, it carries certain drawbacks such as outsourcing vital campaign decision-making to machines and the problem of what to do with the blocked impression. There is usually a default PSA banner on the vendor's server available to take your ad's place, but that does not solve the issue of ad-serving cost, and if the machine is wrong, the cost of missed opportunity, as well.

Once it is established that nothing was technically wrong with the campaign, a whole other set of questions tends to arise. How many people did my campaign reach and did they belong to the desired segments? Which placements and creative drove reach against my targets? What were the dynamics of reach and frequency build-up throughout the campaign?

The in-depth probing of the audience makeup touched by ads remains in the domain of a different group of companies, those that either control large metered panels, or control very large online verified databases. comScore and Nielsen verify campaign audiences by digging into their metered panels for panelists exposed to campaigns. The panelists' sample is then weighted and extrapolated against the U.S. census to provide a representative estimate of basic demographic makeup of the campaign audience, including reach, frequency, and campaign TRPs, along with metered online behavioral insights.

In Quantcast's case, the campaign audience makeup is inferred from analyzing cookied online behavior and other sources such as isp registration data, custom market research samples, etc.

Aperture, a division of Datran Media, developed a process for setting rich cookies at various locations on the Internet where there is user permission. Protecting privacy is a key element of the design of Aperture, and its process for setting cookies cannot tie back to an individual.  The cookie-dropping constantly occurs; it enables Aperture's platform to have 50 million active cookies at any given time. The audience-verification methodology relies on matching these cookies to IXI, Experian, Acxiom and other industry standard databases, as well as vertical specific and custom customer data sets to provide sizable overlap with campaign reach. In Aperture's case, matching cookies with data sets is direct - no "look-alike" modeling involved.

Aperture's large sample sizes enable marketers to accurately estimate the audience that views, interacts, or responds to an ad down to the placement or page level, which makes it easier to tie online data with client and other data sources for further insights.

If we use these tools, we can finally hold the publishers fully accountable for avoiding offensive or inappropriate content and delivering ads against the agreed-upon placements and audience targets. Once the industry achieves full transparency, then we can trust our business partners.

Recommend (30)
1 comment about "Metrics Focus: Verify Your Paranoia".
  1. Kirby Winfield from comScore , August 9, 2010 at 7:58 p.m.

    Great piece Yaakov. I would be remiss if I did not point out the AdXpose offers blocking as well, and has done so since January 2010. Also, AdXpose has solved the problem of what to do with that impression besides serve a PSA...but that's another story :)