RealTime IT News

IAB Unveils Ad Impression Guidelines

The Interactive Advertising Bureau made good on a years-old promise to deliver standards for measuring and auditing ad impressions and unique users -- though it refrained from handing down rules on at least one key issue.

"Today is a great day for the industry," said IAB chief executive Greg Stuart. "This is a major step forward ... the goal is to make media buying easier to do and profitable."

The New York-based IAB, working with groups including the Media Ratings Council, the Audit Bureau of Circulations and the Advertising Research Foundation's Digital Media Measurement Council, commissioned PricewaterhouseCoopers to study the procedures and best practices of eleven leading ad servers, portals and major publishers. The MRC, ABCi and IAB analyzed and consolidated those findings into a list of definitions and guidelines.

Most critically for the industry, the guidelines lay down rules for how best to measure ad impressions -- a mammoth task, considering that there are a multitude of different ad serving systems and procedures in practice.

Still, there is work remaining: the IAB decided on a bi-partite rule on ad impression measurement.

For ad sellers that use server-initiated ad counting -- that is, counting which uses the publisher's Web content server for making requests and redirects -- an "ad impression" is counted after the site's ad server or content server delivers an ad to a browser.

Client-initiated ad counting, which relies on the user's browser to fetch ad content, requires that an impression can be counted after an ad is requested from the publisher's ad server, or from a third-party ad server.

Obviously, the question remains about which process is the more accurate. ABCi senior vice president Dick Bennett said his group is looking to come up with exact differences between server-initiated and client-initiated measurement options. Stuart added that preliminary findings are expected by the end of March.

Publishers also will be measuring "page impressions," which are counted at the content-server level. An important metric in the sales process, the IAB's definition of "page impressions" also requires that sites uphold best practices for cache busting and for filtering out bots.

The IAB's best practices for cache busting are a combination of HTTP header controls, tagged URLs, uniquely-named URLs, and use of 302 Redirects. (A "302 Redirect" occurs when a server sends the location of a requested ad or page to a browser, rather than sending the ad or page itself. 302 Redirects are not cacheable under some circumstances.)

Best practices for filtration include either a combination of Robot Instruction Files and robot exclusion based on user-agent strings (such as specified in the IAB/ABCi Industry Robot List) or 302 Redirects -- though the IAB concedes the effectiveness of the latter procedure needs to be further studied.

Additionally, under the guidelines, pop-ups, pop-unders and interstitials count as generating "ad impressions" rather than "page impressions."

The IAB's guideline for counting click-throughs, a less controversial issue, specifies that they are tracked and reported at the ad server, generally based on a 302 Redirect.

Mouse-overs and clicks within rich media ads (that don't normally result in the loading of a new page) can be recorded either via 302 Redirect, or on a delayed basis.

The IAB also reiterated its previously-released terms for "visits," "users," and "unique."

A "visitor" is defined as a browser having downloaded content from a site, without 30 minutes of inactivity. The definition is the same for a "user," which also includes "pushed" content. Both visitors and users are best measured by registration, or, should a site not ask for registrations, by a combination of tracking cookies, unique IP addresses and user strings.

Calling a measurement "unique" entails ensuring that visitors/users are not counted more than once, which is largely handled through the registration process, or by the use of persistent cookies.

The guidelines also specified that media companies and ad servers must fully disclose and explain their recording process, data collection processes and filtration processes (for spiders, bots and search engine activity). They also must break out figures for pop-under/pop-up impressions, interstitials, surveys and HTML newsletters.

Industry groups have been working on such standards for years. Drafts of the guidelines have been knocking around IAB task forces since mid-1997. Meanwhile, associated groups had their own go at standards, such as those begun in 1999 by the now-defunct Future of Advertising Stakeholders, sponsored by Proctor & Gamble.

It hasn't all been for naught, though: the FAST group's standards and language appears in part in the new IAB standards.

Still, the obvious difficulty in hammering out a unifying document governing impression measurement stems, in large part, from the myriad issues hindering the process. Chiefly, the fact that a host of different computer systems must interact to deliver even a standard static banner ad. Yet the different participants in the process -- publisher, ad trafficker, third-party ad server -- have access only to a small part of the entire system.

Some publishers have equated an impression with a browser's page request, while others look specifically at the request made by their Web server to their in-house ad management server. Then, in many cases, there's a request for an ad either from the publisher's own server, or from a third party -- another opportunity to count an impression.

Some third-party ad servers have favored counting an impression at the point when a user's browser is redirected to fetch a banner from the network's ad "counting server" through a 302 Redirect. The counting server records an impression, then redirects the browser to an image server.

Then again, the actual impression could be recorded at the image server -- or, finally, when the actual image is displayed in the user's browser.

Of course, the entire procedure takes time. So if a user halts the loading of a Web page or clicks to another site, or a technical glitch occurs somewhere along the way that scuttles the process, there's a distinct possibility that one party could count an impression while another does not.

According to some estimates, those discrepancies could cause up to 50 percent of a campaign's purchased impressions to be called into question.

But Stuart said initial findings from the new IAB guidelines suggest that they can reduce those discrepancies by as 70 percent.

Stuart added that these efforts will also serve as "building blocks for future guidelines," including those governing reach and frequency metrics. Many industry-watchers believe that to encourage traditional advertisers and agencies to increase spending on online advertising, Web publishers and ad reps increasingly will sell inventory on the basis of reach and frequency -- which are widely used in traditional broadcast industries.

In addition to capping off a gargantuan industry effort, Tuesday's guidelines come amid a heightened level of activity by the IAB, designed to take down roadblocks hampering the growth of online advertising, and no doubt prompted by the urgency of the space's current woes.

Those steps began in earnest last year, when the IAB elected its first full-time staff and executives, with the appointment of then-CEO Robin Webster. (Webster has since stepped down for personal reasons.)

Last year also saw the IAB produce a Terms & Conditions template designed to simplify and expedite online media buying, new rich-media-friendly ad units, a glossary, and standards for rich media usage.