Measure the Web Like TV and Brand Advertising Will Follow
The web changes at lightning speed, but online advertising remains stuck with the same old metrics we've had since the beginning. "Unique visitors," or 30-day cumulative reach, holds firm as the audience measure. The currency for buying and selling continues to be "ad impressions," which are generally disconnected from any meaningful audience dimensions. Worst of all, the reigning metrics for campaign efficacy are all tied to observable outcomes, as is typified by the dreaded "click-through rate."
Outcome-based measures completely ignore the very real and powerful ability of advertising not just to change people's behavior, but to change people's minds. Indeed, much of what we call online advertising isn't really advertising at all -- it's direct marketing.
Fortunately, there is a way for us to break out of this direct-response ghetto and demonstrate the power of the web as a true advertising medium. The key is to measure web audiences in the same way that TV audiences are measured -- gross rating points, or GRPs. Such a move would provide brand advertisers with a comparable metric to evaluate their media buys across TV and the web. It would provide proof that online display, bought with sufficient reach and frequency, can be as effective as TV at moving brand health measures and cross-channel sales volumetrics.
The downside for publishers is that our audiences will look a lot smaller than they do under the prevailing system. But let's face it: "unique users," calculated on a 30-day basis, is a pretty meaningless measure. To be counted as part of a website's audience, all a person needs do is view a single page once a month. It matters not whether the page is viewed for an instant or for an hour. It doesn't even matter whether the page contains any advertising! If broadcast networks used a comparable metric and reported the number of viewers who surfed past their channel once each month, they would all have roughly the same audience and it would be roughly equal to everyone in the U.S.
Thirty-day cumulative reach was a brilliant strategic choice as the original online audience measure. When the web made its commercial debut in the mid '90s, the user base was understandably quite small. Media buying, based on reach, was a well-established practice, so it made sense to exaggerate the size of our new medium by choosing a long duration over which to measure the audience. Each year, as more consumers became hooked on the net, websites drew hockey-stick charts showing the dizzying growth of their unique users. This created a feel-good mania that worked for everyone. Investors felt secure that their venture capital was being put to good use, mainstream- media companies took pride in their precocious digital progeny, and media buyers, for whom the metric was originally established, had a rudimentary method to rank sites based on their reach.
A lot has changed in the past 15 years. The web has become pervasive. Consumers use it all day long -- at work, at home and on the go. As a result, the audience by day part, in aggregate, now has sufficient weight to compete favorably head-to-head with TV. The time has come for the web GRP.
Online publishers are often evaluated (and compensated) not so much for the audiences they reach as for the results they deliver. Adopting the web GRP would help correct this situation, because it would provide a method to unify the planning currency (reach) with the buying currency (impressions). The current disconnect between planning and buying is a vestige of our original choice of audience metric, and stands as a major reason that online-advertising efficacy has been judged almost exclusively by direct-response metrics.
The advent of behavioral targeting has helped buyers begin to make the connection between the impressions they buy and the users they reach and helped move the industry in the direction of selling audiences.
This is a positive development, but the method by which behavioral segments are constructed relies principally on inferring the intention of users based on keywords they search for or pages they view. This approach has two serious limitations. First, it minimizes the opportunity for individual publishers to deliver segments of sufficient scale within their own sites, thereby pushing advertisers into the arms of multisite aggregators such as Google. Second and more importantly, behavioral segments based solely on inferred user intent keeps the microscope of online-advertising effectiveness focused on the bottom of the marketing funnel, ignoring the very real role that display advertising plays in demand generation.
Demand creation happens at the top of the marketing funnel when brilliant creative ideas are coupled with the tactical buying of reach and frequency, broadly targeted. Advertisers buy the audiences they plan for, and media owners are paid for delivery of those audiences. If we adopt the web GRP as the standard for measuring online audiences, we will be able to clearly demonstrate that online is as effective at the top of the funnel as it is at the bottom. When we do, online advertising will attract the branding dollars that have largely eluded the marketplace thus far.
Here is a blueprint:
Name it. All industry initiatives need a name. Let's call ours the "Werp" (WRP).
Creative unit. TV uses the 30-second spot. Let's start by allowing any of three IAB approved units to count: the medium rectangle, the large rectangle, or the half page.
Time and clutter. TV's use of the 30-second spot solves two problems that have plagued us online. It gives advertisers 100% share of voice, and it (sort of) guarantees an exposure time. Let's propose that for an ad exposure to count towards a WRP, it must be the sole marketing message in focus for a minimum of five seconds.
Calculating WRP's. Works exactly the same as GRPs. Reach frequency.
Audience reporting. Eventually, we'd want to have WRPs by day part. But as a modest beginning, and so as to gracefully transition publishers from the 30-day cumulative measure, let's start with total weekly WRPs at the site level.
Let me know what you think!
|ABOUT THE AUTHOR|
Michael Zimbalist is VP-research and development operations for The New York Times and tweets at twitter.com/zimbalist.