Addressable TV: Myths and Realities

Will TV spots soon be as targeted as online ads?

Anyone who’s been involved in the “advanced media” (read: iTV) space knows the industry has a penchant for announcing vaporware and speaking of the future in the present tense. Often, it’s difficult to cut through the unbridled enthusiasm and get to the reality of what’s deployable today. With that acknowledgement, it’s appropriate to turn our attention to the concept of addressable television advertising.

Marketers are perhaps able to grasp no other facet of advanced media as quickly. The ability to deliver customized messages to households (or to individual set-top boxes within a household) based on specific knowledge about that household (income, ethnicity, presence of children, purchase habits, etc.) is an ideal for most advertisers. Naturally, two immediate questions come to mind: Does the inherent premium associated with addressable advertising deliver incrementally better results? And, how can I personalize my current television campaign to make it relevant to different market segments?

Before looking at the current and future possibilities of addressable advertising, it’s helpful to review what we know already. Arguably, the most thorough addressable testing program conducted to date is the Comcast SpotOn test in Aurora, CO (SpotOn is a solution offered by OpenTV). The test was conducted nearly a year ago and included a number of blue-chip marketers: General Motors, Johnson & Johnson, Coca-Cola, the U.S. Army, and Home Depot.

A few relevant facts about the program:

• Four networks participated: ESPN, TBS, Lifetime, and Discovery.
• The first pod position was the only addressable spot.
• “Command codes” were inserted into the local avail on the four participating networks.
• PRIZM clusters (geographical level) were used as the targeting criteria.

There are two basic addressable architectures in television advertising. One is the “split run” approach, in which one advertiser buys one avail and can send different creative to different homes tuned to that avail. In this instance, the brand must have multitargeted creative to get the maximum benefit, and the brand pays for all homes tuned to that avail. The Aurora trial used this architecture.

The other architecture is very similar to the Internet’s behavioral ad-serving model: The advertiser and seller agree on a price per target impression. One avail is split among multiple advertisers, and the advertiser pays only for the homes in the target (i.e., meet the targeting criteria) the ad is served to. The advertiser can have multiple creative executions in this instance as well. The advertiser pays only for the targeted impressions delivered. Though I’ve not personally seen this in action, we’re told this model can currently be executed in a few cable systems around the country. (This doubting Thomas will believe it when he sees it.)

Without getting into all of the limitations that plagued the Aurora trial, there are a few key takeaways. PRIZM geographic data can be misleading. For post-analyses, the PRIZM data was compared to Acxiom data and was shown to be a poor indicator of true target identification. Had Acxiom (or Experian) data been used, results would have been greatly improved.

Additionally, addressable spots had very limited “switchaway,” meaning a majority of the audience was retained during the commercial breaks. Advertising avoidance continues to be a growing problem for advertisers. Addressability, which increases relevance, could ultimately translate into significant audience retention.

Current opportunities and near-term advances look very promising for addressable advertising. Companies such as Visible World, SeaChange, and OpenTV enable cable operators to offer addressable opportunities to marketers. Ad-serving companies, such as DoubleClick and aQuantive’s Atlas, are also talking to cable operators about ad serving into their broadcast systems. As all things become digital, the differences between platforms become less and less important.

Do you believe the hype? I’ll leave that to you. Today, we have the ability to target unique ad messages to limited (ZIP + 4) geographies. Imagine what this could mean for local retailers working to drive store sales or national advertisers with customized local promotions. Very cool stuff. Individual household targeting and dynamically generated ads aren’t just a subject for science fiction. They’re just around the corner.

I love this business!

Nominations are open for the 2004 ClickZ Marketing Excellence Awards.