Behavioral Optimization

By Published on .

Most Popular

It's almost impossible to pick up a media trade publication today and not see an article about optimizers, the intriguing software applications enabling media planners to analyze vast combinations of television programs and prices. Optimizers are designed to assist agencies in making more informed decisions about which mix of networks, dayparts, and syndicated TV shows yields the highest reach per ad dollar for their clients' brands.

While the current wave of optimizers has rekindled the longstanding debate about the effectiveness of TV advertising-and that's never a bad thing-they fall far short of the mark. And it's not just because many of these off-the-shelf products were designed for use in the European media market, or that there isn't yet enough quality respondent-level TV data to fuel these optimizers.

The main problem is that today's optimizers are being used to optimize the wrong thing!

These optimization programs were fundamentally designed to maximize the reach and/or frequency of a brand's TV schedule. Reach and frequency refers to the number of unduplicated viewers who see an advertisement (reach) and how often they see it (frequency). Getting precisely the right ratio of R to F without wasting media budget dollars has been a chief concern of media planners since computers allowed for quick analysis of a media schedule's reach and frequency. However, there are at least two other critical elements of the marketing equation-brand sales and advertising response levels-across consumer groups that are escaping the current optimization bandwagon altogether.

Maximizing only reach and frequency is inadequate because it perpetuates an inherent disconnect that has long prevailed in the field of media planning. Marketers create plans based on consumer purchase behavior. In fact, they go to inordinate lengths and spend many dollars to understand the underlying drivers of sales within a given geography. It is critically important to understand where sales volume is coming from: loyal brand users, heavy category users, and so on. Understanding the drivers of brand sales levels often facilitates the development of the brand's marketing strategy and, quite often, forms the basis for the creative strategy as well. But advertising agencies create plans based on the TV viewing behavior of people segmented by their age and sex profiles, not their purchasing habits. In other words, if you're selling bar soap or pre-sweetened cereal, women aged 18 to 49 have long been a likely media target.

Numerous academic studies have shown that age/sex descriptors can account for only 15 percent or so of the variance of consumers' purchase behavior, whereas past behavior can help to predict upwards of 80 percent of future purchase behavior. To be fair to the advertising media community, the process of planning media has been an evolutionary one, and the only true measure of the value of a TV program or daypart has historically been the number of people tuning in. However, why should the objectives of the marketing plan (designed to achieve a given sales level for a brand) and the objectives of the media plan (designed to achieve a given level of gross ratings points for a given budget) be so divorced from each other? If a marketer and a media planner watch the same baseball game, does the marketer believe that the winning team scored the most runs and the planner think that the winner threw the most pitches? Before any optimization process can occur, there should be agreement on what is being maximized.

The optimization of reach against a particular age/sex-defined demographic segment seems even more questionable given the availability of advertising response estimates from marketing mix models. Marketing mix models estimate a brand's incremental volume due to various marketing stimuli. Some of these models can even estimate how different consumers respond to TV advertising. For instance, some models have shown that upscale urban households with children are nine times more responsive to TV advertising for yogurt than are blue-collar, rural households with children. Having access to this information should clearly influence the construction of a yogurt brand's TV schedule.

Lastly, the disconnect is compounded when multiple brands are involved, because it's common for individual, demographic-specific brand needs to be subordinated to the cost-per-thousand objectives of a broader corporate buy. Such a consolidation into a single corporate demo greatly erodes a planning optimizer's results, raising the question: What good is a planning optimizer if the buying process limits any efficiencies gained during planning optimization?

Truly targeted, efficient, and effective media plans should be evaluated for both age/sex demos and behavioral-based targets. At MediaPlan, Inc., we're in the development stage of what we hope is the cutting edge of behavior-based optimization. We're converting traditional demographic TV ratings into behavioral ratings using product purchase data from Information Resources, Inc. (a 55,000-household panel) and proprietary MediaPlan, Inc. methodologies. In this manner, media plans for TV can be executed against a consumer behavior profile, in addition to traditional age/sex demographics.

What's truly exciting is that, for the first time since store scanner data became available in the 1980s, we're finally on the verge of developing the sophisticated analytical tools to fully exploit the data in the everyday world of the media planner/buyer. And by engineering an intersection of consumer behavior and traditional age/sex demos, we will help marketers realize a greater return on media dollars invested.

In this article: