A new media-mix strategy

By Published on .

The brain takes comfort in simple sorting. It is less relaxed with conditional complexity. For example, I like apples better than oranges. But, if I've already eaten an apple, then I'd prefer an orange. Media planning today is awash in conditional complexity, not because of fruit, but because of media mix. This is a migraine-inducing change in the way media people need to think about what we do.

The opening words of the planner's bible, "Toward Better Media Comparisons," reminds us that media planning is allocation. "This decision is unavoidable, since . . . the use of any medium to any degree implies the avoidance of some other medium . . ." Today we would have to add, ". . . implies the inclusion of some other medium."

The old way of media planning was about picking individual media. New media planning is about picking combinations of media (and permutations of media, where sequence of exposure is important). This increases relevant media choice from a manageable few hundred to an unruly few hundred thousand. It also means comparing apples with oranges. Both tasks are well beyond the abilities of a planner with a notepad. That is the reason for the great interest in media-mix optimizers (MMOs for short).


Media consolidation, such as the AOL Time Warner merger, has focused sellers on cross-media packaging. And the Internet is certainly destined to be a big player in the media mix. But, seller packaging is to media-mix as a Happy Meal is to dinner. Even the largest sellers have limited menus, while buyers need to shop the market.

Media mix is not a selling tool. It's a buying strategy.

This is an encore. We've been there before with TV. Optimizers were a response to three powerful forces in TV: recency planning, fragmentation and sharp increases in prime-time pricing. Recency established reach as the planning goal. Prime time had become too costly for most brands to use for reach. Fragmentation offered the alternative of buying reach through dispersion, but the many possible combinations were more than planners could handle, so optimizers, like SuperMidas and *Xpert, were brought in to help do the job.

These same forces are active in pushing media mix, but the obvious decline of TV is the whip. In the face of strong demand and shrinking inventories, the networks have raised prices and added commercials. Both make TV less effective. Advertisers know this and are seeking options. The move from TV is not rampant, but it is inevitable, and that is spurring agency interest in better media-mix planning data and optimization.


Just as a second beer never tastes as good, a second dollar never buys as much response. That's why the idea of mixing media has always been attractive. It's in the physics of a flattening sales response curve.

Marketing-mix models will often show a high return on investment for a low-budget medium simply because spending fewer dollars puts the advertiser at a steeper spot on the response curve. The general rule is spend more money and the response-per-dollar goes down. Spend less money and it goes up. But, while sales-per-dollar go up, total sales do not, so a brand can't scrooge its way to growth.

Media mix gives advertisers a way of beating the curve. Where market-driven costs per thousand reflect relatively comparable media value, spending fewer dollars in more media will produce a greater response.

There are also the under-explored benefits of focus and synergies. Some media may just communicate better to some consumers (reading people versus viewing people). Mixed exposures may have a greater added effect (sell the car on the TV and the deal in the paper).

These are the arts of media mix. The science is knowing which medium to add and when to add it. CPM alone offers no guidance, CPM and reach-build together do. This is the routine dynamic of optimization.


Because a media-mix optimizer deals with several media, the problems in building one are far more complex than those encountered with TV alone. The big three are database, comparability and frequency value.

The first big hurdle is reaching consensus on what data to use. This is different from TV optimization, where the only choice is Nielsen Media Research. The media-mix database has to contain two things. The "currency" measurement for each medium (e.g, Nielsen Television Index for TV, Mediamark Research Inc.'s for print), so that the optimized schedules will price out. The database also must have a means of estimating duplication between media, so that it can calculate reach. A database with these qualities does not exist, and creating one is a bear.

The simplest approach is to use the currency for ratings and within-medium duplication and calculate the across-media duplication rates from a single-source study like MRI. The weakness of this approach is the MRI recall measurements of TV do not track with NTI meter measurements of TV. As a result, the duplication with TV data will be poor. This is a fatal flaw. Because TV is so important in media mix, the duplication with TV data controls the optimization.

The alternatives are to use random duplication-even though we know cross-media duplication is not random-or to use data fusion.


Data fusion is not just theory. It is currently used in the U.K., Japan, Latin America and Europe. For years, U.S. media researchers have been told that fusing together surveys designed to measure specific media may be preferable to using a single survey that measures them all. Reigning media statisticians like Gerry Glasser of Statistical Research and Marty Frankel of MRI, have long said these statistical techniques could be appropriate in the U.S. Nevertheless, introducing fusion, or any kind of data modeling here, has proven difficult.

There will probably be both an MRI-based single source optimizer (offered by IMS and Telmar), and a Nielsen/MRI fused database (offered by Kantar, which produces the SuperMidas optimizer).

Data access is another media-mix issue. In TV optimization, buyers and sellers had access to the Nielsen database. Turner Cable was especially aggressive in using the new information to switch-pitch the dominant broadcast networks. In media-mix optimization, only agencies and TV sellers are likely to have access to Nielsen. Competing media that can best use the data (magazines, newspapers, radio, outdoor and Internet) may be shut out by its high cost.

But there is a crack of opportunity. If competing media can interest Nielsen in selling limited, affordable NTI data access through the media-mix database, it could open the way to more intelligent competitive selling.

Today when other media attempt to sell against TV, this lack of access (and lack of familiarity with what the Nielsen data show) is a high brick wall.


Because TV is the primary dollar medium for most advertisers, TV will set the data standard for all media. To be a candidate, a medium will need a planning period and a Target Rating Point definition that conforms reasonably to those used for TV. The critical TV planning period is the week, so a full-function MMO will have to do weekly-reach planning for all media. This requires data reporting magazine-issue readers by week and the ability to calculate the week-by-week reader duplication with other magazines and other media. Weekly planning is simple for time specific media, such as TV, Internet and radio, but it needs to be developed for out of home.


Media mix raises the quiet issue of CPM comparability to a shout. Agencies are far more willing to accept it within a medium (that a cable :30 is equal to a broadcast :30) than across media (that a cable :30 is equal to an Internet banner).

A wealth of experience with media-mix optimizations done in the 1970s showed that without comparability weights, or predefined dollar allocations by medium, the plans produced were so counterintuitive, they were usually deep-sixed.

This flags the importance of adjusting CPMs for media value before attempting to optimize a mix of media.

Several kinds of CPM adjustments need to be considered before an optimization program can produce an intelligent schedule:

Probability of exposure to equalize ratings across media. Out-of-home showings and traffic counts, for example, are far looser measures of exposure than TV's "average minute audience."

Ad exposure weights to equalize the probability of an ad being seen. What percentage of the viewers of "Friends" see the average commercial? What percentage of the readers of Time see the average ad?

Communication weights to equalize the probability of an ad message communicating. For example, a magazine color page compared to a TV 30-second spot. It is generally what we mean if we say "TV, with sight, sound and motion, is more effective than print."

Frequency weights to establish the contribution of the first and each successive exposure in obtaining a response. This is what makes an optimizer value reach over frequency, or vice versa.

Synergy weights to consider the value of exposures in different combinations of media in obtaining a response. It is what we mean when planners say, "Print is more effective after the TV has run," or "Fifteens work better when they follow :30s."

Synergy values add a complication to frequency values, but they are essential, because synergies appear to be an important reason to mix media.

In practice, the five weightings can be compacted into two groups: Exposure, commercial and communication weights, which equalize the different measures of audience, and frequency and synergy weights, which establish the value of reach and repetition.

The greatest benefit of TV reach optimizers has been analytical. They opened up the Nielsen database for the first time and allowed everyone to learn how TV's various pieces work together to build reach. This led agencies to abandon traditional dayparts in optimizing schedules and as a result, TV buying changed. Cable and prime time benefited, daytime lost.

Media-mix optimizers will do the same thing on a larger canvas. They will lead us to combine media in complex ways to reach consumers more cost-effectively. They will help us to think about and use media synergies. They will improve media ROI. All of this will play out in the marketplace.

Media-mix optimizers will show advertisers how and why to move dollars from TV to other media.

Advertisers can't wait.

Erwin Ephron is a partner at Ephron, Papazian & Ephron, a media consultancy. He can be reached at [email protected]

Most Popular
In this article: