Media Planning 4

Media Planning - Rev. 2002
(Part 4 of 5)

Point # 4: Sampling. "I'm damned sure my opinion is worth more than the average of theirs."

So spoke Nobel laureate Physicist Richard Feynman, when asked about taking part in a survey of the Delphi group. He acknowledged that there were probably people smarter than he was in the group, but was unwilling to compromise his expertise to the "average" of a group of surveyed experts.

There are a host of important advances in strategic planning that are related to how we gather information or sampling. Nielsen and MRI are continually asked to do re-contact studies with their samples, to augment its population in order to answer new questions. We are getting better at integrating data between two samples and have undertaken a serious examination of fusion as a process, under the auspices of the ARF Fusion Lab. They are currently testing ten different factors to fuse against forty different independent tests. They will try to determine if demography alone is enough to use to fuse or other behavioral factors must be employed. Results will become available late fall early winter this year.

We have become much more serious about single source research, witness the work done by IRI and Nielsen in this area that led to recency planning. We are however still far behind in coping with the eccentricities of sampling error and what thy do to the data we use.

These areas are major frontiers for the science of our business over the next five years. As we begin to formulate more sophisticated ways of collecting data, we will be able to make a much better guess at the real relationship between planning media and selling product. We will also be in a better position to develop reliable multi-media optimizers.

Point #5: Modeling coincidences

Modeling not only the media we plan, but also the host of other marketing tools at our disposal, is something we have refined pretty well in the US compared to the rest of the world. We are able to pull apart segments of a media plan to see which parts have been most effective in creating a sale of the advertised product. Was it television or magazines? Was it the 30 or the 15, which was more cost effective? Was it flighting or continuity? What daypart worked best? By observing several years of weekly activity with regard to media placement and product sales, we are now able to draw conclusions on these issues. Basically, we look for coincidences. If sales go up and everything else in our plan remains steady except the amount of GRP weight (which doubled) it seems fair to conclude that the media weight was responsible for the jump in sales. Some practitioners even claim they can isolate the effectiveness of an individual spot, but the research companies that manufacture these models claim that is asking a bit too much of the black box. There is just so much magic in there.

But the modeling is far more sophisticated than just media planning. It looks at promotional activity, public relations activity, pricing, competition, the weather, distribution and a large handful of other relevant factors related to the selling process.

Imagine what would happen if you had all this detailed information available to you and you could graph all of this activity on a weekly basis. You would first look for a blip up or down in sales and a corresponding up or down blip in one factor. All other things being equal, you could reasonably attribute the increase in sales to the isolated fortified marketing blip. Of course, the model is a bit more sophisticated than that and has a bunch of built-in statistical safeguards and tolerances, but that is roughly how it works. The problem is that there are many little blips in sales as well as all the other factors and this complicates interpreting of the data.

Advertisers only?

Modeling has become a tool for Marketing Directors at major companies. One major supplier of this information counts only advertisers among its clients, no agencies. Up to now, modeling has been mostly an advertiser driven tool, with agencies occasionally chipping in and some even trying to get into the fray by developing their own systems. In the long run, if agencies do develop these models, it can help them sell their consolidated mega-agencies as true marketing partners, since most of the mega-agencies have companies that operate in every area of marketing. This would dramatically alter how agencies structure and sell themselves. If modeling is adopted in a serious way by mega-agencies, we will witness a revolution in the entire process of account assignments for all marketing areas.