Straight Averages Vs. Weighted Averages
A philosophic side debate
During the course of the Turner RMT NCS study, there was a debate between researchers, as documented in the ARF paper “Quantifying the ROI Impact of DriverTagsTM Context Resonance” (at III.9, IV.3, and V.1), authored by Bill Harvey and Howard Shimmel. The debate was over how best to interpret and present the findings.
The objective of the study was clearly stated before the study was commissioned: looking at historical ad campaigns for major advertised brands predict how much of an ad-produced sales increase should be expected with practical implementation of RMT DriverTagsTM in media planning and buying.
As RMT and Turner requested, for each of two selected brands, Nielsen Catalina reported sales lift indices by impressions grouped into ten groupings of DriverTag™ Ad Resonance. RMT inspected the data and consistent with the forecasting objectives of the study, estimated that an average +36% further increase in sales lift would be pragmatically attainable.
The RMT estimate was computed using straight averages of the NCS sales increase indices across the High Context Resonance grouping. The estimation method is consistent with the concept of representing a hypothetical future buy optimized with the use of RMT DriverTagsTM. The future buy assumes that the central tendencies within each NCS-measured group e.g. “90%-100%”, “80%-90%”, etc. was representative of each group and that by buying purposely to obtain the highest possible Ad Resonance, GRPs would, if anything, be skewed upwards.
NCS correctly pointed out that the two actual brand ad campaigns NCS evaluated, bought without any consideration of DriverTagsTM Ad Resonance, there is a downward-skewing of GRPs; as such, NCS argued that the RMT quoted improvements should be +23% versus +36%. RMT argued that in a future schedule, the GRPs would more likely be skewed upward rather than downward, possibly making the +36% number understated. RMT believes that the NCS argument should be disclosed wherever the RMT +36% estimate is quoted.
For clarity, the +36% estimate assumes that in a buy made based on integrating Resonance considerations in media selection, the GRPs would be neither skewed upward nor downward in Resonance. The +23% Nielsen Catalina measurement reflects an actual buy made without considering Resonance, and the GRPs are skewed downward to lower Resonance placements. If one assumed that in a Resonance-reflective buy the GRPs would be skewed upward to higher Resonance placements, the sales lift would be greater than +36%, probably in the vicinity of +49% (13 points higher than +36%, because the +23% scenario is 13 points lower than +36%).
Whichever way one looks at it, this is compelling proof that the DriverTagTM Resonance metric is a powerful tool for lifting sales effectiveness of advertising, which can be integrated with purchaser targeting, reach, cost efficiency, and all other factors in making better media buys.