by John Pemberton February 10, 2011
InsightExpress has developed a proprietary research methodology, IdeaGauge, which is used to narrow down a large list of items (concepts, names, flavors, etc.) in a cost-efficient manner and provide researchers with preference simulations similar to TURF findings. The most common question we’ve received from clients when talking about IdeaGauge is, “What is the exact benefit in using this over TURF?” Never one to shy away from the tough questions, John Pemberton, our VP of Marketing Sciences, has an answer.
What Does TURF Provide?
TURF, the acronym for Total Unduplicated Reach and Frequency, is a type of statistical analysis used for providing estimates of media or market potential and devising optimal communication and placement strategies given limited resources. Specifically, TURF asks respondents to evaluate concepts on a Purchase Intent scale for each concept tested.
During the analysis, responses on this scale are recoded into a dichotomous 1/0 scale, where 1 is equal to being reached (1=reached) and zero is equal to not being reached (0=not reached). Here is where the first tricky part of TURF comes into play. The definition of “reach” can be different from study to study and from analyst to analyst. In some cases reach is defined as being a top box response on the scale, in some cases a top two box response, and in some cases a top three box response and so on.
Once the reach definition has been determined, an exhaustive search algorithm evaluates all possible combinations of concept subsets from the larger complete set. So, using flavors as an example, if there are 12 total flavors, every possible combination of 1 single flavor is evaluated (12 total), every possible pair is evaluated (66 total), every possible trio (220 total) is evaluated and so on. Each combination is evaluated to see if it “reached” each individual respondent. Then the percentage of respondents reached in the entire sample is calculated for each combination.
Combinations of similar size are compared to see which one reached the widest audience. Across the different combination sizes, reach is examined to see where additional concepts added to the line begin to exhibit diminishing incremental returns. From these two analytical steps, insight for the number of unique concepts and then the particular combination of concepts are obtained.
What Does IdeaGauge Provide?
IdeaGauge asks respondents to evaluate potential concept options within a series of choice exercises. From these choice exercises, a stylized discrete choice model is estimated that creates an affinity score (or more formally, a utility score) for each item evaluated for each respondent in the data set.
Using these utilities, simulations are run for the various potential combinations. These simulations, executed at the individual level, are run against the option that a respondent will choose to purchase no concept at all.
Again, using the flavor example, let’s suppose the combination of Flavor A and Flavor B is being evaluated. The procedure estimates three probabilities: (1) the probability that a respondent will choose option A, (2) the probability that a respondent will choose option B and then (3) the probability that the respondent will choose neither. Individual probabilities are then aggregated across the sample to create aggregate likelihoods that each of the three possible outcomes occur.
Combinations are evaluated using a search methodology to identify how many items constitute the ideal set and which specific concepts are in that set. Ultimately, the ideal mix will identify the combination that minimizes the probability that customers will walk away with none of the concepts presented them.
Implications of the Differences Between IdeaGauge and TURF
Both techniques are attempting to answer the same question and create analytic output that is similar in format.
However, InsightExpress typically prefers to utilize IdeaGauge simulations as opposed to TURF evaluations based on the underlying mechanics of the methodologies.
In TURF, whether a respondent is reached or not is a decision that is arbitrarily made by the researcher. Usually, there is no reliable or consistent reason to believe that a person recording a “Top Two” box score is really likely to purchase the flavor if it is presented to him. TURF assumes that there is some purchase threshold that can be defined relative to a ratings scale and then homogenously generalized to the entire sample. Based on how different people use scales differently, a top two box definition of reach may be appropriate for one respondent, but not another.
With IdeaGauge, preferences are being revealed through the presented choice tasks and respondents are also presented the option of selecting none of the presented options. This “none” choice is explicitly modeled in the Stylized Discrete Choice model. It allows for the threshold separating rejection from selection to be endogenously calculated for each individual. There is no broadly wielded assumption made to translate intention to action, upon which analytic decisions are made.
An additional benefit of the IdeaGauge simulations is that the technique provides input not only as to which combinations of concepts will yield the widest appeal, but also which specific concepts in the combination are doing the heavy lifting. If a TURF run identifies flavors A, C and H as the best combination, the researcher has no idea how many of each flavor will be demanded by consumers. In an IdeaGauge simulation, preference percentages allow the researcher to understand which of the three flavors are relatively more likely to move in what relative quantity off the shelf. This allows estimates of initial relative volume to be created.
Another benefit of IdeaGauge relative to TURF is that TURF provides no information on cannibalization. When an additional flavor is added to a combination, TURF calculates the additional consumers reached but says nothing about how the introduction of the flavor into the line will cannibalize share of flavors already in the line. Since IdeaGauge creates selection probabilities, the introduction of a new flavor will affect the probability that existing flavors were selected and hence gives an excellent read on potential cannibalization.
I hope this has given you a good look at the differences between TURF and IdeaGauge. I’d love to hear from you on this topic, should you have any questions.