As researchers, our job is to provide answers. “This is the packaging treatment that maximizes trial.” “Here is the optimal price.” etc. So, when we can’t identify a definitive answer, it can be frustrating. It goes against our nature!
That is why flavor sorts, or line optimization studies, can be challenging. We want so badly to say “this is the optimal line up.” Black and white, no questions asked.
When you are working from a small set of alternatives, it is more feasible to identify the “perfect set.” However, most companies are working with much larger sets, especially early in the development cycle. A recent example: A CPG client wanted to explore which of 100 potential new flavors should be added to the line. Even if you want to find the best combination of 20 flavors, guess how many possible combinations there are?
Would you believe 536 Quintillion?
Even if you don’t know what “quintillion” is, you know it’s “a lot.” (In case you are curious, it is: 536,000,000,000,000,000,000.) And, when you are working with a set that voluminous, it is not realistic to assume there is one “best” combination… or one hundred best, or even one thousand best.
Many researchers would immediately recommend TURF for flavor sorts or line optimization. TURF is “Total Unduplicated Reach & Frequency.” TURF essentially looks for combinations that maximize the number of people who would be “reached” by at least one of the available flavors. So, for example, if you have Chocolate and Vanilla Ice Cream, adding Chocolate Chip probably does not “reach” people who are not already reached by one of those flavors. But, adding Strawberry may be more effective in reaching people who don’t like Chocolate or Vanilla.
But there are several potential flaws with TURF. One is that, for many brands, “reach” is maximized with just a handful of the top selling varieties – thus, improving reach can be an unrealistic goal. Still, the right research approach can help to sort through it.
Decision Insight has found an alternative approach that can provide substantially more value.
Using a series of simple exercises, we can efficiently evaluate a large quantity of products (we’ve tested over 100 products at a time). The output centers around two key components.
First, we conduct “redundancy” analysis looking for statistical associations across varieties. This analysis identifies and groups varieties that are highly duplicated or, in other words, appeal to the same consumers.
This organizational structure provides a framework and one input into the recommendations: by increasing the number of “groups” represented in a recommended assortment, we are maximizing reach.
Then, within each group, all potential varieties are prioritized based on how they perform on a number of key measures. We typically include exercises that measure output such as breadth of appeal, depth of appeal, craveability, substitutability, etc. We can even take it a step further by using Shapley Value Analysis, to understand how much value each individual flavor adds to the line.
The process provides a clear framework for our clients to give every variety a “green light,” “yellow light” or “red light.” Ultimately, operational factors can be combined with the consumer insights to begin to narrow down the set and provide the intelligence needed to make informed go-forward flavor decisions.