Category Management Plan Example: Category Definition and Segmentation
Attribute-Driven Category Definition
After internal alignment has been reached (step 1 of the CatMan 2.0 process), covered in detail in this post here, it is time to define the category. This step was previously step one in the CatMan 1.0 process and can be considered a critical part of the process. It is in this stage where the category management team will take a look at organizing and defining the types of products that are to be considered in the category, and what sub categories of products it will include.
Initially this sounds like a fairly standard step, but in fact, there are critical tools such as Markov Chain Analysis, Clustering and the Category Decision Tree (CDT), that are utilized that will have fundamental implications on the entire category management plan and in particular the category definition. It is therefore very important to evaluate these tools to understand how they may benefit from high-order attribute data and an attribute-driven approach.
Types of Segmentation
Traditionally, categories were segmented in two or more ways by attributes such as size, packaging, form, user, or in some advanced cases, by ingredient or formulation. This type of segmentation is shown below on the left. In most cases, if product data beyond typical attributes was leveraged, it was limited to very broad ingredient data, such as ice cream, ice milk, vs. yogurt.
With the availability of high-order attribute data as supported by the Label Insight retailer solution, category managers are now able to to dive much deeper into the types of segmentation than previously. In an attribute-driven market, segmentation can now include such attributes as detailed ingredient properties, "free from" claims, allergen call-outs, certificates, or supply chain claims, just to mention a few. This new approach to segmentation has the potential to fundamentally change the way the category gets defined, as understanding the consumer's decision hierarchy by segment is crucial.
Standard Ice Cream Category Decision Tree (CDT)
In a traditional CatMan process, as has been previously executed by Jones Grocery to manage the category, the below CDT has been sufficient. The category is defined first by product form such as Ice Cream vs. Ice Milk vs. Yogurt, then defined by price, and lastly by flavor. This has a lot of value when one considers a traditional decision-making process which was probably defined by limited choice. Therefore, it would make sense that people were making decisions based on the product types they wanted according to price, then flavor.
However, in an attribute-driven market where choice is infinitely expanding, the way shoppers make decisions is also fundamentally changing, so it becomes important to leverage the tools at hand and combine them with an attribute-driven approach to see if we can evolve the CDT to better reflect the modern decision-making process.
Markov Chain Analysis
One of the tools that can be used to better understand how people are making decisions about products and what attributes may or may not be driving those decisions is Markov Chain Analysis. Although this may sound complicated, the concept behind the mathematics is fairly simple. A Markov Chain Analysis is used to determine the probability of a new state existing given the pattern of previous states.
In the category management process, a list of historical purchases of a type of product is analyzed for a person or family, then used to predict what product they might purchase next. The probability of the different potential choices are calculated to identify the likelihood of one choice over another, as is described in the diagram below.
Scaling and ClusteringTaking this data and making it actionable is critical. as "insight without action is worthless". The next step in the process would be to leverage multi-dimensional scaling and cluster analysis techniques based on attributes to better understand interactions between brands and to group similarly interacting items into clusters. Once these clusters are formed, the category hierarchical structure can be defined in terms of common attributes of brands within clusters. It is from this type of analysis that the hierarchical tree diagram can be created.
As with the Markov Chain Analysis, the multi-dimensional scaling and clustering takes on a new dimension (pun intended) with the addition of high-order attribute data. So for instance, the Markov Chain Analysis may demonstrate that many people are demonstrating that "lactose free" is a key decision influencing attribute, which, when added to the cluster analysis results in an entirely different Category Decision Tree.
Attribute-Driven CDTThis is exactly what Jones Grocery found. Upon completing the attribute-driven Category Definition process, the Jones Grocery team created the following CDT. They found that there were three key ways people were making decisions about ice cream in their stores: they were either looking for natural ice creams with nothing artificial; they were looking for ice creams with added "good ingredients"; or they were looking for ice creams that didn't contain certain things - "things to avoid". Once this critical decision-making criteria had been met, shoppers would then choose the flavor, safe in the knowledge that their core decision-influencing criteria had been met. The resulting CDT, as shown below, is considerably different from the traditional ice cream CDT that Jones Grocery had previously operated from. This fundamentally changes the way that the Jones Grocery team looks at the ice cream category and, as we will see as this series progresses, the implications are significant.
Category definition in an attribute-driven market
In an attribute-driven CatMan 2.0 process, category definition can lead to a fundamentally different category decision tree. Put another way, being attribute-driven can lead to a more sophisticated understanding of a category and how consumers shop that category - one of the fundamentals of category management. In this recent post, we explored the idea that high-order attributes may even go so far as to invert the category decision tree, making the way we have traditionally thought of categories as redundant. But as you'll see in the post, this was neither a new idea, nor a valid one.
Irrespective of the disruptive nature to the profession, there is no doubt that changing the CDT in this way will have significant implications as we progress through the Jones Grocery ice cream process.
What's next? Category Role
In the next post in this series, we will look at the the Category Role step in the process. This stage seeks to set the context for the category in the wider picture of a total store experience. We will aim to understand how important the ice cream category is to the Jones Grocery team, and how important is it to their shoppers.