Drilling Down And Refining Behavioral Clusters

The product clusters defined in the “RFM grid-based clustering to ensure the growing retail business“ are pre-defined in the Enalito. Any retailer with Enalito can find all of these product clusters ready to give insights into their business. However, Enalito cares much more for a  business than retailers think. Enalito aims at figuring out and resolving every ‘What If’ that comes in a retailer’s mind. So, what if retailers want to dig deeper and find some other product cluster that can provide more clarity in their business insights? Enalito has already resolved this question and serves the desired freedom to drill down and refine behavioral clusters, as perfect as retailers want to get a relieving and helpful answer to their each and every What if.

Let’s understand this by an example:

There’s a retailer who sells fine and popular wines on his online store. He uses Enalito to analyze the product clusters to get detailed insights into his business and sends campaigns of those products that turn customers in often and motivate them to buy more from his store. The retailer figures out that the ‘Remove’ cluster in the RF (Recency-Frequency) Grid needs to be more refined, as it shows the products that aren’t sold recently and aren’t bought often either. As the retailer is a seller of wine (rich and expensive), many of his products aren’t sold much or often but whenever they get sold, they contribute well in monetary terms. So for his business, the products which aren’t bought recently and frequently can’t be fitting to ‘Remove’ if they are providing a good monetary contribution.

In such a case and in any other need of drilling down the pre-defined product clusters, Enalito provides the retailer with the provision of refining clusters on their own. For the above example, the retailer can introduce an additional attribute M (Monetary Value) to the pre-existing RF-based Remove product cluster. By doing so, the retailer can set up a condition to filter out only those products as removable that also haven’t provided any monetary contribution in addition to the already low recency and frequency. The retailer can set up the condition as clustering those customers as removable who have the M score as M<=3 (where M is lesser than or equal to 3). After setting this condition, the retailer will yield a cluster of products that aren’t sold recently and frequently, and also haven’t contributed significant monetary value.

Similarly, the retailer can also figure out products that might have not sold recently or frequently but have contributed good money. To do this, the retailer can add up the monetary (M) attribute to the RF-based cluster and set up the condition of M>=3 (where M is greater than or equal to 3). This way, the retailer can yield a product cluster that doesn’t count as removable when the monetary value is good.

The retailer can even drill the ‘Best’ product cluster to get the products that have the best recency, frequency, and monetary value to find those products that are BEST in terms of monetary contribution (M>=3). Similarly, the retailer can go the other way around to find those products that have the best recency and frequency but not the best monetary value (M<=3).

From the above example, it is clear how to add an additional attribute to a pre-defined cluster and make it more refined. Using Enalito, the retailers get the freedom to add as many attributes they need to make their product clusters as refined as they want them to be.

Likewise, any retailer can drill and refine any product clusters as per their needs and yield the best of the profoundly innovative Enalito.