Retail store clustering can be a tricky subject to wrap your mind around, but it doesn’t need to be. In the simplest of terms, clustering is just the process of grouping stores by their ability to sell different types of products so you can make decisions about clusters rather than individual stores. Many retailers use sales volume as their primary clustering method, but some industry thought leaders have recently suggested sales volume misses isn’t enough for retail store clustering. At daVinci Retail, we argue sales volume is the right method to cluster by, but you need to take it a step further to get value out of this strategy.
If you’re unfamiliar with clustering by sales volume, the concept is fairly straightforward. Stores are clustered according to their sales figures, usually into four or five different sales volume groups. The powerhouse flagship stores that move the most inventory fall into the “A” cluster, while the sleepy stores that don’t sell as much might be labeled “D” stores. Using this strategy makes sure the stores that sell the most merchandise get the most merchandise.
But there are severe shortcomings to this method. Most importantly, it doesn’t take into account customer preferences. It simply groups stores by their ability to sell without diving any deeper. It assumes an “A” store in Brooklyn is the same as an “A” store in Beverly Hills, or Miami or Vancouver. But the fact is the customers shopping on those “A” stores likely have very different preferences. That’s why sales volume group clustering has come under fire.
See also: Why Your Retail Store Clustering Method Doesn’t Work
We agree that this method of retail store clustering fails to account for customer preference. But we still defend using sales volume as a clustering method. But we recommend clustering by sales volume group at a product class level.
Rather than treating all A stores the same, you instead create clusters that reflect locations’ ability to sell different types of products. That Brooklyn store might be an “A” for outerwear, but a “C” for accessories. Meanwhile, the Beverly Hills store might be a “D” for outerwear, but an “A” for woven dresses. Rather than treating each location as if it’s an “A” for every single product, this method does a much better job of accounting for customer preferences.
|Location||Short sleeve tops||Boots||Bedding||Cookware|
This table shows how stores can be assigned different clusters for each product category
To take this method another step further, we also recommend clustering before you buy assortments, rather than waiting until the allocation process. This gives your buyers the ability to determine much more precise quantities before they commit to buys, and offers allocators a better roadmap for how buys are intended to be distributed.
You can also give your buyers more control over which stores to buy products for by allowing them to tailor buys according to additional dimensions of customer demand. For example, a buyer may choose to exclude a product from certain stores based on factors like country, currency, and climate. Tailoring buys in this way ensures the quantity purchased matches up with the quantity likely to be sold.
Even with this vast improvement over the traditional method of sales volume group clustering, there are still challenges that need to be addressed. The most pressing challenge is that clustering and buying based on product class exponentially increases the number of unique buy plans that must be generated. Tailoring buys by additional layers exacerbates this problem to the point where you could have thousands of unique assortments represented within your buy plan.
The traditional tools used by merchants (i.e. spreadsheets) aren’t capable of managing the quantity of information generated by such a system. To take sales volume clustering to a product class level, assortment planning software is required. Fortunately, in addition to improving buy quantities through better clustering, there are many other benefits to assortment planning software that make the investment worthwhile.
In conclusion, sales volume is more than satisfactory for retail store clustering, but only if you take it beyond a store level. To use sales volume groups correctly, you need to cluster at a product class level and use your clusters to inform the buying process. For maximum success, you should further tailor your buys to ensure you’re feeding the right stores.