Assortment rationalization isn’t what it used to be

RSR Research’s Brian Kilcourse discusses why, when it comes to decisions around assortment, the customer dimension can’t be ignored. And AI has the answer.

A recent “Viewpoint” on the Rule of 17 from Symphony RetailAI highlighted two important facts for FMCG retailers to keep in mind when it comes to assortment rationalization:

  • Up to 17% of category items are duplicative, according to Symphony RetailAI’s own research
  • Consumers are 90% less likely to make a purchase decision when the choice of “like” items is quadrupled – a finding reported in a 2006 article from the Harvard Business Review, can this still be true 13 years later?

Catering to “best choice” can waste valuable shelf space

Merchants, intuitively if not in fact, know that these challenges are real. While retailers want to give consumers choices, there can be “too much of a good thing.”

A real-life example of this is the hot pepper sauce selection at my local grocer. There are sixteen linear feet of shelf space devoted to offering a wide assortment of hot pepper sauces, everything from Louisiana favorites to exotic Thai products. Shoppers can be particularly finicky about which is the best choice, and so the grocer keeps all the options on the shelf. But when I apply one of the oldest tricks from a retailer’s playbook – checking the bottletops for dust – I know that my grocer is wasting valuable space by offering too many choices in a way that is actually hurting category sales.

Retailers deal with “option overload” by “redlining” assortments – limiting the number of items in a category based on sell-through. But that may be costing the retailer in sales and loyalty in some surprising ways. There’s a famous example from the days of sales data warehouses that points to the danger. The story follows.

Who removed my cheese??

In the late 1990’s, merchants at large UK supermarket chain decided to redline the cheese selection to the top-100 products. To make that decision, the category managers queried the company’s data warehouse, which organized data along the dimensions of “Product/Location/Time”, i.e. a classic decision support data cube. Based on that analysis, one item that the retailer redlined was a Brie cheese from a small independent French producer. But not long after the category had been rationalized, the retailer began to notice a measurable drop off in total sales for one of its top-producing stores at a high-end location. Stymied, the retailer brought in an expert team of data analysts from one of the big-house consulting firms, and after several weeks the team hit upon the idea of match item sales against the company’s loyalty database. By doing so, the company was able to learn that the redlined French Brie cheese was a particular favorite of customers in the top-two deciles of the loyalty system.

In other words, by dropping a particular product, the retailer had managed to alienate its very best customers.

Factor in the customer dimension to assortment rationalization

The industry and the technologies that support decision making have come a long way from those early days of business analytics. But many retailers continue to make assortment decisions based on product sales without considering the new customer dimension in those decisions. There are several reasons for that; for one thing, those retailers that mastered the mass merchandising model did so by optimizing supply chains by limiting product assortments and buying centrally – and they are loathe to give up the gains that they achieved. Another reason is that rules-based data analysis tools – even if they consider the four dimensions of Customer/Product/Location/Time – aren’t good at finding unanticipated patterns that might be important in the decisioning process.

That’s where AI engines can play a big part. AI is particularly good at finding patterns in data that might be important but aren’t immediately observable “with the naked eye.” And that brings me back to the Symphony RetailAI’s article on the Rule of 17, which states that, “Looking at shopping behaviors across sales, basket and loyalty data, AI learns shopper behaviors, understanding what a customer buys as a complement to another product, or even as a substitute. That data is used to drive assortment decisions, helping retailers to understand the motivations, priorities and needs of its customers.”

Move fast, keep it simple, delight the customer

That’s exactly right. But additionally, finding important relationships between customers’ shopping behaviors and product sales doesn’t have to be a science project. The new technologies can deliver insights that once took a small army of data experts to discover, and it can deliver those insights in hours instead of weeks.

Taken together, the ability to gain important insights into category performance based on the relationship between customer behaviors and product movement, and the ability to perform those analyses very quickly, vastly improves retailers’ ability to respond to changes in consumer demand in a way that will maximize the effectiveness of assortment decisions and delight customers.

Want to learn more about assortment rationalization and effectiveness? Read eBook on the “Rule of 17.”