Retailers, no longer limited to selling products from within brick-and-mortar stores, are adding online channels to create endless aisles of products for customers. The endless aisle is desirable because it facilitates sales from anywhere, anytime, but it can be difficult for merchants and customers to manage. To succeed, retailers must know what products to present and to whom, and they must make sure customers can find what they want. In this interview, Jeroen Lievens explains how retailers can bring data science to the art of retailing to enhance the effectiveness of their endless aisle strategies. Jeroen is practice lead for retail at Lynx Analytics.
Q: The endless aisle is an evolving retail concept. How do you explain it? What does it mean to you?
I like to approach the concept from both in-store and online perspectives. It is a strategy for overcoming the trillion dollars in sales lost every year because customers can’t purchase what they want when they want it—whether they’re in a store or somewhere else at that moment. Anything involving e-commerce, including online marketplaces, is a solution to that. And obviously it includes ordering out-of-stock products from staff or even digital kiosks from within a store.
Because the endless aisle expands distribution opportunities and can accommodate a wider range of products, these channels raise new merchandizing challenges that retailers didn’t have 50 years ago. Now, they need to control the “endlessness” of their inventory to make sure it focuses on what is relevant to customers. They can add data science to the traditional art of retailing to ensure relevance.
Q: What should retailers do to determine the most relevant assortment for their customers?
Like anything else in retail, you start with the customer: who is the customer, what do they need, what do they like, and what do they buy? A new trend among brick-and-mortar retailers is to use mobility data—data generated by customers who are accessing smartphone applications from within stores—to understand their customers’ ages, genders, income categories, and what brands they like. By comparing these data along with in-store sales data to local demographics, a retailer can strategically expand its assortment to better serve existing customers and attract new ones.
You do need to be careful and do this in a manageable way, however, so you can meet expected increases in customer demand and track the business impacts. For example, it’s best to expand the assortment for a particular brand incrementally, adding items season by season based on what you learn from each iteration. When measuring the impacts, keep in mind that sales of new products might cannibalize sales of traditional items. Watch for this and consider its implications when optimizing your
Q: How important is pricing in the endless aisle? Should you have the same prices across channels?
This is super important, and you should absolutely strive to have consistent pricing because it ensures clarity for the customer. It matters whether you’re referring to a standard price, pack price or MSRP. There is room for some flexibility, however. For example, customers usually understand that you might have a special promotion in one channel but not another. And, of course, different channels often have different pack sizes.
Nonetheless you can get big synergies from multi-channel learnings, especially around prices. In an online channel, for example, you can easily test different price points or discounts for a specific time period or a certain city. The tests will tell you a lot about your customers, even by segment. You can then apply these learnings to the physical store when setting new in-store prices. Even if your pack sizes aren’t the same, you can still do some of this. Newer pricing models, such as those we use here at Lynx Analytics, can help guide these types of pricing decisions to help improve margins.
Q: Often, the inventory from different channels is managed in different systems. How can you use back-end systems to ensure the consumer can get what they want at a particular time?
First, you need to break the silos: At the very least, the systems should talk to each other and know what products they have in common. For example, if you’re selling a particular food item in a grocery store and via an e-commerce-based home delivery service, both channels should use the same ID code for the same product.
Second, you should organize your logistics with this in mind. This is especially important during the pandemic. For example, you might need to move items from your physical stores to a central warehouse for distribution to e-commerce customers. Alternatively, you might want to allow customers to pick up online orders from their local stores. You might want to ship parcels from local stores to customers in the same city to expedite deliveries. Good, integrated logistics will facilitate these strategies. The logistics will also reduce the frequency of shipping mistakes.
Third, you need to have a good sense of the demand you’ll be getting, whether or not you have separate inventories. Importantly, today’s demand models can consider much more than traditional sales variables. In fact, our models can consider regional COVID-19 conditions and social listening analytics from sites like Twitter and Facebook when projecting demand. By embedding these types of analytics in your demand forecasts, you’ll know better where to place your products.
Q: What advice do you give retailers on how to work with suppliers?
Working with consumer-packaged goods (CPG) companies is evolving along with the industry. In the traditional retail model, a grocery chain with finite physical space could leverage the value of its shelf space when negotiating agreements with CPGs. The retailer could run virtual assortment optimization algorithms to determine the best assortment for its shelves and use the analyses to negotiate placement, the number of cases, discounts, and other terms with suppliers. The retailer might not actually carry out the assortment plan but could use it as a negotiating tool.
In the endless aisle context, physical space is less important and now you need to focus on new parameters, such as page rankings, to attract the most attention to your products. You do have new levers to use to your advantage, though. For example, many e-commerce sites use a page ranking algorithm as an incentive to motivate their suppliers to adhere to their business terms. If a supplier complies with the retailer’s terms to maintain price parity with competitors’ sites, the retailer will rank the supplier higher. This type of approach is mutually beneficial to both sides, but it must be super transparent. For example, if a supplier offers a product for a lower price to a competing platform, its compliance score will drop and it will be informed immediately that its ranking has fallen.
Q: What are some of the new platforms or services Lynx is offering now? Do you have solutions that help retailers during the pandemic?
I would mention two things. One that I discussed before is mobility data analytics that help retailers better understand customers. Retailers are taking advantage of this and adjusting their forecasts.
The second thing we are offering is “one-off” analytics that help retailers with unusual decision-making situations. Let’s say a fashion brand sold only half of its inventory during the pandemic and now it has new inventory coming to market. Suddenly it has an unprecedented problem—a double inventory—and it must go back to the drawing board to make critical merchandising decisions in a short timeframe. This is what I would call a one-off problem. We offer specialized analytics solutions for these types of situations.