When I took my first Sales Manager job with Circuit City at a store in Silverdale, Washington, I was made aware of a report available to store managers that I hadn’t previously seen. I learned the true reason that employees were forced to use the side entrance: to avoid skewing the close rate. On either side of the main entrance were two small devices that counted any time the beam was broken. This kept a count of roughly how many people entered the store during business hours, and was compared to sales transactions to calculate a “close rate” and metrics such as average spend across all visitors. We were held accountable to the metrics that used it, regardless of our level of trust in the data.
Counting the number of potential customers walking through the door, and comparing it to transaction data, is an interesting exercise in trying to divine what went through the minds of those who passed through the door but failed to buy anything. In Circuit City’s case, they wanted to understand if stores were capitalizing on the opportunities generated by local marketing efforts to generate store traffic. If a store only “closed” 15% of customers who entered, but their district averaged 19% and the total company averaged 21%, there would be some tough questions.
Retailers still track customer traffic, but the sources of non-purchase signals have continued to grow over time. We now have web traffic, email views, Tweets, Likes, and even Bluetooth and WiFi tracking using in-store beacons. All of these technologies generate data that is a demand signal, but not typically considered part of the demand signal. The reports reviewed daily within retailers are dominated by summarizations of transactions. As mentioned above, some analysis is done on traffic data, but it’s rarely incorporated into the data that drives decisions on a daily basis. In this regard, eCommerce is slightly different. Instead of “close rate”, conversion rate is used. The number of views, add-to-carts, and purchases are typically published with similar importance to sales data. This helps merchants better understand if products are resonating with customers that view them, and if they’re getting the views they need to begin with.
Despite this incorporation of additional elements of the demand signal, something is missing. The typical data requirements for forecasting, demand planning, and optimization systems used by merchants includes transactional data only. This isn’t to say that these additional demand signals are entirely ignored by science-driven tools, it’s just that those tools are mostly targeted specifically toward optimizing specific elements of eCommerce visual merchandising such as page layout, design, images, and direct marketing—elements that drive engagement and improved user experience on the storefront, rather than the larger world of optimization solutions available for price, promotion, assortment, and inventory.
This provides an opportunity to improve merchant and marketer tooling by finding novel uses for the rest of the demand signal. Knowing what a customer viewed on the storefront before they made the purchase is valuable. It can tell you things about affinity and cannibalization that are hard to derive from sales data lone. Understanding the journey a customer took across social media, your storefront, and app that finally led to an in-store purchase can help the merchant better manage multi-channel pricing and promotions. These data can bolster models that sometimes strain to find meaning in sales data, and provide powerful insights that address more than customers and extend to prospects.
What are some examples of these missing demand signals? The most readily available for retailers with an eCommerce storefront is the browse path taken by the customer prior to purchase. A customer may perform a search on Google for a type of product and follow a link to the storefront. From there they may click on a few recommended products, or perhaps use the site’s search functionality. Eventually, hopefully, they add a product to their cart and complete the purchase. Today, that transaction tells us one thing: Product X was purchased for price N. What else could we learn from this?
Why didn’t the customer purchase the product they found through Google search? Why didn’t they buy a product that appeared in the recommendations, and instead had to use the site’s search functionality? Was the product ultimately purchased substitutable or comparable with the one viewed initially? What search term was used on Google to find the first product, and what search term was used on-site to eventually find the purchased product?
Considering data generated through direct email marketing, coupon redemption, social media, third party shopping portals, and much more, the world of underutilized demand signals begins to look like an endless buffet of potential insights and discoveries.