Guy Yehiav is CEO of Profitect, a well-known data-analytics solution provider in the retail loss prevention industry. LPM asked the 25-plus-year retail supply chain industry veteran to explain how advanced analytics is contributing to positive changes in omni-channel retailing.
Data analytics has been one of the major technology applications driving the evolution of retailing. While many of the analytics platforms are focused on in-store operations, the same techniques can help retailers fine-tune their logistics and supply chain operations to maximize merchandise on-shelf availability and support omni-channel retailing.
LPM: In layman’s terms, what is predictive versus prescriptive analytics?
YEHIAV: Predictive and prescriptive analytics differ in their end purposes and user types. Predictive analytics uses many techniques from machine learning, pragmatic artificial intelligence (AI), modeling, and more to process data, find trends within it, and use those trends to make predictions about future business performance and generate reports to be interpreted by data analysts. Prescriptive analytics does the same, except that it uses those trends to tell the right person the right action to take for an optimal outcome. A prescriptive analytics solution might have thousands of users versus just a few with a predictive analytics solution.
LPM: How is analytics used differently in supply chain versus store operations?
YEHIAV: Supply chain is using analytics to maximize on-shelf availability and optimize their processes through just-in-time delivery, full truckloads, on-time shipment complete, forecast accuracy, efficiency at the distribution centers (DCs), and other factors.
So in a nutshell, predictive analytics makes predictions, while prescriptive analytics makes predictions and tells you what to do about them to drive optimal ROI, revenue, and margin improvements.
Store operations is using analytics to maximize customer satisfaction, optimize P&L, optimize labor, minimize shrink, execute promotions, optimize merchandise displays and promotions, and other examples.
As both areas are leveraging a significant amount of data, the common solution has been predictive analytics and/or exception-based reports. However, these methods are not ideal, with the following six common points of failure:
- Did anyone open the email or report?
- Do they understand the reports, fields, and columns?
- Do they see the insights? Do they understand them?
- Do they know what is expected of them in responding to this insight?
- Did they follow through and execute the task(s)?
- What is the value of doing a given task compared to others?
Prescriptive analytics has turned this upside down. It provides those who receive the reports with clear, actionable directives in plain text that anyone can understand and follow through on. By tracking both the opportunity and the response to it, prescriptive analytics provides a level of comprehension and visibility that predictive analytics and exception-based reporting can’t match. Combine that with a user-feedback loop, and you have a systematic process with built-in continuous improvement. The solution cuts through any biases that may arise during the interpretation of the reports.
LPM: What is the “Amazon model,” and how is it different from the normal retail model?
YEHIAV: Amazon’s flexible technology stack allows it to offer consumers a broader product assortment (endless aisles, the long tail of merchandising), greater convenience, highly competitive pricing, and other services like movie streaming and cloud management through Amazon Web Services-all of which make Amazon a formidable competitor for traditional multichannel retailers. But when compared to the normal retail model, retailers have something that Amazon doesn’t-actual stores (although it seems to be on their radar, given recent deals with brick-and-mortar stores like acquiring Whole Foods, partnering with Kohl’s, and leveraging the physical supply chain with companies like Best Buy to sell its own, private-label merchandise like TVs).
Despite Amazon’s convenience, consumers have proven that they still enjoy a fully immersive shopping experience-like at Apple. According to the latest US Census data released in August 2017, nearly 90 percent of all retail purchases in America were made in brick-and-mortar locations during the last quarter. Stores offer immediate gratification-something Amazon has attempted to do with one-hour delivery options for Prime Now combined with free returns.
LPM: Beyond these, how are specific retailers improving e-commerce as well as brick-and-mortar operations with prescriptive analytics?
YEHIAV: When it comes to prescriptive analytics, there are many examples from retail and e-commerce. Here are a few success stories from leveraging a prescriptive analytics solution.
A large retailer with a popular loyalty-card program used prescriptive analytics to uncover an issue they weren’t even aware was a problem. Leveraging prescriptive analytics, they were able to discover a trend that involved shoppers cheating their reward points for free rewards. The retailer’s previous system did not integrate points accumulation until the end of each day, so if a customer brought their receipt to multiple stores in a single day, it was easy to receive many times more reward points than were due, then convert it to real money, just like a free ATM machine. The retailer’s prescriptive analytics solution caught the recurrence and saved millions within the first week of use.
Prescriptive analytics identified and broke up an organized retail crime (ORC) ring that turned out to consist of student employees at the call center of a fashion retailer. The ORC group would legitimately buy a product online and, after receiving it, would call the center to complain they never got the product. Thus, they would receive a second product plus a $20 gift card as a customer-appeasement gesture. Prescriptive analytics stopped the ring, saving the retailer an average of $50,000 a month.
Another fashion retailer used prescriptive analytics to identify a glitch in their transaction process. This glitch purged transaction records from the retailer’s point-of-sale (POS) system after seven minutes of idle time. Cashiers who knew about this glitch found a way to exploit it. When the cashiers loaded a live gift card, they waited the required idle time, generating cash out of thin air. Leveraging prescriptive analytics, the data revealed that the cashiers were loading money onto gift cards but not ringing up the amount. Instead they waited, and seven minutes later, the glitch kicked in, leaving active funds on the gift card but no evidence of a completed transaction. From there the cashiers could return the gift card for cash or spend the funds on merchandise without issue. Thanks to prescriptive analytics flagging this, the company was made aware the glitch even existed, took corrective action on the cashiers, and fixed the issue. The savings and future losses avoided were significant.
