Q&A—The growth of prescriptive analytics

data infographic
Profitect is helping retailers to better analyze and put into action their collected data. (Pixabay)

Retailers have been diligently capturing data, but many still struggle with harnessing that information to take needed action. In the past, most retailers turned to business intelligence and reports to help take action. But Guy Yehiav, CEO of prescriptive analytics provider Profitect, says that reports are continuously delivered and employees are spending countless hours trying to interpret them. 

"The volume and time required leaves people feeling like they’re drowning in reports," he said. Profitect recognized this issue and created prescriptive analytics technology leverage pattern detection and machine learning. This solution takes a retail or CPG company’s data and identifies areas for improvement such as inventory accuracy, out-of-stocks, pricing accuracy, unsellable merchandise and assortment discrepancies. As a result, identified improvements are delivered in plain language and integrated workflow ensures recommended actions are taken.

FierceRetail sat down with Yehiav to learn more about how Profitect is changing the ways in which retailers look at data. 

FierceRetail (FR): What retailers do you think are in need of your technology in 2018?

Guy Yehiav (GY): This question depends on two factors: size and analytical capability or structure. In terms of size, prescriptive analytics tends to benefit retailers making more than $100 million in annual revenue. At this point, retailers typically have multiple stores, employees, SKU’s, distribution centers (DCs) and are dealing with complex data issues. With all this available data, prescriptive analytics provides a way for them to streamline operations and helps enable future growth.

From an analytical capability, prescriptive analytics helps them be more efficient by eliminating the confusion around traditional analytics reports. Prescriptive analytics helps retailers identify and resolve unknown/unknowns, known/unknowns and known/knowns. Unknown/unknowns are issues that retailers don’t realize they have and, therefore, don’t know how to fix. Known/unknowns represent an issue that the retailer has identified, but is not sure how or why it is occurring. Lastly, there are known/knowns. These known/knowns are issues that have been identified, and their root cause is known. The key to prescriptive analytics is not just finding the area to improve or duplicate a success, but it is as important to find what is being done about it and guide/monitor all the way to execution.

FR: Can you explain how Profitect works to improve the data analytics process?

GY: Prescriptive analytics leverages machine learning and pattern detection to analyze a retailer’s data and sort what is and isn’t working well. It helps duplicate what is working well and/or eliminate what is not working. Prescriptive analytics describes in plain language what aspects of the business need to be fixed/duplicated and then prescribes the steps needed to increase sales and margin. This organized, synchronized system allows people to understand data and business needs the same way across the retail organization.

FR: When it comes to analyzing data, where do you think most retailers are "missing the boat?" 

GY: Some retailers only have partial analytical capabilities, and unfortunately, this means they still have siloed data. In today’s world of data processing, it is no longer enough to find the anomaly throughout siloed data. It’s essential to have full task management workflow capabilities that take you through the execution step-by-step and in a way that ensures the correct actions are being taken. It can also help facilitate operational efficiencies such as job training. Because of attrition or maybe part-time labor, retailers want to ensure the business-process execution is flawless every time, no matter who completes a task.

FR: How quickly can your technology be integrated with what retailers currently have?

GY: Profitect can easily be integrated into any existing retail technology, with little IT involvement. Once the retailer provides the data, Profitect’s prescriptive analytics solution is immediately available for displaying insights and analyzing information. The implementation process can be completed in days, and our software-as-a-service model means you can access the solution from any internet-connected device. The engagement is at no risk to the retailer or CPG company.

FR: Where do you think analytics are headed in the near future, and what should retailers do to prepare?

GY: Analytics are going to continue to evolve, but based on what we’ve seen, the answer for most of the industry continues to move towards creating the best-looking report out there. The goal of retailers should be viewing analytics as very smart tools that can be democratized and used by everyone. We also believe using sophisticated technology like machine learning can help identify issues that were previously seen as hard to predict and/or measure like hidden demand and shrink forecasting.

Hidden demand (lost sales due to insufficient inventory) isn’t easy to quantify and therefore it’s priority, but Profitect can help. Our machine learning capabilities can predict the hidden demand for a particular item at a given store (or online) on a daily/instance level. Our proprietary model uses historical sales data during time periods where the item is consistently in stock as well as using other “like” stores from the same cluster behavior sales patterns, to learn the behavioral patterns underlying sales trends. Once taught, this trained model is then used to forecast the demand for time periods (even within the day) where the item is not sufficiently stocked. Profitect’s machine learning capabilities also help teams drive sales by improving shrink and inventory accuracy.

Like predicting hidden demand, machine learning can also be used to look at other areas of the business, like shrink and operational compliance. In order to predict shrink, Profitect analyzes how shrink data behaved at the time of the last cycle count, combined with other data type behaviors of activities over the time frame. Our machine learning identifies smart correlations between data streams and additional shrink indicators, such as theft or operational loss, and generates a prediction accordingly. With Profitect, retailers can report predicted shrink and do more with less in terms of guiding teams because they will be focused on the right issues at the right stores and within the right categories.

Retailers have data all around them. They should look to automated analytics to solve their problems, even ones they may not realize they have, in an easy way.

FR: What else can you tell us about improving the analytics process?

GY: Two myths have appeared since the advent of prescriptive analytics. First, many retailers think the promise of prescriptive analytics is too good to be true. However, Profitect has many customer examples that can backup the benefits with testimonials speaking to results seen and ROI provided.

The second misconception comes from retailers believing that simplifying and condensing the massive amounts of data on the backside of operations leads to long implementation times, as they are often used to working with technology vendors with six-, 9-, or 12-month or more deployments. This is not the case with Profitect, as we’ve deployed tier-one retailers as quickly as two weeks.