A lot of retailers and manufacturers are turning to data scientists in the hope that they can make heads or tails of data. That’s a great thing- but a carefully crafted strategy and goal on exactly how data can be used to improve business is a necessity. Furthermore, data scientists may have great technical skills but lack the necessary retail knowledge and experience to make use of those skills. It’s not about coding or using one of the many visualization tools on the market that matters. It’s the domain expertise that makes the difference. Not knowing what to look for is putting a lot of dollars to waste. Luckily, learning the fundamentals of retail is not hard. I offer the following five recommendations when putting your data scientist and strategy to work.
- Know what questions need to be asked. Learn the fundamentals of retail. Determine benchmarks for performance- what is good, bad, relative to the business you are studying. When I introduce our interns to retail analysis for the first time I focus on two simple statistics. The number of stores that have inventory and no sales, and the number of stores with sales and no inventory. One implies there is an execution (presentation) problem and the other indicates a stock out. Fixing both have an immediate positive impact on business. There are many scenarios like this for both brick & mortar and e-commerce POS (point of sale) data. Our retail primer can help.
- Don’t lose sight of the end consumer. The POS data we analyze exists because somebody bought something, somewhere, for some reason. The more we can explain each of those, the more we can improve our assortment, allocation, and timing. Studying POS by itself is introspective. It does help improve basic decision making but we have long integrated 3rd party data with POS to reveal more game-changing insights. Understanding where to place tests, target products to the right demographics, improve inventory flow of seasonal products are just a few examples. What else affects shopping habits? Weather? Demographics? The price of a gallon of gas? This is essentially applying “Big Data” principles on a smaller, yet more usable level. A lot of 3rd party data is free or cheap if you look hard enough.
- Focus on inventory productivity. On both the wholesale and retail side of the business, the largest capital investment is inventory. The delicate balance of not having too much or too little is what every company strives for. Mistakes can be extremely costly, so improving inventory productivity by even a small percentage could mean millions of dollars added to the bottom line. Having the ability to analyze every SKU-store combination can bring to light many opportunities and liabilities that would otherwise go unnoticed.
- Provide the required data to take action. Many analysts get caught up in the trap of thinking their role is to just uncover information. But what yields the most value is providing the supplementary data needed to take action. For example, if a particular SKU-store is determined to be under-inventoried, the action would be calculating a replenishment order. Then communicating what action is required with simple instructions and in a format that is easily implemented provides the greatest chance for that action to be taken. Creating a process for that should be part of the data strategy.
- Evaluate the role of Artificial Intelligence. It comes in many forms and flavors from the use of simple algorithms to deep machine learning. We are just now reaching the point where AI can provide valuable insights depending on how large your business is and the tools you choose to use. The next generation of data tools will utilize it (like our RetailNarrative) so you need to understand what it can do for you. There are some encouraging developments in AI that I think will speed up our ability to catch and predict trends more accurately. Like humans, it must develop the knowledge and experience to know what to look for (We call it “Augmented Intelligence”).
The abundance and availability of data doesn’t necessarily relate to better decision making. You must know how to use the data. I have focused primarily on POS data here, but there are many other facets and data points at retail where similar data strategies can be used to yield positive results.