Many retailers are depending on disconnected systems for demand forecasting and are missing the big picture when it comes to how fresh impacts center store categories and vice-versa. Joe Skorupa, Editorial Director for RIS News, and Kevin Sterneckert, Chief Marketing Officer for Symphony RetailAI discuss why retailers need a holistic view of consumer demand across all categories.
Interested to learn more?
- Read the full targeted research report from RIS News on AI-Driven merchandise management
- Learn more about how fresh items are impacting the center store in new whitepaper
- Read Podcast Transcript
Hello, I’m Joe Skorupa, editorial director of RIS News. I recently did a study of interest that has been well received by retailers, which we are tracking on our website and that is a report about AI driven merchandising.
Now, Merchants used to succeed by being brilliant guessers, brilliant wagerers. They really had a chance to use their gut-level instincts, certainly playing all of the experience that they’ve had throughout the years into each decision for assortment, for purchasing, for brand image, and all of that.
But now, I can just imagine merchants ready to throw up their hands and just saying, “You know. There are just too many personalized assortments I have to think about here. I have to think about all the customer segments. I have to think about all the stores” and not just stores in general, but individually by planograms. They’re going to throw up their hands because they have to identify meaningful patterns in the shape and the pace of customer demand. They have to throw up their hands because they have to determine price sensitivity for all of these products and all of these shifts, and of course there’s omni channel complexity. What they’re looking for is a solution and AI has that capability to provide it to them.
I think the big change here is, is that in past decades, consumers behaved in ways that allowed for the averages to work, the mean. If I understood the mean, and I understood the trend, I could do a good job in meeting the needs of my customer, because my customers generally behaved in a predictable way. But today, that’s just completely different.
And I was thinking through this around how in the grocery departments are impacting the center store, and I was thinking about how the change and the influence of fresh, and the prominence of fresh for a lot of retailers, they’re using it as a competitive weapon in an eCommerce environment. That prominence of fresh changes the nature of the rest of the store. If customers are buying more fresh, they’re shopping more frequently, their transaction sizes are smaller, but they’re also buying less frozen and packaged, and boxed, and canned goods because they’re buying more fresh.
But if your forecasting engine, and your demand engine doesn’t understand the changes in fresh as it relates to frozen, canned and boxed, then your understanding of demand is incomplete. We created a whole concept around the importance of making sure that your forecasts understand the whole customer that understand the whole store and not just segments, or clusters, or categories, or sub departments of demand. Because the customer isn’t thinking in silos when they shop, they have a mission. They want to satisfy meals, or solutions and they’re thinking holistically and if you’re thinking by category, or if you’re thinking within the sub department, you’re missing a big part of who the customer is and what’s driving behaviors.
You can’t understand demand for forecasting and replenishment if you’re thinking about, “What’s the demand of this item in this category,” and you’re not including, “What’s the behavior of the customer in the fresh categories,” because there is an influence. Most grocers today, their demand engines are not thinking that way and they can’t. It requires this AI and the machine learning in order to do that.
While we’re talking about AI, I do want to point out that most retailers in this study, including grocers, cites reducing out-of-stocks as a major mission, a major goal that they want to achieve. And I have to say that as a research analyst, which is one of the hats I wear, I often think of, whenever I get an answer that shows up number one on every question, there’s something wrong with that answer because it’s too simplistic. I do believe that reducing out-of-stocks is a big problem and solving that problem would result in more sales, would result in better performance in every possible way, but I think it’s how you do it.
Make sure you’re not ignoring the other key factors that are in play here. Now just out-of-stocks, what about overstocks? Inventory is resellers’ biggest asset. It’s actually a monetary figure. If you have overstocks, you’re using your finances and the actual money on hand poorly. Another thing that’s just as important as reducing out-of-stocks, is reducing markdowns and if you reduce markdowns, you improve margins. There are so many other factors at play in here and I just think that if retailers, their first automatic impulse is focus on out-of-stocks, they’re missing out on all the strategies and tactics that lead into it that could have a significant impact on business performance. AI can certainly help on improving initial pricing accuracy and speeding up pricing changes, reducing markdowns to improve inventory and if you do all that, you’re going to reduce out-of-stocks. So I look at all the things that tie into it as opposed to the overall header there.
Yeah, and this is quite interesting. I think what a lot of retailers try do is they’ll say, “Let’s pick a area of problem and let’s go after it and prove out how AI can benefit the business,” so they pick this out-of-stocks. Here’s the reality of it. I was talking with a very large grocery retailer just a couple weeks ago, and they were talking about how they would use AI to apply it to their forecasting. Because of my experience in the industry and where I’ve been involved with forecasting and replenishment, I know that it’s an important aspect to make sure that you manage your perpetual inventory information, and you gain accuracy and compliance with your planograms as well as accuracy and discipline in perpetual inventory.
I suggested and shared with this retailer that the first and best way to gain improvement in out-of-stock is to gain discipline in perpetual inventory and planogram compliance. The retailer replied to me, “Well, you know, we have a lot of retailer leaders and store managers that believe that they don’t have to pay attention to that because their stores where profitable and their sales are growing so it’s optional for them.” I said, “Well, how will AI applied to your demand forecasting improve the situation if the data and the information and the compliance isn’t in place for the AI to leverage?”
I think this is an important aspect when you think about how you apply AI. AI is not the silver bullet that fixes all woes. If you applied AI to a retailer who has poor compliance to the planogram and poor perpetual inventory accuracy, AI is not going to fix your out-of-stock problem, because the root, and the source of information is dirty. It’s not healthy information, so the AI engine will not learn properly, it will not behave properly, it will make proper decisions.
There’s so many places as you pointed out where AI can be applied and machine learning. What’s the right assortment by store? What are the right everyday prices by store? What are the right promotions? What are the right personalized promotions? All of those are very important aspects as well, but thinking that you can apply AI to a problem and not having solved the foundational root causes of that problem to begin with is failure waiting to happen.
To the point I made, and the point we’re talking about here is, would your out-of-stock problem improve if you improved your category management by using artificial intelligence if you improve your personalization by customer, by store, by localized purchasing and inventory, would your out-of-stocks improve? Would your out-of-stocks improve if you used AI to develop exclusive label of private label products in-house? Would all of those things contribute to reducing out-of-stocks?
Absolutely, so focusing your AI initiatives on out-of-stocks without focusing on all of the critical applications that play into it – It’s not a silver bullet that’s going to work, so focusing on those specific areas is probably a better strategy to deploy.
Yeah, it’s going further upstream, right? It’s getting at root causes, so in essence, being out-of-stock is a result of many decisions that occurred upstream. Either it wasn’t the right number of facings, it wasn’t the right forecast, it wasn’t the right arrival time of products in the warehouse, it wasn’t a recognition of competitive activity, of pricing, of competitive promotions, or the response by consumers to your good prices, or promotions. All of the decisions upstream are contributors to the result of being overstock or out-of-stock. If all that your doing is saying, “I want a better forecast so that I can have in-stock conditions,” you’re missing all of the causal and direct impact on why you have an out-of-stock problem to begin with.
I think it’s worthy to pick an area to apply AI. I think that’s worthy. I think the problem is, is that focusing it on the end result versus a foundational, upstream decision, I think that’s the challenge and where we should really be focused on is further upstream to improve the things that contribute to the condition of being out-of-stock or overstocked.
That’s a great idea there. It is possible to conceive of solving the out-of-stock problem. It is possible that you would have all of your shelves stocked, ready for customers to buy, but customers don’t want that product. If you have the wrong product on those shelves, whether they’re fully stocked or not, you’ve got a problem.
That’s exactly right. We could look at a lot of examples of companies who solved the out-of-stock problem by solving the upstream. That’s one of the things that Best Buy worked really hard on is being in-stock. They didn’t just work on their forecast to get better in-stock conditions, they worked through their whole supply chain, and every part of their assortment, the number of facings, their holding power, their holding in the store behind the wall and on the floor. There was a large number of contributors to being in-stock for the customer.
I think a tactic in applying AI is saying, “I want to improve my assortment,” or “I want to improve my understanding of how my suppliers deliver.” If my system has a setting that says it’s a three-day lead time to get merchandise from this vendor, but the vendor always shows up in four days, I should change that, because that affects everything downstream. If I’m constantly running out of the product in the warehouse and I can’t fulfill it to the store, AI isn’t going to fix it because I had a better forecast. AI fixes it because I tuned my true lead times versus my set lead times and I go through and I actually get the lead times right, then the whole system downstream can work properly.
Kevin, it was great talking about artificial intelligence especially as it can help the merchandise management area.
One thing I have to call out in summary is that the AI can identify meaningful patterns in such critical areas as customers demand. Customer demand, demand forecasting is something that is absolutely essential to retailers and it encompasses so many factors that it really takes an AI engine to shift through and make very fast decisions.
Great discussion and appreciate the study that you did and as always Joe, it’s phenominal to have an opportunity to visit with you. And I hope that our listeners have found this valuable and we stand ready to help at Symphony in any way where customers are looking to connect with their customers even better.