In today’s economy, retailers are hard-pressed to increase revenues. Among the biggest challenges retailers face is matching supply with demand. When a customer walks into your store, do you have the item they are looking for? Or do they walk out because you are out of the product, you don’t carry it or they couldn’t locate it on the shelf? Stock excess is just as problematic and leads retailers to heavily discounted pricing.
In his book, The New Science of Retailing: How Analytics Are Transforming the Supply Chain and Improving Performance, co-authored with Harvard Business School professor Ananth Raman, Wharton operations and information management professor Marshall Fisher argues that retailers have the data they need to manage supply chains more efficiently and increase sales and profits. Knowledge at Wharton recently spoke with Fisher about what types of data are most important for retailers to collect, how they can use this information to identify home-run products and why the retailing industry might be missing as much as one-third of potential sales.
The following is an edited version of the conversation.
Knowledge at Wharton: In your book, you say retailers often fail to match supply and demand. What are the consequences of that failure?
Fisher: You can see one of the consequences when you walk into many stores: huge piles of merchandise on sale at deep discounts, as much as 80%. There is a department store I visit from time to time, and its aisles are sorted into three sections: 50% off, 70% off and 80% off. That is one consequence.
Another consequence for retailers is obviously lower profit, which can come in one of two ways. First, there is the excess we just talked about, which means you frequently sell things below what it cost you to buy them and get them into the store, so you lose money. On the shortage side, you don’t have enough of something and a customer can’t find what they are looking for. That is a lost sale.
Department stores used to collect and report the markdown percentage, until 1995, when it became embarrassing. If I remember correctly, in the mid-1970s, the number was about 6%. By 1995, it had grown to something like 33%. It is not publicly reported anymore. But I heard a statistic that, for one leading department store, that number would now be 40%. This means that the average item sells for 40% off its full price. That is on the excess side.
On the shortage side, there is a consulting firm that does an annual survey of consumers. This is in apparel, where it is maybe hardest to match supply with demand. But they routinely find that people who respond to the survey will say that, about a third of the time, they walk into a store with a clear idea of what they want to buy and walk out empty-handed because they couldn’t find it. This is truly amazing if you think about it — that a $2 billion retailer is really a $3 billion retailer but they are not getting the third $1 billion in revenues because one third of the people walk out empty-handed.
Knowledge at Wharton: Does this imply that if retailers are able to make even small improvements in matching supply with demand, that this would have a fairly big impact on profits?
Fisher: Yes, absolutely. For a simple reason: Retailing is a high fixed-cost business. It costs a lot of money to maintain a store base and pay the associates who work in those stores. You incur those costs whether you sell $1 of merchandise in the store or $10 million, so small increases in revenue have a big impact on profit. Typical numbers for gross margin would be somewhere between 30% and 50%. Take a retailer whose gross margin is 50% — a 5% increase in sales with a 50% margin is 2.5% of revenue increase in profit, which, for a lot of retailers, would double their profits. When you consider the one-third who walk out empty-handed because of stock-outs or because they can’t find the product in the store, just correcting a little bit of that could double the retailer’s profit.
Knowledge at Wharton: What implications does the relationship of the stock market valuation of a retailer have to its ability to manage its inventory?
Fisher: That is an interesting question because most analysts don’t pay enough attention to inventory or other operating variables of a retailer. You can think of a number of them. But if we take inventory as an example, a retailer can enhance its revenue line with more inventory. To come back to the example I described, you would like to have better in-stocks so that people find what they came for when they walk in the store. The right way to accomplish that is more effective management of your inventory, better forecasting and better analysis of the margin of error around the forecast, so you can risk-adjust.
The heavy-handed way to improve in-stocks is to just jack up your inventory levels. A retailer who does that will see an increase in revenue. It is not a real improvement in the effectiveness of their performance, but it can look that way to the stock market. If the market doesn’t take into account the relationship between revenue and inventory, people can be fooled that a retailer is having a good year when all they are doing is forcing sales by pumping excess inventory into an inefficient system.
Knowledge at Wharton: One of the really interesting points in your book is that many retailers are drowning in numbers but lacking in insight. How can they correct this problem?
Fisher: Well, that was actually a poster on the door of a woman who we worked with whose title was vice president for merchandising technology, I believe. The quote was, “We’re awash in data and starved for information.” The short answer to your question is they should buy this book, read it and implement the results.
That is essentially what we tried to do in this book. We reported more than ten years of experience we have had working with retailers who were awash in data but starved for information and helping them to interpret all kinds of data, starting with point-of-sale data, customer satisfaction surveys and demographic data about their stores. If you think about what would cause you to be awash in data but starved for information, there are many things. I’ll mention a couple.
One is that a lot of retailers, until recently, have not had sufficiently granular data. This may surprise you. Take this shirt I’m wearing as an example. They would see sales of this shirt for the entire chain, across all sizes. They would not see the sales of this shirt in a medium in XYZ store last week. So they are left to guess the level of in-stocks across the stores. They are left to guess at the right size mix. Point one is, until recently, the data was too aggregated to be useful. You need it at the microscopic level. How is this shirt selling in the Iowa store in a medium size today?
The second point is that it is not enough to know an item’s sales level. You have to know its sales level in other conditions that might affect the sales of that product: competitive activity, weather, how it is priced, how it is presented in the store. Is it well presented, or is it somehow hidden in the back room? Most retailers don’t collect that data, and they don’t interpret it. You need to look at sales and a handful of factors that are sales drivers in order to accurately predict how an item will sell in the future.
I could go on and on and on. It is doable, but non-trivial, to look at the huge amount of data retailers have available to them now and make sense out of it. But you can, and then it is very powerful.
Knowledge at Wharton: Your book talks about how retailers can crunch their sales numbers to identify home-run products that they may be missing. How can they do that?
Fisher: The retail assortment is the set of products that is in the store at any point in time when you walk in. Retailers will periodically update their assortment. They will get rid of products that are not selling so well and add new products that they think will sell better.
When you think about those two decisions — get rid of the worst sellers and add some new products — which is easier? The decision about which products to get rid of is easier because you have sales data. But you have to be a little careful because that product that is not selling very well may well be the favorite product of some of your best customers. If you get rid of it, you not only lose the sales on that product, but you lose the customer and everything else they are buying. It is a serious problem in the grocery business. But other than that kind of subtlety, it is easy to figure out what to get rid of.
What to add is harder. A lot of retailers do something kind of like gin rummy. In that card game, you discard your worst card, and you draw randomly another card from the deck. So, they will add another product to the assortment. My colleagues and I asked ourselves, “How could we give better guidance on that?” The idea we deployed is to think of a product, a stock-keeping unit, a SKU as we call it, as defined by its attributes. So this shirt I’m wearing would be a knit, it is short sleeved, it is a certain color, it is a certain size. Then you use the sales of your current products, in each and every store, to estimate the demand that customers have for those attributes.
Then you can do something very powerful. If I know the sales of this color and of short-sleeved shirts and of mediums, I can figure out sales of other products that are different combinations of attributes that I am not currently selling. We found, amazingly, that there will be products that the retailer thought there was no demand for, so they didn’t have very much of it. But if you analyze the data correctly, you see that there is a huge demand.
Knowledge at Wharton: Can you give an example?
Fisher: We worked with a tire retailer. One of the attributes was the price and quality level. They had six different price quality levels from cheapest to highest price. They didn’t think it appropriate to carry very much of the cheapest tire. Across approximately 100 different sizes, they carried the lowest price, lowest-quality tire in only nine of these sizes. They didn’t sell much of it: It was only 5% of their sales, confirming their belief that customers really didn’t want to buy this cheap tire.
If you looked at the sizes, the nine sizes that they carried this lowest price point in, across the six, it represented 60% of revenue. It outsold the next bestselling tire by ten to one. The customers, if you look carefully at the data, were screaming, “Hey! Price matters to us.” The neighborhoods that these stores were in were less-wealthy neighborhoods, so it was not surprising. There was a whole set of home-run products that they were not aware of because of this self-fulfilling prophecy that, “We don’t think our customers will want to buy low-priced, cheap tires, so we are not going to carry them. We don’t sell a lot of them, and we were right.” In fact, they were dead wrong. There was a huge opportunity there.
Knowledge at Wharton: Another issue you bring up in your book is that inflexible supply chains are the bane of all retailers. What are some ways in which a supply chain can become more agile?
Fisher: Well, I would start answering your question by referring back to the beginning of this interview, where we talked about the average markdown in a department store. Now it is 40%, signaling a huge amount of excess supply of some products, yet customers who fill out a survey say one-third of the time, “We can’t find what we came for.” So there is also shortage. How do you correct that? How do you better match supply with demand, have less of the wrong products and more of the right ones? There are really three things you need to do — and all three must be done together.
You need more accurate forecasts, better choice of the right inventory levels and a flexible supply chain, which means short lead time, the ability to supply the quantity the market needs efficiently, which might be a lot, or it might be a little. A number of factors have caused supply chains to become inflexible over time. But probably the biggest one is the pursuit of low cost by sourcing from Asian countries, China being at the head of the class. I first visited China while teaching in a Wharton program for six weeks. It was in 1982, and I stayed in Shanghai with my family. Pretty much the only other Westerners we saw were buyers from apparel companies. They were the first wave. That was really the beginning. China normalized relations with the United States in 1979.
That is when apparel companies realized that wage rates in China were 3% of what they were in the United States. That caused a flood of product sourcing from China and other Asian countries, including apparel, toys, consumer electronics and many other things. That lengthened the supply chain from miles to halfway around the world. That is probably the biggest factor that has caused supply chains to be inflexible. There are lots of other choices you make in a supply chain, between slow and cheap versus fast and expensive. For example, you ship by boat or by air.
You can measure cost. And companies are very conscious of cost. The value of speed is harder to measure. That value is better in-stocks, better revenue. But you don’t have a direct measurement. It costs you an extra $3 to ship something by air. The gain you get from that is unknown. If you pay the $3 today, the gain is in the future. There is a tendency for what’s immediate and measurable cost to be over-weighted relative to what’s harder to measure, though it is valuable: speed and flexibility of your supply chain. Every time there is a fork in the road for companies — “Do I pick slow and cheap or fast and expensive?” — the bias is to take the slow-and-cheap route.
Knowledge at Wharton: You cite examples of companies like World, a Japanese clothing company, and Spanish clothing and accessories retailer Zara that have done innovative things with their supply chains. Are there lessons that other companies could learn from their experience?
Fisher: Of course. I think the first lesson is to know — in this choice between slow and cheap and fast and expensive — when is it appropriate to choose fast and expensive? It is not appropriate for stable products with very predictable demand. Take the example of Campbell’s Chicken Noodle Soup. It has been around for more than 100 years and has a long shelf life. Access to inventory is not going to go obsolete. You will sell it eventually. You may have to hold it a little longer. There is very predictable demand. You don’t need speed and flexibility in the supply chain for that.
You can contrast that with fashion apparel, toys and many consumer electronics products. There is a huge spike of demand at Christmas. These are products that have a gross margin of 50% or more, so missing a sale is extremely expensive. For those products, you need flexibility. Zara is pretty well known. World is a Japanese retailer, less well known in the West because they sell only in Japan, but it has similar practices to Zara. Just a couple miles from where we are sitting in Philadelphia is Destination Maternity, a company started by Rebecca Matthias, a graduate of the University of Pennsylvania, with a crushingly dominant 50% share of the women’s maternity market.
All three apparel firms have adopted a DNA of speed and flexibility. They can produce more and get back in stock on a hot selling item in two weeks, whereas that number for other retailers would be measured in months, which is game-over for a seasonal product. If something is selling well, too bad, you can’t get back in stock. These firms do a whole set of things. Some of them are sort of obvious and mechanistic. They will buy un-dyed fabric and keep it in inventory so they can quickly cut it and dye it. They use a laser for cutting, so they can cut a single layer of fabric.
The other thing they do, which I think many retailers miss, is they empower product teams in the trenches, the lowest rung of the organizational ladder, to make decisions. You want to make more of a product, and you can’t get the same button you originally used. But you can get one like it. Is that okay? They can decide yes or no. The team will have a designer on it, and they will say, “Yes, that will work.” So they will be somewhat scrappy and street smart in matching the materials they can get with what the market is telling them is needed
Knowledge at Wharton: Any thoughts on how retailers can improve their store-level execution?
Fisher: First, we all shop, and we all experience store execution. When we go to a store, can we find a parking spot? Is it easy to find the entrance? When we get in, is the store easy to navigate? Does it look neat and tidy or messy? If we need help, can we find somebody? There is a whole set of components of our shopping experience that affect how we feel. The next time we get to the end of our driveway, are we going to turn left and go back to that retailer, or turn right and go to the competitor?
When I talk about store execution and I ask my students, “Who’s had a bad experience shopping?” Every hand goes up. “Who’s had a bad experience that made you really angry?” Every hand goes up. I could spend an 80-minute class just listening to viscerally angry stories about bad experiences they have had shopping. I would just suspect you have had that. I know I have. Everybody has. And it is a huge, huge problem for retailers.
There are a couple of ideas described in the book that I think are important. One is getting staffing levels right in the store, or the quantity and experience of the store associates. We worked with a number of retailers who collected customer satisfaction data, surveys from their customers, on shopping experience. And you could correlate how satisfied was the customer in their overall number with factors that might drive that satisfaction. And almost always, across a variety of retailers, it was three things. “If I need help, can I find it? Is the person knowledgeable and helpful to me, the store associate? And can I find the product I came to buy?” So at the end of the day, people who shop have a mission. And if they can accomplish their mission, they are happy.
We’ve found that stores, on average, are understaffed. Again, it comes back to the measurable being over-weighted relative to the un-measurable. You write a check at the end of the month for salary for your store associates. That is measurable, known and immediate. The value of having 10% more payroll budget, and therefore, 10% more people in the stores, is harder to measure. So retailers tend to under-staff.
Based, in part, on these customer satisfaction surveys, we developed a statistical technique for correlating revenue with level of staffing in a retailer. We found, in one example, that every dollar more you spent on payroll added $10 in revenue, on average. It varied by store and by month, so you had to apply it in a granular way. Because some stores were understaffed, you wanted to cut them. But this technique lets you measure the revenue impact of payroll. Now you think about the economics: to spend a dollar this month to get $10 in revenue this month, with a 50% margin, you’re getting $5 in profit for a dollar in expense. That is a great deal.
And it is not an investment because it happens within the current month. So that was what we saw as an execution void: number one, an overweighting of the payroll expense, viewing labor as sort of an expense rather than an investment, and number two, not being able to staff in a way that takes into account the impact of staffing on revenue.
Knowledge at Wharton: One final question, Marshall. What advice would you give managers of retail companies to help them succeed in the new normal global economy?
Fisher: This new normal is hard to define. That phrase is often used to describe the worst economy that you and I have experienced. Who knows how this is all going to play out? There is this sense that the economy will recover, but that it will be somehow fundamentally changed in ways that are hard to discern.
I’ll just mention one possibility, which many retailers believe. During a recession, people become more price-conscious. If you look at the impact of the recession on retailers, for the vast majority, maybe 90%, it was bad news. Revenue went down. Surprisingly, some retailers are counter-cyclic and do as well or better in a downturn. So who would you guess would be a counter-cyclic retailer that does better in tough times?
Knowledge at Wharton: Probably the discount retailers?
Fisher: Yes. So who would be at the head of the class?
Knowledge at Wharton: Walmart.
Fisher: Walmart, yes. Walmart, auto parts retailers, any do-it-yourself retailer. They did as well or better during the recession. Will that persist is a very interesting question. And of course, the Walmarts of the world would say, “Of course it will.” Fundamentally, people have changed their buying behavior, and price matters more now. There is a whole generation that went through this lousy economy, and they are going to think about life differently. They are going to be more price-conscious.
The other thing that is going on now, and I think will persist for a long time, is de-leveraging. People are paying off their credit card bills, which means they are spending less. There is a great reduction in consumer-spending levels. That is going to have a big, big impact on retailing.