Friday, August 31, 2012

Walmart’s Selection and Long Tails

Walmart has pressed hard on the selection growth lever, packing upwards of 150,000 individual items in each of its supercenters. Its ideal is getting you in the door for that one thing you want, and loading you up with a shopping cart full of stuff you didn’t know you needed…until you came through the doors. 

And that’s priority number one for Walmart: get shoppers through those doors. 

Castro-Wright’s Folly or Walmart Hits the Limits of the Pareto Principle 

Eduardo Castro-Wright, former CEO of U.S. Stores, perpetrated a great folly by reducing Walmart’s selection in an attempt to improve the aesthetic experience for the shopper and thereby compete with Target for upscale discount customers. 

His strategy made sense in theory. I’m sure his analysts poured through the data and saw clear patterns demonstrating some derivative of the Pareto Principle, something that pointed to 20 percent of the inventory generating 80 percent of the sales. You can fiddle with the ratios however you please – perhaps it was 40 percent of the inventory driving 90 percent of sales or 60 percent of the items responsible for 75 percent – the point is that the data highlighted a glut of slow-moving items. If you could just reduce the low-demand inventory you could simultaneously make the stores more spacious and stock shelves with items that sell quickly. 

And there’s a strategy an MBA could love! 

But that which sounds good in theory, to paraphrase Yogi Berra, often has a secret dark side in practice. Yes, your statistical models have blind spots. They show that you’ll create more space getting rid of those garish holiday sweaters, but it didn’t show that the lady buying those sweaters was also loading up on pallets of cat food for those dozen felines sharing her home. By reducing that part of your selection, even though the sales it generated looked miniscule in your data sets, you drove away a customer who would have otherwise filled her cart with plenty more stuff. She’s now shopping elsewhere. 

The Long Tail of Selection 

If a retailer could satisfy its druthers, it would stack high and deep only those products with the highest demand, quickest rate of consumption, and fattest margins. Everyone would sell Coke, Tide detergent, Pert Plus shampoo, and Old Spice deodorant. They would spend all their time restocking shelves because the demand for these consumables would keep the products flying out the doors. Selection would be thin, product mix easy to figure out, and all the retail battles would be waged over price. 

Alas, that is not retail reality. 

As we assume Walmart’s Castro-Wright figured out in his ill-fated tenure as U.S. stores chief, a minority of products are responsible for the majority of sales…that Pareto Principle. The curve for demand versus selection looks something like this: 


Most of the demand stacks up with a few products and then tapers dramatically into a long tail where the rest of the selection just doesn’t sell very much. 

When it comes to pressing the selection lever, Costco – with its 4,000 or so SKUs in any given store – only competes in the highest demand, fastest moving inventory items. It clings tightly to the left-hand side of the demand curve, sells in bulk, and competes on price. It’s been a good business, resonating with plenty of shoppers. (See the article, Costco and the Paradox of Choice here.) 

Walmart followed a different path and won the patronage of many more consumers by extending itself much further down the demand slope. It offers about 150,000 unique items in its supercenters. Because of this, it can’t offer the cheapest per unit price like Costco, but the tradeoff it makes with such a wider selection is that shoppers are more likely to pick Walmart as the one-stop place to take care of all their shopping needs. They go in for a festive sweater and load up their carts with everything else they need (or vice-versa). And the prices – while not the cheapest – are cheap enough that they don’t bother adding a Costco stop to their shopping errands this week. 

Why Lose Money on Groceries? 

It’s the reason Walmart introduced groceries to its product mix in the 1990’s. This is a notoriously hard business to be in, requiring a much different, more complex supply chain for sourcing, distributing and selling perishable goods (as opposed to those that can sit in a warehouse for months at a time without spoiling). The logistics of transporting ice cream are orders of magnitude more difficult than those for Old Spice deodorant. 

And it’s not like Walmart sells groceries at a premium, ringing up big margins on the sale of each bag of apples or gallon of milk. To the contrary, Walmart’s prices tend to be 15 percent cheaper* than regular grocers like Kroger. It’s even possible that groceries are a loss leader for the company. 

So why would a business invest so heavily in constructing new additions to its stores, handing over valuable floor space, and increase the complexities of its supply chain on merchandise that loses money? 

Because groceries are the ultimate consumable. People have to buy them. And if Walmart’s best competitive advantage is being the one-stop place to shop, you have to do groceries to get people in the door. 

Offering groceries was the clearest sign of Walmart’s commitment to offering the widest selection. 

As Castro-Wright discovered with his folly, this commitment creates a cascade effect with customers. The wide selection gets shoppers through the doors looking for that low-demand, slow-moving item. They then fill their baskets with more of the high-demand, fast-moving stuff. 

But if you get too fancy with your analytics, stripping out the inventory further down the long tail of the demand curve, the cascade can flow in the opposite direction, too. When shoppers can’t find that esoteric item you pulled from inventory, they might just take their entire shopping cart elsewhere. 

Stores Can Only Have So Much Selection 

But even Walmart, with its nearly 200,000 square feet of supercenter selling space stacked high and deep with 150,000 SKU’s, can only go so far down the demand curve when determining its ideal product mix. It can fit only so much in the confines of those four walls. 

This makes shelf space a valuable commodity. It cannot stuff it with slow-moving items that sit for days or weeks just waiting to be purchased. Its own feedback loop – the productivity loop – depends on high velocity selling…making a bunch of inventory turns to keep up the volume of sales and help press costs down. 

But Castro-Wright came to understand that one cannot be too aggressive in cutting out those slow-moving products either. He pushed too hard, customers left, and he soon followed. 

This is the nature of the selection game for store-based retailers. It’s defined by physical constraints. They can only stock so many items; go so far down the long tail of the demand curve. It requires constant evaluation to find the right balance of products at the right time to keep customers satisfied. Even for mighty Walmart and its commitment to providing the widest selection, it involves trade-offs in deciding what products go on the shelves and which are left out. 

These challenges do not exist for Amazon. We’ll explore that next. 

* The 15 percent estimate is pulled from Charles Fishman’s excellent 2006 book, The Wal-Mart Effect.