SKU proliferation: Too much or not enough?

Xiang Wan and his team were trying to find out if managerial intuition about the pros and cons of SKU proliferation is correct—in effect, whether it is good or bad, or if there is a middle ground.

My twelve-year-old daughter filled her cup with cherry cola at our traditional Saturday lunch spot.  In a revelatory tone she exclaimed, “Someone should make banana flavored soda!  I would buy that.”

The operations manager inside me cringed. "Sure," I asked, "but who else would? What about the inventory, the spoilage, and the transport?"

I explained that her idea had more fundamental implications than it might seem. She was invoking an important debate pertaining to SKU proliferation: When examining a product portfolio, when is more variety needed, and how much is too much?

I told her about an article published in the Journal of Operations Management called “Too much of a good thing: The impact of product variety on operations and sales performance.” In it, Xiang Wan and his colleagues examine the sales and operational performance of a major US soft drink bottler that distributed 328 SKUs from a network of 108 distribution centers.

Wan and his team were trying to find out whether managerial intuition about the pros and cons of proliferation is correct, and if so, which force is more dominant. In effect, he asks whether proliferation is good or bad, or if there is a middle ground.

The theoretical pros and cons of SKU proliferation are pretty straightforward. Proponents argue that the more you segment demand, the better you become at matching the needs of your customers. Customers like this, so sales increase.

Alternatively, SKU-proliferation skeptics suggest that it is costly to manage product variety. Forecasts are less accurate, more administration is required, and warehouse management becomes more complex. As a result, SKU proliferation can compromise operational performance. The study measures operational performance using “fill rate,” which is the percentage of product orders that are actually fulfilled (as opposed to stocking out of the product).

My daughter silently adjusted her straw, so I continued. The impact of increased SKU proliferation on operational performance (i.e., fill rates) is pretty clear—it hurts. Wan finds the costs associated with managing more things are as predicted. The good news, as I explained, is that the degradation of operational performance occurs at a decreasing rate. The worst of it occurs at the lowest levels of SKU proliferation when sales improvements are more likely to offset compromised operational performance. Nonetheless, fill rates continue to fall as product variety increases.

The sales impact of SKU proliferation is where things get interesting. The logic is that sales increase with product variety as a company better serves the needs of unique customer segments. The data support this view. However, as with operational performance, the rate of improvement decreases as proliferation increases. Research suggests this is because new products eventually cannibalize demand for existing products. In other words, if you go to buy regular cola and end up buying it with a “hint of lime” instead, then overall sales have not increased; instead, sales merely shift from one product to another.

“So, sales would increase (but at a decreasing rate) if a company decided to add banana cola,” she said. “It’s all good.”

“Not really,” I responded. We also have to deal with the indirect impact of operational performance on sales. Remember, you cannot sell what you do not have in stock, and the soft drink market is highly substitutable. You might have a preferred brand, but if it is out of stock, you are more likely to choose the other brand than to choose nothing. Therefore, a decreasing fill rate—our measure of operational performance—can have an indirect and negative impact on sales.

Pulling this all together, the critical finding of Wan’s study is that SKU proliferation requires a delicate balance. Variety can be good, but it is possible to have too much of a good thing. In the context of one major soft drink bottler, where the range of SKUs ran between eight and 114, the researchers determined that the optimal number of SKUs was 84. Below that number, the bottler could improve its position by better serving customer niches; above that number, the damage caused by lower operational performance overwhelms those benefits.

My daughter continued to play with her straw, so I continued my explanation.  At this point, I introduced a second critical insight to come from the study: optimal SKU proliferation is something that can be effectively estimated. Wan’s study uses data acquired from a real company. Other companies have the opportunity to use their data to perform similar analyses. Understanding the tradeoffs involved in SKU proliferation can provide critical insight to managers, and this insight is within their reach. “It’s just a math problem.” I concluded, “and you love math!”

She raised her eyes in response. “You’re such a geek, Dad,” she concluded. “All I said was a banana cola would be good.”

“Indeed,” I replied. “Nonetheless, successful companies look for ways to use the data they have to understand the tradeoffs between top-line and operational performance to maximize their most important metric—the bottom line.”

And with that, our impending feast of chicken fingers and fries led us to other distractions.