Reprinted from the original article publish by the Institute for Supply Management, December 2018.
A change of the calendar is always a time when supply management professionals think about challenges in the coming year. However, if recent history is a guide, there won’t be much debate on the biggest challenge for procurement in 2019.
Throughout the decade, surveys and studies have identified demand variability as the biggest impediment to supply management success. The variability is the difference between a demand forecast’s purchase projections and the amount of a product that was shipped. As an analytic, it makes many other metrics necessary — if customer demands never changed, and there were no natural disasters or other unexpected events to impact production and distribution, there would be little to measure because supply chains would run smoothly and consistently.
What makes demand variability so difficult is that, while you can plug in data and get a reading, it involves more than math, says Tracey Smith, MBA, MAS, CPSM, president of Numerical Insights LLC, a boutique analytics firm in Charlotte, North Carolina. Judgment and instinct also are required, she says, when weighing how much to spend for safety stock of a product against the risk of not having it when demand spikes.
“You get a little bit of math and a little bit of judgment, and that’s how you have to create your demand forecast,” Smith says. “You can’t forecast the unknown … There may be a surprise election result. There are normal variations in the economic cycle. There are always various business disruptions that happen. That’s why it’s always going to be a challenge.”
Meaning of the Metric
Demand variability is calculated with the S/X ratio — the standard deviation (S) divided by the mean (X). A low S/X indicates consistent demand; high S/X more sporadic. However, Smith says there is no benchmark number — S/X depends on the product and its cost, and each company must decide how much variability it can afford to plan for. This task gets more challenging for a manufacturer that serves several industries, when demand for a part from one industry increases and goes down in others.
Some manufacturers, Smith says, divide variability into categories of parts. “There might be a highly volatile part, and it’s expensive, but it also might be the most profitable product a company produces,” she says. “So, there’s more consideration to put into it after you’ve done the math to figure out variability. … How much does the part cost? If it’s cheap, it might not be a big deal. Can the supplier ship within a week, or will the lead time be longer if you need extra parts? So, there’s a lot of things (a supply manager) needs to discuss.”
Demand variability is a term that is often used interchangeably with demand volatility, even though the latter sounds much more daunting. Smith says she prefers to use variability until the S/X ratio on a part gets high enough to suggest true volatility. “When you get to the high levels of variability, you can call it volatile because that’s when you have to decide what to do about it,” Smith says.
Some supply management professionals make a distinction — calling demand variability a test of an organization’s planning and demand volatility a test of its response.
Getting a handle on demand variability can provide a manufacturer more clarity on inventory levels. “For a potentially volatile part, it can be hard to determine what to do,” Smith says. “If (a company’s) average is 1,000 parts out the door each month, but it bounces between 400 and 1,200, there needs to be a decision (on a monthly basis) on covering the 1,200 parts or taking a risk on having fewer in stock. So, that’s when it starts to get statistical on what to do.”
Smith cites a manufacturing company that conducted a demand-variability analysis for each of the more than 6,000 parts it uses to make products. Customer-shipment data was used to determine mean and standard deviations, which helped identify parts with highly variable demand. With big help from a “near-real-time” purchasing data dashboard that enabled parts analysis by category, supplier and other features, each highly variable part was classified, based on (1) the profitability of the product(s) it is part of, (2) lead time, (3) cost and (4) its recent-years purchase trajectory (up or down). The manufacturer got a clearer picture on the parts it needed to safety stock and which ones it was willing to take risks on.
“(With that kind of data), a company can prioritize based on the variation relative to the mean and the worth of a part,” Smith says. “It can prioritize and (focus) on parts that suddenly emerge as sporadic. Sometimes, the more parts a company has, there’s less visibility into the information about them.”
Until a crystal ball becomes standard procurement equipment, demand variability will likely never stop being a challenge. While the right mix of math and judgment will not eliminate the unknowns, Smith says, it can help companies better navigate them.