The Monthly Metric: Demand Forecast Error Percentage
The eighth year of The Monthly Metric begins with a look at one of the 17 analytics in the Hierarchy of Supply Chain Metrics pyramid by Gartner, the Stamford, Connecticut-based global business research and advisory firm.
In fact, the metric, demand forecast error percentage, sits at the top. However, that lofty position should not be misinterpreted, says Cristina Carvallo, MBA, a senior director analyst at Gartner. “It’s definitely an important metric,” she says, “but it shouldn’t be overemphasized to a point where (a company) tries to hit targets that aren’t feasible.”
Its position on the pyramid is appropriate: For years, demand variability has perhaps been the biggest impediment to inventory management success. Demand forecast error percentage gauges how much a company missed its projections, whether by overestimating (which keeps money tied up in inventory) or underestimating (which increases the risk of a stockout).
Demand variability, as this space noted in 2018, “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.”
Demand forecast challenges were intensified by the coronavirus pandemic. And as companies began 2024 by waiting for demand to return, the metric is due for a reexamination. “You should monitor forecast error. You should attempt to decrease forecast error,” Carvallo says. “You should ensure that you have a supply chain network and plan that are aligned to whatever your forecast error is.”
She continues, “It is a starting point for everything we do right when we build supply chain plans and determine safety stocks. That means we prepare our supply chains to deliver against demand, projecting it to stay within a certain range of error. But the (error percentage) will never be zero, so don’t become obsessed with trying to get it there.”
Meaning of the Metric
Demand forecast error percentage measures the difference in forecast versus actual sales of a product or SKU. Gartner suggests weighting the percentage based on product, Carvallo says.
“What might look like a big error might not be as big if the percentage is small related to overall sales,” she says. “We recommend a weighted percentage based on what products are most relevant based on revenue cost or sometimes simply volume. That helps planners try to reduce that error where it makes the most difference.”
According to Gartner data, benchmarks vary by industry. In food and beverages, the median error rate is about 25 percent, with the upper quartile at 20 percent, Carvallo says. In the durable consumer products industry, the benchmark is about 50 percent. “That’s a huge error between forecast and actual,” she says.
A badly missed demand forecast can be the result of the economic environment, sudden consumer behavioral shifts and other external factors beyond a company’s control. Or it could be the result of forecast bias, which Carvallo says tends to lean more positive, leading to over-ordering inventory.
“Organizations go through a consensus demand planning process, which means aligning the forecast with the sales and finance organizations,” she says. “Sales wants to be sure that we have enough inventory to support whatever orders come in. It’s more likely that they are inflating that forecast to ensure more inventory is in place.”
That dynamic has been especially prevalent in recent years, with the pandemic’s “stuff economy” and e-commerce boom leading to sky-high demand, which eventually cooled off. As Institute for Supply Management®’s Manufacturing ISM® Report On Business® has detailed in the last year, demand has returned more slowly than purchasing executives anticipated.
“A key is mitigating forecast bias behaviors, which has been a bigger problem lately,” Carvallo says. “A lot of companies are struggling with bias right now, with inflated demand forecasts because hopes are high that demand will come back. But we keep waiting for it.”
How to Fix the Forecasts
Reducing demand forecast error percentage can ease inventory management pressures, but it’s not a panacea: “It’s important for organizations and planning leaders to understand that fixing demand forecast error will not fix all problems,” Carvallo says.
The first step is determining the root cause of a badly missed forecast — over- or under-ordering, or a supply chain breakdown that could be uncovered by another metric. “A lot of organizations report a metric but don’t use it to drive improvement,” she says. “You need to understand why the error occurred before you can make a better forecast.”
Also: (1) collaborate with customers and suppliers, using more external data, (2) develop forecasts using machine learning-enabled algorithms, which better recognize data patterns, and (3) segment your planning approach, so the right level of effort is dedicated to each forecasting combination, based on its value and volatility.
Most importantly, know where accuracy is most critical, Carvallo says: “Try to improve your forecast with the products that really matter, rather than trying to get the same level across your portfolio.”
Reducing demand forecast error percentage is not a cure-all, but the metric’s position at the top of the Gartner pyramid reflects its strategic importance to companies. And more accurate forecasts lead to better decision-making.
“Most supply chain planning leaders are under pressure, especially over the past year, as there’s been higher emphasis on cost and inventory reductions,” Carvallo says. “But demand variability is not going away. We’re not expecting to go back to the worst times of the pandemic, but there are many consumer dynamics impacting demand.
“That represents a big challenge for a lot of organizations.”
To suggest a metric to be covered, email me at dzeiger@ismworld.org.