When analyzing supply chain risk, logistics managers tend to focus on widespread disruptions such as environmental disasters and pandemics. Indeed, it’s these events that most directly impact the end consumer and capture the attention of mainstream news media.

There are, however, less visible but no less powerful changes that can likewise derail an unprepared supply chain. Among these is demand-led supply chain disruption, dictated by sudden changes in consumer behavior. In our demand-driven supply chain economy, this type of disruption is just as important to plan for.

Though different in nature, demand-led risk can be affected by supply risk. As a result of the COVID-19 pandemic, for instance, consumer markets have seen an influx of designer face masks, HEPA air purifiers and hand sanitizer.

How consumers eat, and from whom they buy their food, has also changed as a result of lockdowns. According to a recent study by EIT, eating habits shifted in 2020, with more people eating at home, ordering delivery or buying groceries online. Those who did eat out in cities chose local, independent restaurants over larger chains. For in-person shopping, many opted for smaller stores instead of supermarkets.

Many logistics managers are well-versed on the risks posed to supply sources in the wake of environmental disasters. What they might be overlooking is the effect such events are having on long-term consumer demand. Faced with the realities of climate change, many buyers are choosing more sustainable products and services. Customers today are asking themselves if there’s a more sustainable version of the product they want, if they really need that item tomorrow, or if they need the product at all.

One of the fastest ways demand can change is through popular culture. Celebrity influencers can change what kinds of clothes or cosmetics consumers want on a dime, when photographed on the street or at a red-carpet event. If the brand benefiting from this exposure hasn’t planned ahead for a spike in sales, it could face backlogs and lose business to its more prepared competitors.

Though these trends aren’t always easy to forecast — who knew that fidget spinners would take off? — certain industry insiders have learned to read the market and can see changes coming earlier than most.

Fortunately, demand-led risk is easier to forecast than supply risk, due to the wide selection of data sources available and the analytics technology designed to process that data. Depending on your supply chain’s maturity, you might need to make changes to your processes and people to effectively utilize these tools. Following are five essential approaches to incorporate into your risk-mitigation strategy.

Capturing new demand inputs. Demand forecasts typically look ahead about 30–90 days. But a window this large is too broad to capture usable insights. By tracking short-term sales history and related demand causals, companies can get near-real-time insights within the month to make more relevant forecasts.

Businesses should also maximize the volume and variety of the data sources they collect. Details such as social sentiment, point-of-sale (POS), inventory and on-shelf availability all help improve short-term demand visibility.

Demand modeling. A demand model helps predict future customer behavior based on past experience. The more external sources you incorporate into your model, the more accurate and predictive it becomes. External sources can include social media feeds, competitive information, weather forecasts and POS data. Coupled with internal data sources such as sales history, promotions and new product introductions, this information can paint a much more accurate picture of past behavior and future trends.

Probabilistic forecasting. When forecasting is based on multiple variables, the old “one number” deterministic approach is too simplistic. By contrast, a probabilistic, or stochastic, forecasting process takes uncertainty into account to help manage risk. With probabilistic forecasting, advanced algorithms analyze multiple demand variables to calculate the probability of each possible outcome, then determine which is most likely to occur. This offers a much more reliable way to make predictions when demand patterns are variable, order history is limited (as, for instance, with new products), or factors like seasonality come into play.

Demand-forecasting software. Choosing the right software is key to effectively analyzing the data you collect. Demand forecasting software that employs a probabilistic approach automatically models bottoms-up demand for individual items. It analyzes order lines to model both historical demand quantities and demand frequency, to give an accurate estimate of volatility. The right system will understand the difference between bulk-ordering 20 units and selling single units of the same product 20 times. It also handles intermittent “long-tail” demand for slower-moving products, which are difficult to forecast. It plans for market factors such as trends or seasonality, as well as organizational factors such as demand-shaping promotions, new products, forecast bias and the bullwhip effect.

Human insight and cross-functional planning. Once you’ve generated a baseline probability forecast, you need people in the business to refine it by adding their knowledge and expertise. Complex demand factors, such as contradictions in consumer behavior, are best unpacked by an entire team of human analysts.

Take an example from the fashion industry. Gen Z buyers tend to prioritize sustainability and prefer “upcycled” secondhand clothing. They are also the most likely demographic to buy from “fast-fashion” companies to keep up with changing trends. In order to make sense of such contradictions, the fashion buyers who analyze trends need to compare notes with merchandisers who are closer to the actual sales figures. In any supply chain, the more people you can involve in refining demand forecasts across finance, marketing, sales, operations and your channel partners, the more accurate these forecasts will become over time.

As the supply chain guru Martin Christopher once said, “Individual businesses no longer compete as stand-alone entities, but rather as supply chains.” Never has this sentiment been truer than it is today. Those companies whose supply chains are best able to sense and react to demand variability will not only be the most resilient to risk; they’ll also be able to ensure the best overall business outcomes.

David Barton is general manager, Americas, at ToolsGroup.