Three Ways to Increase Demand Forecast Accuracy in a Volatile World
Has there ever been a more volatile global supply chain environment? As new headlines constantly remind us, the world’s supply chains are operating in an increasingly precarious business landscape. The high costs and uncertainty of raw materials, components, labor and transportation have led to some significant product shortages ― most visibly, semiconductor chips impacting the automotive industry, but other products as well. For example, as consumers return to travel, they may not be able to find airport snacks.
When products are scarce, it becomes even more critical to create an accurate demand forecast, so you can match supply and demand in a way that optimizes both profitability and service levels. Sending your limited supply of products to the wrong place in today’s world can be disastrous, as you have stock-outs and lost sales in most markets, but pockets of excess inventory sitting around eroding your margins somewhere else.
Legacy planning systems and spreadsheet-based, manual demand planning processes used by many of the world’s manufacturers and retailers haven’t caused today’s product shortages and other supply chain issues — but they’re only exacerbating the situation.
In a typical organization, multiple functional teams develop their own siloed forecasts, leading to a lack of alignment and consensus across the organization. In addition, every team runs its own forward-looking scenarios, importing data from different sources and, thus, arriving at different conclusions. No two teams can agree on a single demand forecast, let alone a plan for allocating products. And, because manual processes and data silos are prone to error and not all data has been vetted for accuracy, there are plenty of forecasting mistakes along the way. The result? Stocks-outs, overstocks, lost sales, eroded margins and missed revenue targets.
When both supply and demand were more predictable, these manual, time- and labor-intensive processes might have produced results that were “good enough” to roughly match products with customers. But, in today’s world of massive upstream and downstream volatility, new modern solutions and best planning practices are required for companies to survive, let alone thrive. Demand forecasters and planners need to do better.
Best Practices for an Unpredictable Marketplace
While there’s no simple answer to precisely matching scarce products with customers in today’s environment of extreme uncertainty, Blue Yonder offers three best practices in demand forecasting that can drive a significant improvement in accuracy.
1. Collect direct customer insights.
What are your customers looking for? Maybe you should start by asking them. B2B companies can often access supplier portals and data interchanges that include a detailed projection of future customer demand. B2C companies can gather a range of data about consumers’ needs and behaviors on a localized basis — and it goes without saying that, in today’s fast-changing landscape, this data should be gathered as near to real-time as possible. These direct customer insights provide a strong foundation for a single demand forecast that will be shared among all stakeholders.
2. Leverage third-party data.
Demand may be unpredictable, but there are nearly always signals of upcoming demand changes. Your data stream should include inputs from customers, but also reflect market trends, competitor insights, social media, events, weather and news. Because this third-party data is going to be both structured and unstructured, it’s critical to have the right tools and processes in place to capture it, store it and begin the task of interpretation. This third-party data, gathered in real-time, can be combined with direct customer sales projections to create a shared data repository that’s the basis for the forecast.
3. Apply artificial intelligence/machine learning.
Without a doubt, this is the most important best practice ― and the one where most forecasting organizations fall short today. The current generation of demand analytics engines, supported by artificial intelligence/machine learning (AI/ML), offer truly incredible capabilities for scanning enormous volumes of data, applying proprietary algorithms, identifying patterns, sensing deviations and arriving at an extremely accurate, localized demand forecast.
Today’s advanced analytics not only produce a single forecast that’s reliable enough to share across the business, but they create a dynamic forecast that’s rapidly, seamlessly and autonomously updated in near real-time. As market conditions inevitably shift, the forecast evolves to reflect the current reality — enabling manufacturers and retailers to quickly pivot and re-align supply with demand.
The Blue Yonder Difference
While forecasting teams have done their best to make sense of today’s unpredictable business world, the current level of demand complexity is simply too much for human cognition. Spreadsheets, manual calculations and labor-intensive processes can’t keep pace with the rapid and continuous changes in today’s markets ― and they can’t manage the complexity of accurately predicting demand in every location, every day.
Blue Yonder’s demand planning solution on the other hand, is purpose-built for this task. AI-enabled optimization engines ingest huge volumes of real-time data from different sources, apply intelligent algorithms and sophisticated segmentation strategies, run scenarios autonomously and create highly accurate forecasts with a glass box approach. And the forecast is updated easily as conditions change, even multiple times per day, without the need to wait for manual processes and consensus-building meetings. Thanks to ML capabilities, the forecasting engine learns from real-world results to become more and more accurate all the time.
“Demand planning from Blue Yonder enables us to meet every customer where and when they want to shop. We can create a broad network strategy that’s based on meeting very specific, granular customer needs,” says the Vice President Solution Delivery for Sally Beauty. “We know what levers we can pull to make some adjustments, we learn from a new set of configurations, and then we go forward from there. We don’t have to make one decision and live with it forever. If the pandemic has taught us anything, it’s that we’ve got to be nimble. And Blue Yonder has given us that agility.”
When your organization replaces manual processes with the automated power of AI/ML, the role of your human planners becomes much more strategic. Instead of doing hands-on tuning and cleansing activities to generate the forecast, planners can now focus on the result. As they review the forecasting outputs and manage exceptions, employees leave their functional siloes behind and focus on what’s best for the organization as a whole.
Unfortunately, both demand and supply volatility seem to be here to stay — at least for the foreseeable future. But, by relying on AI/ML and best practices, your forecasting team can pivot with both intelligence and agility as conditions shift. In today’s challenging environment, the winners will be those companies that act with speed to adopt new forecasting solutions and processes that help transform uncertainty from a hurdle into a competitive advantage.