Most organizations do a poor job forecasting, with just one in five coming within 5% of forecasts. This statistic is staggering, and it implies that although many organizations understand the value of forecasting, the majority of them are doing it inaccurately.
Forecasting is a combination of science, technology and business process, but few organizations take a holistic approach to improve their forecast accuracy in a sustained manner. In this post we’ll discuss some of the major reasons CPG companies struggle to improve forecasting accuracy.
1. Information technology is not enabling
Believe it or not, there’s such a thing as too much technological enthusiasm, and some CPG companies have reached a point of over-technologization. Examples include system over-engineering by IT and unfriendly user interfaces.
Other companies are crippled in the opposite way – they’re overly reliant on outdated IT. Many organizations rely exclusively or partially on spreadsheets for demand forecasting. Sure, these forecasts may seem helpful, but they’re much less accurate than more advanced forecasting tools.
2. Internal data isn’t perfect
Companies need to understand that raw data isn’t perfect, and it may be incomplete, and at present, many CPG companies’ data cleansing, outlier treatment, imputation and analysis efforts fall short. Considering the SKU-location combination for CPG companies, the volume of data tends to be huge. In some situations, the most granular combination of SKU- location can go up to several million records.
Data correction is an intensive and ongoing effort, but the undertaking is worth it, and the payoff is more accurate forecasting.
When building forecasts, CPG companies should also look beyond internal data to include reports from government agencies, economic organizations and other trade groups, for an added layer of context. External resources can help you understand consumer and economic drivers.
3. The demand planning process is not streamlined
Without well-defined processes and goals, no technology can deliver desirable results, and CPG companies oftentimes fall victim to this common misstep. For more reliable forecasting results, CPG companies must delfine their goal at the outset, while also defining their processes, ones that include data cleansing.
4. Science is not up to the mark
Frequently, the underlying math or algorithms used in forecasting aren’t cutting-edge.
Data science is a field separate from IT, marketing, and finance, so CPG companies must bring in data experts, like data scientists and statisticians, to construct useful models for their data.
Data scientists don’t come cheap, with salaries in the six-figure range, but the opportunity cost is lost profits from unreliable statistical modeling, and these tend to significantly outweigh the salaries of data scientists.
5. Overlooking the cash cows
Demand forecasting tools are great at uncovering hidden insights, and many CPG companies use them to better understand their worst-selling products and anticipate unexpected changes in consumer trends.
But companies are quick to overlook their “cash cow” products, and they generally forget forecasting can also help reap even more revenue from their best-selling products. Furthermore, segmentation is rarely implemented properly, and thus, not leveraged fully.
When defining your forecasting goals, it’s important that you don’t ignore the products contributing most to your bottom line.
6. Inherent business bias
In spite of the many benefits of advanced data science and leveraging statistics, there’s sometimes resistance from management and leadership who may even say, “Sure, that’s what the analysis says, but my hunch says otherwise.”
Whatever business knowledge and experience they have should not be overriding, as human intuition is inherently flawed. For example, many companies still use historical trends, ones that don’t necessarily repeat themselves in the future. What’s more, leadership may be viewing forecast accuracy solely in isolation, and in some instances, a 2% improvement in forecasting accuracy means significant improvements in the number of expedited orders and more.
Improving forecast accuracy is possible
Demand forecasting takes more than simply executing an a “ magical” algorithm for results; data needs to be cleansed, management needs to lay out concrete processes and best practices for analyzing and understanding results, and in some cases, companies must overhaul their approach to demand forecasting altogether.
Don’t be unsettled if some of the factors above resonate a bit too much with your CPG company – understanding the challenges tied to forecast accuracy is the first step in improving them.