Though demand forecasting and demand planning are often conflated, they are not the same thing. Demand planning is the process of creating forecasts—the more effective the demand planning process, the more accurate the forecasts—and implementing a supply chain to support that vision of future sales.
The research and data science strategy a company uses is therefore of the utmost importance for retailers and CPG brands alike. Accurate demand forecasting can establish trends to help retailers and brands predict how much inventory is likely to be sold at various locations—taking into account seasonal trends, major recurring events, and other factors. For example, Nestle has saved millions of dollars with accurate demand forecasts, while multi-billionaire retail company Walmart has repeatedly cut their warehouse staff, not realizing that this money-saving measure would lead to empty shelves in their stores. This case study about how the largest importer of tilapia in the world saved millions on carrying costs during Covid-19 is also a lesson in the importance of accuracy.
Why demand forecasting is essential to brands with a retail presence
An accurate, SKU-level forecast is the key for a CPG brand’s production plan, budgets, and other supply chain strategies. It's essential to know much cash and resources each department will be using, from manufacturing to marketing and beyond. If a demand forecast is inaccurate, it throws every aspect of the business off that can hinder growth potentially even shut down business.
Target, Walmart, Costco, and Kroger are all examples of retailers who tend to place large orders. As a brand, you need to make sure that you have enough product to meet their demand. If you can't meet retailer demands, then they will turn to a competitor for their needs. Some of them may stop purchasing from you all together if you can't consistently fulfill orders. And fulfilling orders—keeping inventory moving—is necessary, as many brands cannot just have a ton of extra inventory sitting around. The perishable goods they sell will go to waste.
Why high forecast accuracy is vital in retail
Extra inventory sitting around also ties up working capital: a product that costs $100 to make requires a production run, then shipment to your warehouse, then shipment to your customer, and then waiting typically about 60-120 days for your customer to pay you. That entire time, the $100 you spent on the product is tied up.
Often CPG brands borrow that money (bank loans, commercial paper market, fixed income market, other lender), but this borrowed money, of course, has an associated interest rate. Therefore, it both costs you money to keep extra inventory to meet large potential orders, and the loss of a potential customer if you don't have enough inventory to satisfy a customer order.
Accurate and consistent (low deviation) demand forecasts are critical to financial planning. Most CPG investors and executives want financial plans and projections that typically begin with sales forecasts, and also involve conversations regarding how much profit to expect and how much working capital is needed. Basically, by having more accurate forecasts, you need less working capital for the same amount of sales.
For further insight, you can read about for the impact over excess inventory HERE, and the impact of over/under-forecasting HERE.
The dangers of inaccurate forecasts in retail
Non-automated demand forecasting methods, especially manually keeping track of data on spreadsheets, like many brands do, do not typically take different, parallel data sets into account or to provide a singular view. These data sets include historical sales data, of course, but also market research, consumer research, resource and capacity planning, stock replenishment, and possible societal factors ranging, such as a recession or public health crisis.
Old-school methods of forecasting cannot optimize all of this data as market trends change, and brands then take hits to their sales because of less inventory, or over-stocking. Holiday seasons, promotions, Black Friday, and Cyber Monday, while exciting opportunities for brands, prevent most from accurately forecasting sales.
Imagine a situation in which a product is sold out at one store while excess inventory is costing you warehouse space or, worse, nearing spoilage, and you can see how slightly inaccurate forecasting can cost a company thousands or even millions. This includes a deviation of as little as one percent. One last loss brands take when their forecasts are off: they have to rely on safety stock when they’re unsure about potential sales, because they want to insure their business against losing sales stemming from a slim inventory.
The CPG world dealing with COVID-19 gives insight into how important demand forecasts can be, and how many variables are involved in demand planning. When the lockdown was brand-new, brick-and-mortar retailers saw panicked customers pantry-loading, and their consumption preferences changed rapidly. There was also a bottleneck effect: As CNN explained back in March: “Food that had been destined for restaurants, bars, offices and other gathering places will need to go to homes instead, and the system will have to account for the increased volume of groceries Americans cooking at home are suddenly buying.”
Some small businesses had trouble keeping up with customers purchasing more low-cost items because of fears about economic health. But COVID-19 isn’t the only reason retailers should invest in demand planning software; the pandemic is only an unfortunate example of how unpredictable demand can be and how fraught with error most demand forecasting is.
How can CPG brands accurately forecast demand and get the most out of their software?
Organizations should adapt now and put into place the tools that will enable them to remain competitive into the future. Unforeseen variables will continue to drastically impact consumption and the global supply chain. The next step towards building up your demand planning process to match? Demand planning software.
AI-driven platforms take in data and learn on the spot. Besides enough products on the shelves, another point towards using forecasting software is that brands can use the data they provide to plan their upcoming product launches and promotions, which as mentioned above, can be the most tricky events to plan for. And forecasting platforms offer cleaning algorithms that are applied to your data sources prior to being pushed to predictive models to shore up any gaps in data that may exist. AI-powered demand planning software allows retailers to predict sales across all channels and drill into predictions by customer, ship-to, brand, segment, product and more, view accuracy and error reports at the SKU level, and filter SKUs by customer, channel or segment, and easily adjust your forecast or create your own manual forecast based on different sales scenarios, plans, or budgets.
Best practices, tips and techniques for demand forecasting in retail
- Get the most value out of your data. You need to be able to collect insights from your historical sales data, but also your product and customer data to make accurate demand forecasts. While vital, historical sales alone (these are your shipments at the SKU-level by customer, ship-to, and ship-from locations) cannot necessarily enable you to predict rapidly changing consumer behavior. This is why consumption-driven forecasts are a must; depending solely on supply data to predict future is no longer good enough. Comparing your forecasts to real market results and identifying the data sets or variables that weren’t considered in the original forecast allows brands to lower their chances of blind spots for the next forecast. It’s important to be discerning about metrics: the capabilities of a successful software include measuring your historical sales (shipments at the SKU-level by customer, ship-to and ship-from locs.), marketing and trade investments (promotions, etc.), online search behavior and social media mentions, point-of-sale (POS) consumption data at both chain and store levels (foot traffic), seasonality (and other fluctuations related to time), weather patterns, and ingredient and health trends. For businesses without much historical data, qualitative metrics like focus groups and expert opinion can be the way to go.
- Get the right software. When you settle on metrics that will be most productive for your company, it’s time to decide how often to collect data—your software could measure your data points weekly, monthly, or you could choose different rates of collection for different data categories. Currently both brands and retailers are attempting to forecast demand three, six, nine, twelve, etc., months in advance, which, as COVID-19 made clear, is not always the most reliable way to plan. Ideally, software that collects data as close to real-time as possible is the best option. If a brand uses omni-channel marketing and sales strategies, they will need to aggregate all of their collected data into one set, too, so a demand planning platform that can clean your data for- you is of the utmost importance. Aggregating data for all of a brand’s prominent SKUs will let them figure out the most valuable channels. The primary driver of demand is customer requirements, so retailers shouldn’t forget to use their demand planning platform of choice to forecast at the SKU, location and customer planning levels.
- Call in the data scientists: AI, automation, and algorithms at work. After aggregating and cleaning your data, data scientists will complete a process of combining their in-house data with various external data points, like the ones discussed above. Different machine learning algorithms will be tested, until the data scientists make a predictive model that can meet or even exceed standards. From there, a team of software engineers can use these models to create a scalable solution that interfaces with your CPG brand’s specific data and systems. Essentially, combined with today’s powerful computing capabilities, machine learning algorithms allow us to view and leverage data in ways that weren’t before possible. Since predicting SKU-level sales using spreadsheets is not tenable, given the amount and diversity of data that should be accounted for when forecasting sales, inventory, manufacturing and distribution, it’s in CPG brands’ best interests to reimagine their demand planning processes and improve their forecasts.