Each January, the pundits are abuzz with talk about the new food trends for the year. Whether it’s antibiotics, sugar levels, food waste, GMO-labeling or cauliflower being the new kale, one thing is for certain: the pace of change in the food industry continues to get faster every year. The implications are significant for manufacturers. Shifts in consumer tastes, the rapid rise of e-commerce and steady SKU proliferation all contribute to increased volatility. This makes matching supply to demand considerably more challenging than just a few years ago. The time-honored practice of simply adding more inventory to buffer against uncertainty runs counter to today’s financial objectives. It ties up cash in an unproductive working capital and erodes margins with excess holding costs and spoilage. As we kick off the new year, what considerations can we make to get a better handle on demand?
E2open’s annual Forecasting Benchmark Study encompasses more than $250 billion in trade from 17 multinational consumer goods companies and published as a public resource to help companies on their journey to forecast excellence. According to the study, the average demand planning forecast error for food items is high, at almost 50%. This means that if your company forecasts ten thousand cases of a particular product will sell this week, half of the time the real number is between 5,000 and 15,000 and half the time it is outside that range. The root cause of this uncertainty is the reliance by traditional demand planning systems. They rely on historical sales to predict customer orders. As we all know, historical performance is a poor indicator of future sales, especially in times of rapid change. So what can be done?
1. Get Current with Demand.
Augment historical sales data with current information across your supply chain. Include data from your channel partners if available. Using daily data puts a finger on the pulse of what is actually happening in your supply chain. It’s also particularly helpful for hard-to-forecast items like new product introductions and seasonal goods. Current information lets you understand how much consumers really like that new yogurt flavor; or just when (and where) the flu season arrives to drive soup sales.
2. Go Hands-free.
Using current information means analyzing lots of data with applications specifically developed for the task. Traditional demand planning systems were designed decades ago to analyze limited amounts of historical data. At the time, this was state-of-the-art technology but now these systems are ill-equipped to handle today’s masses of real-time data. Likewise, trying to do it manually with spreadsheets is error prone and more importantly, would require an army of planners. This is a job for an automated algorithm. As an analogy, consider how Google Maps uses algorithms to sort through masses of real-time traffic data to get you to your destination with the best possible route. Demand sensing algorithms do the same for demand planning. They sort through masses of supply chain data to determine the most likely demand based on current market realities. Daily forecasts are created for every item in every location and get published directly to supply systems for execution. The entire process is hands-free and touchless. It’s as if Google Maps was combined with a driverless car but for the supply chain.
Few of us use maps anymore because our smartphones are much better at navigation and paper maps can’t give real-time traffic redirections while you drive. The takeaway is that automated, data-driven systems create better outcomes. As such, it is only a matter of time before the adoption of demand sensing is commonplace in the food industry. But unlike the paper map example, demand sensing augments, instead of replaces, existing demand planning systems. This protects existing investments while enabling new capabilities designed to capitalize on the rapid proliferation of data all around us.
Case in point, leaders like Campbell Soup, ConAgra Brands, and Kellogg have been quietly sensing the demand for years with excellent results. The average reduction in forecast error compared to demand planning is 41%. Safety stock is proportional to forecast error, so a 41% reduction in forecast error translates into the opportunity to reduce safety stock by 41%, improve customer service or a combination of the two. The use of automation algorithms is an important step in evolving to a digital supply chain. It has become a compelling proposition that has the attention of the analyst community. At its Supply Chain Conference last year, Gartner predicted that by 2018, “25% of companies will have deployed demand sensing and short-term response planning technologies.”
As we start up the New Year, take a moment to pick up your smartphone and reflect on the rapid pace of technology advances you have experienced in the past few years. Consider the masses of real-time data now available across your supply chain. What could it mean to upgrade your demand planning capabilities from paper maps to Google Maps?
This article was originally published by FoodChain Magazine here.