5 Mistakes Demand Planners Should Avoid

Companies base most of their important operating decisions based on the predictions of their demand planners. Demand Planners play a critical role in making a company proactively absorb data from multiple data sources to produce accurate forecasts that are as close to the actual demand while also providing the agility to make changes on the fly and do course corrections to the supply chain. Their work directly influences and impacts customer satisfaction levels and the overall performance in terms of revenue impact and operational cost.?

Essentially, being a demand planner is not an easy job, especially in the last few years!

The pandemic has completely changed the way the modern supply chain works and ensuring supply chain continuity in volatile markets is an uphill task. Traditional ways of approaching the supply chain do not work in the post-pandemic era. Demand planners need to realign themselves, their demand planning, and inventory planning processes to be future-proof.?

In this article, we shall see five pitfalls that?demand planners should avoid:

1. Only Factoring Historical Sales Data for Demand Prediction- Likely To Repeat Past Mistakes

Traditionally retailers used to look at the historical sales data to chart future demand prediction but modern demand planning is a?completely different task. The ever-so-evolving demand drivers, the emergence of new channels, and digitally informed customers have changed the game for demand planners completely. To add to this global healthcare and economic volatility have made it infeasible to predict future demand solely on historical data.

Demand Sensing?is the way to go in the future. It is a new-age method of utilizing AI/ML and other advanced technologies to provide more accurate real-time demand predictions. Demand Sensing provides higher accuracy in demand predictions, and day-by-day sensing by leveraging a high level of data granularity to analyze daily demand information as close as possible to the end customer and immediately detect changes in demand behavior. Factors like cyclicity, seasonality, price changes, promotions, Holidays, and even weather can be utilized to make accurate demand predictions.

2. Sticking to the same old spreadsheet - Genesis Of Rule-Based Rigid Planning

Demand prediction for thousands of SKUs over hundreds of warehouses and stores is a really challenging job, and accurately predicting demand is even more so. Most demand planners like you may be scouring through hundreds of excel sheets, running into thousands of rows, even with missing data, to manually calculate demand predictions and develop replenishment plans. According to the latest study, conducted by a leading supply chain think tank,?70% of demand planners still rely on spreadsheets and manual processes for demand forecasting.?

Demand prediction done on spreadsheets can be a real nightmare as calculations are made via. rigid rules-based approach. These systems are unyielding, inflexible, and computationally limited to factor in the many influencers of demand for a product. Demand prediction over spreadsheets is notorious for low prediction accuracy from rule-based demand estimation & manual adjustments.

This needs to change and is changing. As per the same report,?accurate demand planning capability was the 2nd most important aspect of companies’ supply chain planning process and 75% of them are already employing or planning to employ AI & ML to manage the complexities of supply chain planning. Companies are investing in digital forecasting and planning capabilities where they can plan and adjust demand and inventory planning to the most granular levels.?

3.?Top-down Approach Towards Demand Planning - Little Visibility Into Granular Micros

In the top-down approach, the predictions are made by the demand planners at the highest level in the supply chain, then individual demands are calculated/attributed according to a percentage of the total for each warehouse + channel + store/dark store + category + SKU level. These forecasts fail to take into account the change in demand at the granular levels of the supply chain. A major issue with this approach is that even if the high-level predictions are accurate, inventory operations will not improve unless granular levels of predictions are made accurately. Moreover, the proportions themselves may change, especially if one product starts to sell at a faster rate than others either because of changes in demand trends or because of promotions.?

Today’s demand planning needs a more pragmatic and reliable approach where artificial intelligence is augmented by human intelligence to collect demand data at the most granular levels. Demand is predicted at such granularity and is then aggregated up for overall demand forecasts. In this bottom-up approach, demand is predicted as close to the customer as possible and then rolled up to the top for more accurate demand prediction numbers.

4.?One Fits All Template For Safety Stocks - Sure Way To A Bloated Supply Chain

The size of the safety stock buffer is a crucial factor in enhancing availability, service levels, elimination of out-of-stock occurrences, and customer satisfaction.

The majority of companies operate in a situation where the supplier lead time is significantly longer than the consumer tolerance time. Therefore, to maintain uninterrupted product availability during demand changes, they need to maintain safety stocks at different nodes in the supply chain. This may result in a bullwhip effect where each party in the supply chain carries excess inventory, resulting in locked up working capital and bloated demand numbers for production. On the other hand, if safety stock is not sized properly, it could lead to a situation of stockouts and loss of revenue. So how do you choose the appropriate inventory buffers in a situation with fluctuating demand?

This is where continuous demand sensing at the most granular levels comes into play. AI/ML models are exceptional in identifying patterns and can accurately predict demand at each node of the supply chain and the safety stock can be set dynamically on the basis of demand variability and service levels.

5.?Obsessing Over Forecasting Accuracy At Higher Levels Instead Of Optimizing Demand Forecasts - Looking At The Bigger Picture, Missing Out On The Finer Details.

The perception of forecast accuracy often depends on where you are positioned in the company. At a high aggregation level forecast accuracy looks a lot more refined than it actually is at the granular level. So when executives use forecasts to predict quarterly-country-BU level performance, figures are typically rolled up and the accuracy level is often high, Therefore, the perceived benefit of investing to improve forecast performance is low. This same forecast at the more granular SKU-Store-Channel level is below 50%.

Accuracy in demand predictions is very essential for granular-level purchase and replenishment planning, so while everything might look good to the top executives, a sub-50% demand prediction accuracy can result in excess inventory or stockouts, resulting in higher operating costs, lost revenue, and putting service at risk.

In conclusion, all of the above-listed mistakes can be resolved by the use of machine learning and Artificial intelligence. By implementing machine learning into supply chain planning, the accuracy of forecasting in demand planning can increase 20-30%, enabling reductions in inventory and improvements in revenue.

Kronoscope?is an all-capable, modern AI-powered Demand Sensing and Inventory Planning cloud-based software that can save you from out-of-stock losses and waste, ensuring that you are optimally stocked through all points in your supply chain.