What are the most important demand planning metrics and how do you measure them?

In my view, it's crucial to consider the level at which you conduct forecasting and how it aligns with measuring forecast accuracy. Forecast accuracy can vary significantly when working with detailed versus aggregated data. For example, comparing forecasts at the product group level to individual product forecasts often shows that using product group data provides better accuracy. However, strategies can be developed to break down forecasts to a granular level for supply planning at the product level. Additionally, it's essential to calculate accuracy for forecasts using lag-based approach to assess the variance in forecast for a future time period over S&OP cycles and the actual sales done for the same.

There are various KPIs like Forecast Accuracy, Bias, Forecast Value addition, Inventory Management KPIs But the most important thing must be these factors while we calculate the KPIs 1. Right data from right source must be used to calculate the KPI at the right level of planning in Supply Chain. 2. KPIs must be automated and viewed in powerful dashboard together than manual calculation and viewing them in silos. Power BI is one such example. 3. Deep dive the KPIs number every Week on week or Month or Month to understand the change in number %. 4. The Externa factors must be studied in depth causing the change in these KPIs like over forecasting or under forecasting due to the local climate factor leading to less or more product sales.

Key demand planning metrics go beyond basic Forecast Error and BIAS. I leverage a blend of metrics to uncover drivers of errors and actionable insights. These include Forecast Evolution Waterfall to assess stability over 12 months, Forecast Value Added (FVA), Sales Loss vs. Demand Loss analysis to differentiate demand creation from supply constraints, ABC-XYZ analysis to classify demand patterns, and top contributors in both volume and value variations. This comprehensive approach sharpens forecasting and drives continuous improvement. My analytics team have developed a dashboard of these metrics and this is a continuous exercise to look into demand and not just at a point in a month.

One method of generating new product forecasts is to use demand variations or extensions from existing products, families or brands. Consequently, they draw on the historical data of existing products or families. When combined with causal effects it has to accommodate introductory promotions to derive a realistic and dynamic forecast for new products. Using this approach, new products are assigned a percentage of the parent, family brand, enabling them to proportionately draw a forecast that contains the base, trend, and seasonal elements of the associated category. If the forecast for the category demand deviate from the forecast and exceed a user-defined tolerance, then early warning alert system can notify to take corrective action

It's important to consider the level of granularity of data when conducting the analysis. The forecast accuracy can vary significantly depending on whether we are working with detailed or aggregated data. For instance, comparing forecasts at the category-country level to SKU-ZipCode forecasts often shows that higher-level data provides better accuracy. However, it's possible to develop strategies to break down forecasts to a granular level for supply planning at the product level. Moreover, it's crucial to calculate the accuracy of forecasts using the lag-based approach. This helps to assess the variance in the forecast for a future period over S&OP cycles and the actual sales made for the same period.

There can be situations where you may intentionally seek forecasting bias, depending on your organization's goals. A positive forecast bias might be intentional if you're working to build inventory to insulate yourself from supply chain issues, or have more capacity to go after new business. Similarly, you might be pleased with a negative forecast bias if your focus is on running significantly leaner.

Bias is an important factor to consider when measuring accuracy. It helps to comprehend the tendency of data, which has led to varying outputs in accuracy. For instance, while measuring positive and negative bias, we can keep a check on the continuous occurrence of either of these cases for a certain period, let's say 3 months. This indicates the presence of one-way bias at the level we are measuring. To control this bias and maintain accuracy, it is crucial to focus more on this level in the upcoming cycle. By doing so, we can identify the underlying factors that contribute to the bias and take corrective actions accordingly.

Forecast bias is a very important measure for demand planning. To be a good forecaster, your bias should be 50 50, which means that you shouldn't be biased to over shipment /under shipment. Over shipment: When your Forecast is lower than the actual demand. Consistent over shipment makes you run out of stock as you don't plan for sufficient supply so this causes you service hurt (Customer wants your product but you are out of stock. Customer may go to your competitor to satisfy his need) Under shipment: When your forecast is higher than actual demand. You over estimate your forecast so you build supply that is not materializing into shipment so it costs you unperformed inventory which causes you a tied up cash.

In simple terms, this metric gives a sense if we are over-forecasting or under-forecasting month on month; which directly questions the statistical models we use for forecasting. Let's say we have high +ve bias (Forecast>demand) in last 3-4 months. This signifies that we have been over-forecasting since last 3/4 months and hence we should revisit the model being used to forecast. Normally organizations use bias signal (setting a threshold indicator) to identify such cases and take actions. To sum up, this metric is the patrolling officer of the metric "Forecast Accuracy".

The most important demand planning metrics include: Forecast Accuracy: Measure the variance between predicted and actual demand to assess the effectiveness of forecasting methods. Bias: Evaluate the tendency of forecasts to consistently overestimate or underestimate demand, indicating potential systemic errors. Mean Absolute Percentage Error (MAPE): Calculate the average percentage difference between forecasted and actual demand to gauge overall accuracy. Tracking Signal: Monitor the stability of forecasts by measuring the cumulative deviation from expected values over time, identifying when adjustments are needed.

FVA, or Forecast Value Added, represents the enhancement a demand planner brings to a system-generated or model-based forecast, encompassing methods like ARIMA, Exponential Smoothing, Linear Regression, NBEATS, among others. A positive FVA indicates improvements made by the demand planner, while a negative FVA reflects a less accurate forecast.Notably, even an excessively high positive FVA suggests a limitation in the base forecasting model, prompting the need for manual intervention. In such cases, it is crucial to identify and address the underlying issues affecting the forecasting model's performance.

Forecast Value Added (FVA) is a metric that helps measure the accuracy of forecasting. It calculates the value added by each forecast to the overall forecast. It is crucial in demand planning as it helps identify which forecasts are contributing positively or negatively to the overall accuracy. We measure FVA by calculating the difference between the actual demand and forecasted demand, and then assigning that difference to the specific forecast responsible for it. A higher FVA indicates a more accurate forecast, and it helps improve demand planning and decision-making.

I used this metric in one of my org. to create accountability on folks who provided the bottom up intelligence to planners in terms of forecast. To cut the long story short, we can identify individuals / planners who were able to enhance the forecast using ground intelligence compared to those who wanted to overachieve but were unable to do so. This indicator keeps the planners at bay by considering only validated external inputs which would in turn give a better forecast.

One crucial demand planning metric is the Expected Value Added (EVA), measured by assessing the value contributed by forecasting and planning activities. It involves comparing the projected benefits, such as cost savings or revenue gains, against the resources expended in demand planning processes. EVA provides insights into the efficiency and effectiveness of demand planning efforts in delivering tangible value to the organization.

Service Level metrics are important in demand planning because they measure the effectiveness of meeting customer demand. They help us to understand how well we are performing against our goals and identify areas for improvement. Common metrics include order fill rate and on-time delivery. We measure them by tracking data and calculating the percentage of orders that are fulfilled within a specified timeframe. This allows us to make informed decisions about inventory levels and resource allocation.

In my experience, while service level can be influenced by Demand Planning (forecast accuracy) it is not the only variable that can have an impact on results. Supply Chain performance can destroy service levels, and requires a collaboration between demand and supply to ensure good decisions are made in short supply situations.

Inventory turnover ratios can vary significantly across industries, so it's essential to compare ratios within the same industry or sector for meaningful analysis. Below are some metrics to be considered. 1. Calculate Cost of Goods Sold (COGS): COGS = Beginning Inventory + Purchases ? Ending Inventory 2. Calculate Average Inventory: Average Inventory = (Beginning Inventory + Ending Inventory) / 2 3. Calculate Inventory Turnover: Inventory Turnover = COGS / Average Inventory By regularly measuring inventory turnover and analyzing trends over time, companies can make informed decisions about inventory management, procurement, and sales strategies to optimize their operations and improve financial performance.

Similar to service level, Inventory Turnover can be impacted by Supply Chain performance as well as forecast accuracy. What can be even more impactful is aligning the demand plan not just with customer sales but also with the projected shipment plan timing through the supply chain network. Some will call this a "load in" vs "sell through" demand plan and something that may help most specifically with innovation or new product launches. Which can create point-in-time excess or slower turnover if not aligned through all the metrics.

Inventory Turnover metrics are crucial in demand planning as they help in assessing the efficiency of inventory management. It measures the number of times inventory is sold and replaced within a given time period. A higher turnover ratio indicates that the company is selling products quickly, which is a good sign. On the other hand, a lower ratio indicates that the products are not moving off the shelves fast enough, which could result in inventory buildup. This can lead to a decrease in profits due to increased storage costs and potential spoilage or obsolescence. To calculate inventory turnover, divide the cost of goods sold by the average inventory value.

One of the key demand planning metrics is Inventory Turnover, measured by dividing the cost of goods sold by the average inventory level during a specific period. It indicates how efficiently a company manages its inventory by showing how many times inventory is sold and replaced within a given timeframe. A higher turnover ratio generally signifies more efficient inventory management, whereas a lower ratio may indicate excess inventory or slow-moving goods.

Stockout rate can increase with poor supply chain performance or as a result of isolated events, which, if not tracked, may cause people to adjust the demand plan in error..when the real issue is available supply or a point in time customer buy (that would not be repeated in the future, like increase in purchases of sandbags, due an unusual increase in water level in an area that has not seen flooding for 100 years). The common theme here is that each metric has a part it plays in the cause and effect of another, but is not the whole story, and a deeper analysis of additional leading indicators should be added.

Stockout rate metrics are crucial in demand planning as they help in monitoring the inventory levels of a product and ensure that the demand for the product is being met. Measuring stockout rates involves calculating the percentage of time that a product is out of stock or unavailable for customers. This metric provides insights into the effectiveness of inventory management and helps businesses make informed decisions about inventory levels to maintain customer satisfaction and reduce lost sales. Several methods can be used to measure stockout rates, including customer surveys, point-of-sale data analysis, and inventory tracking systems.

One crucial demand planning metric is the Stockout Rate, measured by dividing the number of instances when a product is out of stock by the total number of opportunities for that product to be available. It indicates the frequency at which demand exceeds supply, leading to unfulfilled orders. Monitoring the stockout rate helps assess inventory management effectiveness and ensures adequate stock levels to meet customer demand without excessive overstocking.

Key demand planning metrics: 1. Forecast Accuracy: Precision of forecasts vs. actuals, using MAPE or RMSE. 2. Inventory Turnover Rate: Frequency of inventory sales, calculated as COGS divided by Average Inventory. 3. Service Level: Percentage of demand met without stockouts, calculated as Orders Fulfilled On Time divided by Total Orders. 4. Stockout Rate: Frequency of stockouts, calculated as Number of Stockouts divided by Total Order Opportunities. 5. Safety Stock Level: Minimum inventory to prevent stockouts, based on demand variability and lead time. 6. Forecast Bias: Average deviation of forecasts from actual demand. 7. Order Cycle Time: Average time from order placement to fulfillment.

Another good KPI for checking demand forecasting is Tracking signal.?A tracking signal is?a statistical measure that indicates whether a forecasting model is generating unbiased and stable forecasts. A tracking signal is calculated by dividing the cumulative sum of forecast errors (CSE) by the mean absolute deviation (MAD).

Apart from the commonly used demand planning metrics, as discussed above, a few other important metrics like Customer Fulfillment Rate, Order Lead Time, Perfect Order Rate, and Sales and Operations Planning (S&OP) Effectiveness. Customer Fulfillment Rate measures the percentage of customer orders that are delivered in full and on time. Order Lead Time tracks the time taken from order placement to delivery. Perfect Order Rate measures the percentage of orders that are fulfilled without any errors. S&OP Effectiveness measures the efficiency and effectiveness of your sales and operations planning process. These metrics can be measured using various methods, including surveys, system reporting, and data analysis.

As with any data measurement, it is vital to begin to look at more leading indicators vs. lagging indicators to get the real story and start to be able to predict the future. Keep in mind that the past is not always an indicator of what the future will bring, we should include it in our analysis.

It’s also good practice to separate your promotional demand from your everyday replenishment. When keeping the data type separate, it’s easier to modify the schedule and calculate how the promotion performed against expectations. We can also monitor how long a promotion took to sell-thru and use those insights to tighten the plan for the next promotion.