The importance to a business 

Demand planning directly impacts your bottom line in two major ways: 

You need enough stock to meet demand, but not so much that you’re wasting space or letting products expire or become obsolete.  

The impact of a better demand plan depends on the type of business, but here are some resources that try to translate the benefits into financial results:  

McKinsey estimates that improving the forecast accuracy by 10-20% translates into a potential 5 percent reduction in inventory costs and revenue increases of 2 to 3 percent. 

The Insitute of Business Forecasting estimates that a 15% forecast accuracy improvement will deliver a 3% or higher pre-tax improvement.  

Blueridge’s ROI of forecast accuracy (Summary: A 15% forecast accuracy improvement will deliver a 3% or higher pre-tax improvement is the summary)  

Nicolas Vandeput’s translation of a better forecast into inventory levels & product shortages. (Summary: a 10% forecast error reduction would result in a 4% inventory reduction or a 6% shortage reduction) 

Note that when we speak of improvements of a few percentage points in the domain of demand planning, this could have a huge impact on the overall business.  

Distinction from supply planning  

Several businesses have demand planners who are essentially also executing the role of a supply planner. Therefore, a brief traditional distinction between demand planning & supply planning seems necessary.  

Put simply, demand planning is about figuring out how many products customers will need. Supply planning, on the other hand, is about whether you can meet that demand and how best to do it. 

Key differences between demand planning and supply planning 

From a demand planning perspective, it’s crucial to translate customer demand into actionable information for supply planning. While demand planning usually focuses on product categories, supply planning requires more granular details, like knowing exactly which warehouse will need which specific product, or which plant will produce it and when. 

Demand planning language 

If you want to become familiar with demand planning terminology, this article, “The language of demand” by Lora Cecere, is a great read.  

You'll get to know terms like demand sensing, demand latency, independent demand, dependent demand, demand shaping, demand shifting, forecastability, forecast value add (FVA), naive forecast, demand visibility, demand consumption, integration, and harmonization. 

For more in-depth resources on how demand planning is different from supply planning: 

The Insitute of Business Forecasting & Planning (IBF) on the difference of a demand planner vs a supply planner (in organizations, these roles are sometimes intertwined so this is an interesting one if you want to split responsibilities) 

If you’d like to learn about IBF’s elaboration of differences in the role of a demand & supply planner in video/audio format: Supply planner vs demand planner: the difference in role (Youtube podcast by IBF) 

Blueridge’s article on the differences between demand planning & supply planning 

Intuendi’s blog article on demand planning vs supply planning 

Two key steps of demand planning  

At a high level, a demand planning cycle involves two basic steps: baseline generation and adjustments. Let's break these down. 

This step is all about building a foundational forecast using historical data. The goal is to capture the baseline demand, identify any seasonality, and spot trends for each product or product category. As demand planning matures, companies typically evolve their methods through several stages: 

Stage 1: Basic historical averages 

Early on, companies rely on simple methods like using last year’s sales or averaging past performance to create their forecasts. It’s basic, and it doesn’t consider external factors or trends. 

Stage 2: Statistical methods 

As things progress, statistical techniques like moving averages or exponential smoothing come into play. These help account for seasonality and trends, making the forecast a bit more dynamic. 

Stage 3: Machine learning models 

Machine learning models handle large datasets, detect patterns, and improve accuracy over time by learning from historical data. They also adapt more quickly to changes in customer behavior than traditional models. 

Stage 4: Best-fit models 

At this point, companies start comparing different forecasting models (statistical or machine learning) for each product or category and select the best-performing one. You're letting the variety of models compete to see which works best. 

Stage 5: Blended models 

The most advanced companies mix different models for different periods or product categories. For example, they might use statistical models during stable times and machine learning for more volatile periods. 

Once the baseline is set, you’ll make adjustments to take into account important drivers like promotions, customer contracts, or special deals that could impact demand. This step requires collaboration across teams and evolves in maturity as well: 

Stage 1: No input from sales 

Early on, companies might just rely on the baseline without any input from the sales team. This can lead to missed opportunities to improve the forecast with market conditions in mind. Sometimes, there is no statistical baseline and the input from sales is directly used for the demand plan. This is also a low-maturity level.  

Stage 2: Ad hoc sales input 

At this stage, companies start incorporating sales input, but it's informal. If a big promotion or deal is happening, adjustments are made on the fly, but there’s no structured process and the change in demand is not actually linked to the driver, so that someone looking at it next year will not know why an increase happened. 

Stage 3: Structured sales input 

In a mature setup, companies have a formal process for collecting sales insights regularly, like monthly. This ensures that the forecast is updated in a more organized way, using market intelligence and customer feedback.