When you see a large inventory of any unsold product catching dust, you know its supply has exceeded demand. It could have happened because of multiple reasons – one of them being an overestimation of demand.

This image has an empty alt attribute; its file name is 5e9df8659ea4f32b22ded67b_Group%2052%402x.png

When businesses overestimate demand, they produce a lot more than they can sell. Often the unsold stock is cleared at a reduced margin or loss. Demand planning reduces the probability of such situations occurring. 

What is Demand Planning, and why is it important?

Demand planning is part of the supply chain management function. It predicts demand and optimizes supply accordingly.  

Not planning for demand leads to two scenarios. Either you do not fulfill it due to supply shortage or clog up the supply chain with unsold stock. Both situations have a negative financial impact. On the other hand, just by forecasting right, companies have registered a 3 percent revenue increase.

What is a Demand Plan used for?

With a demand plan in place, the next step is to match supply to demand.

A supply planner uses the demand plan to create a purchase or procurement plan. Further, it is also used to plan other supply chain activities like manufacturing, logistics, and distribution.

How is Demand Forecasting different from Planning?

Forecasting is a part of the planning exercise that deals with the analysis of data. It relies on several internal and external factors. Following are a few relevant to retail business:

  • External factors – weather changes, competition, product lifecycle, changing consumer behavior and preferences 
  • Internal factors – Unfulfilled demand and unsold inventory from the previous year, channel performance and volatility, shelf space, new product launches

For example, while forecasting the demand for laptops, a demand planner considers past data for various market segments. It could include students, small and medium enterprises, and large companies. Additionally, she has to split this demand by product type, specifications, price points, geography, and sales channels. She also has to account for macroeconomic changes, competition, and market trends across different regions. Next, depending on the type of business, a demand planner uses four different types of forecasts with varying time ranges:

  • Micro-level forecast – usually done at SKU or category level 
  • Macro-level forecast – based on channel, regional, or business 
  • Short-term forecast – which ranges from few weeks to months
  • Long-term forecast – involving seasonal, quarterly, or annual forecast

When done right, demand planning is a complex exercise and requires demand planning software.

What are the different demand forecasting approaches?

Forecasting approaches vary with the business size. Let’s understand this with the retail industry example.

Following approaches have been popular in the different sizes of retail business:

1. Qualitative: Small retailers typically rely on knowledge accumulated through experience. Individual preferences and qualitative information influence the plan. Understandably, the data is subjective and biased. It works well for single-store businesses but not for larger chains.

2. Time series analysis: Typically used by mid-sized retailers, it predicts demand based on past values. However, it doesn’t include externalities like competition or fads. Given the rapidity of changes today, it leaves businesses vulnerable to a supply scarcity problem.

3. Causality: Popular among large retail chains, this approach establishes a relationship between various variables and demand. Few variables include weather, competition, and substitutes from the same brand or the competition.

4. Simulation: It’s a result of combining the above three approaches, which leads to relatively more accurate forecasts. It’s usually powered by machine learning algorithms. It is used by chains with a nationwide or international footprint.

How does machine-learning-powered demand forecasting work?

Most traditional demand forecasting solutions assume a linear relationship between cause and effect. Also, they predict demand based on a limited number of variables. However, in the real world, many variables are at play and have little linearity between them. This complexity often leads to a high margin of forecasting error. Machine learning algorithms help address the problem. We can understand it through a common scenario in the retail industry.

Retail demand planning depends on data from various parts of the supply chain. For example, if a particular style of a shirt sells well on the western coast and clocks low sales on the eastern, the overall demand might appear low to the demand planner. Consequently, it leads to a downward spiral of lesser procurement, production, and even lesser sales.

An ML-based algorithm, on the other hand, would analyze several other data points from point-of-sale devices. The analysis could present surprising results which would have otherwise remained undiscovered. Say, the regions that sold a lot of the particular style of the shirt might show high sales for trousers that complement the shirt well. In effect, the combination of the shirt and trousers was a hit. Whereas, the stores on the eastern coast didn’t stock the trousers and hence registered low sales. Therefore, an ML-powered demand planning application helps uncover insights that help businesses boost revenues and avoid opportunity losses.

How to create a Demand Plan that works?

If you wish to create a plan that has a positive growth impact, follow these three rules:

1. Don’t rely on qualitative assessments

Many small and medium-sized businesses rely on expert opinion or simply the leader’s gut feeling to predict demand. Needless to say, it is fraught with the risk of biases and lacks accuracy. 

2. Look beyond numbers, don’t assume incremental growth

A common pitfall in demand planning is the assumption that history repeats itself in increments. When a demand planner uses historic numbers without understanding the larger context, the result is generally flawed.
For instance, the office-wear sales could have been low in 2020. If one were to plan demand with 2020 as the baseline, one must be aware that it was a year of work from home. Adding a percentage increment to it could lead to a severe shortfall in supply.

3. Use intelligent demand planning software

An AI-based software considers many variables leading to an accurate forecast. Using machine learning, it learns the relationship patterns from the historical data. It does not assume an a priori relationship between variables. Hence, it is free of subjective biases.  

Choosing the right Demand Planning software

Manual demand forecasting is both time-consuming and prone to errors. Even if you are a small business, demand planning based on data is always better than a purely gut-based decision. For medium and large-sized businesses, demand planning software could help boost margins and growth.

If you are in the market for demand planning software, you must have come across several options. While picking up a solution, take into account your current business needs and future growth plans.

Opting for a machine learning-based solution takes the guesswork out of your plan. It helps avoid situations where inventory(=money) is stuck in the supply chain. 

Go ahead and choose the demand planning software for your business.