Guides / Demand Forecast Best Practices

The complete guide to Demand Forecasting Best Practices

This guide provides an introduction to Demand Forecast Best Practices in Retail

Introduction

This guide presents a structured approach to demand planning that emphasizes getting the right products to the right customers faster through deep customer insight, a clear demand planning strategy, and a data driven approach.

There are 6 levels to Demand Forecasting maturity that capture Best Practice approaches, and companies that master them are well positioned to truly deliver excellent customer experience.

This guide for Demand Forecasting Best Practices places special emphasis on building the Demand Forecast the right way for your company.

The 5 Levels of Demand Planning Maturity

Simply knowing what the key components of your demand planning strategy are isn’t enough. There must be a process and a culture in place that give demand planning teams the framework to put the strategy into place. And Demand Planning Best Practices are just a set-and-forget thing, it’s an ongoing process - one that retailers must persist in to continuously anticipate customer behavior, changes in product range, and changes in the market.

In these Levels of Demand Planning Maturity, Level 1 represents inexperienced Supply Chain Teams, and Level 5, the mastery of Demand Planning and Forecasting. As Demand Planning teams progress through each level, their technology and process become more sophisticated and increasingly accurate.

Level 1. Reactive
Level 2. Excel Driven
Level 3. Forecast Driven
Level 4. Intelligent Forecasting
Level 5. Forecasting Excellence

Starting out on the demand forecasting Journey

Historically, Retailers have built their forecasting with statistical approaches, using historical sales to build a prediction of future demand. These forecasts were then modified by manual overrides. More recently, modern retailers have been using machine learning, which provides more accurate forecasts.

A Step-by-Step guide to building an Excellent Demand

Forecast Strategy
Step 1 - Do an Audit of your Demand drivers

The first step is to understand your customer’s buying behavior and audit all the different drivers that influence the demand across your product range. There are a large number of factors that influence the Customer buying behavior for a retailer, this includes internal business activities such as Price Changes, Promotions, Discounts and Range changes. It also includes external factors such as Holidays, Weather, Local Events, Competitor Promotions.

During this audit, it is recommended that you review the following demand drivers:

Demand Sensing Inputs

Once you’ve completed this audit, you then need to understand whether there are reliable and regular datasets that capture these different demand signals. Machine learning algorithms are subject to the rule of garbage in, garbage out. If you give them an inconsistent or dirty dataset, it can degrade the forecast to a point that it becomes unusable.

Step 2 - Identify the Key Decision Horizons

This is a critical step as it helps define the Forecast Horizons that will be chosen in your Demand Planning strategy, what demand forecasting technology can be leveraged and what data streams will be relevant when building your forecast.

The first thing to review during this step is to review your Supplier Locations and the range of Leadtimes. If you're a business that sources everything locally, it completely changes the way you build your forecast as opposed to businesses who source everything overseas.

Decision Horizons for a Retailer that Sources locally are typically defined like this:

Demand Sensing vs Demand Forecasting

Once you’ve worked out the Key Decisions your making at each horizon, you can use the following guide:

1. If you make key decisions across the short-term, I.e. less than 3 months, then you should be investigating Demand Sensing Technology. This technology uses additional data streams, such as weather, events, and competitor monitoring, to provide hyper-accurate short term forecasts.

2. If all your decision making is fixed well before the 3 month mark, i.e if you import all your products or manufacture them months in advance, then you should be looking at more traditional forecasting approaches. Focusing on improving your short-term forecast accuracy is a waste of time because even if you have a highly accurate forecast, you can’t make changes to your plans across the short-term.

Step 3 - Identifying the correct Forecast granularity

Once you’ve identified the key planning decisions you’re making across your supply chain operations, you can then make decisions on the correct forecasting granularity.

Demand forecasts can be set up with different levels of granularity—monthly, weekly, daily, or even hourly—to support different planning processes and business decisions, but highly granular forecasts are always very beneficial (assuming they’re accurate).

If you’re a business who is making short-term allocation decisions or decisions about products with short-shelf life, then you should consider daily or even intra-day forecasts. Whereas if you’re a manufacturer with a 6 month manufacturing plan, then weekly or monthly forecasts would be most suitable.

Getting the right technology

From Fortune 500 Companies, to small ecommerce companies, most retailers use demand forecasts to help guide their purchasing decisions, based on their best estimate of their upcoming sales. Demand Sensing and modern forecasting is an advanced machine learning task that ingests various data streams to provide better predictions. 

While some larger retailers still rely on spreadsheets or traditional statistical forecasts, such forecasting is best executed by software that has been built to be able to handle large data sets and provide forecasts at scale. 

The most advanced of these softwares explains how the forecast has been derived by machine learning, and transparently explains how the forecasts are being calculated.

Once you're Forecasts are setup

Deploy ABC Analysis

ABC analysis is a method of inventory classification that divides items into three categories based on their importance or value. The categories are typically labeled "A," "B," and "C," and they are used to prioritize inventory management activities.

Items in the "A" category are the most valuable or important, and they typically represent a small percentage of the total number of items in the inventory. These items may require frequent monitoring and attention in order to ensure that they are in stock and readily available to meet customer demand.

Items in the "B" category are less valuable or important than "A" items, but they still represent a significant portion of the inventory. These items may require less frequent monitoring and attention than "A" items, but they should still be managed carefully.

Items in the "C" category are the least valuable or important, and they typically represent the largest percentage of the total number of items in the inventory. These items may require minimal monitoring and attention, as they are not considered critical to the business.

ABC analysis is often used in conjunction with other inventory management techniques, such as XYZ Analysis, in order to optimize inventory levels and minimize waste. It can help businesses to focus their efforts on the most important items and ensure that they are always in stock, while also reducing the cost and effort required to manage lower-value items.

Setup both Constrained and unconstrained forecasts

Unconstrained forecasts and constrained forecasts are two types of demand forecasts that can be used by businesses to understand future demand for their products.

Unconstrained forecasts are the view of projected sales, unconstrained by supply capacity. That is to say, if you had perfect instock position and not a single product was out of stock, then you would achieve sales the matched or beat your unconstrained forecast. 

Constrained forecasts, on the other hand, are demand forecasts that take into account supply constraints and other limitations that may impact the company's ability to meet demand. Constrained forecasts are typically used to help companies identify potential bottlenecks or constraints that may impact their ability to meet demand, and to develop strategies to address these constraints.

Overall, unconstrained forecasts are unlikely to actually be achieved. Constrained forecasts, on the other hand, provide a more realistic view of the company's ability to meet demand, and can be more useful for planning and decision-making purposes.

Closing Remarks

In summary, demand sensing offers retailers a significant opportunity to improve their operations by better anticipating demand and therefore enable more effective planning.

Demand Sensing is driving real change for a lot of retailers: A good demand sensing tool can identify the key demand drivers and not only provide more accurate demand forecasts but also provide insights into their inter-relationship.

That being said, Demand Sensing is only as good as the Data you can provide it. So if you have the Key Data streams available and in a clean state, you can look to implement Demand Sensing to liberate your Demand Planners and Managers from the depths of data analysis so they can actually use their expertise, making the critical decisions that a computer can't and making your customers love you.

Find opportunities in your data