How to Forecast Demand | Step-by-Step Guide with Examples

If your goal is to grow profitably, it’s crucial to know how to forecast demand accurately for your business.

Demand forecasting helps you understand how your market will behave, allowing you to align your strategic areas and adjust operations, from sourcing and suppliers to distribution to your customers.

Here are some key aspects to consider when forecasting demand for products or services:

What is demand forecasting in a company?

The demand forecasting process analyzes the quantity of products or services that consumers are expected to purchase across different combinations or hierarchies within your company, including factors like locations, channels, prices, and other market conditions relevant to your business.

Why is it important to forecast demand?

By gathering information about your consumers’ behavior and forecasting the demand for a product or service, your company will be able to understand customer needs and expectations, analyze the competition, identify high and low seasons, and recognize which factors impact the sales of a specific product.

Demand types

At an economic level, various types of demand are distinguished in the market. Here are the most important ones:

Direct demand: Refers to the simplest type of demand, where the focus is on identifying how many people plan to buy a product.
Indirect demand: Related to products that are necessary for the production of others. If demand for the final product increases, the demand for the materials used in its production will also be affected.
Latent demand: Occurs when demand for a product or service cannot be met because consumers are unaware of the solution, cannot afford it, or the product is unavailable.
Joint demand: When two products complement or are directly related to each other, their demand is interconnected.
Irregular demand: This demand fluctuates unpredictably over time, often influenced by fashion trends, events, or seasonal factors.
Negative demand: Refers to products or services that potential consumers tend to avoid.
Full demand: Occurs when demand is equal to or exceeds the production or supply capacity of a product or service.
Over-demand: Arises when demand exceeds the available supply in a market.
Excessive demand: In this case, demand exceeds supply, but at a level that is not healthy or sustainable for society.

Information to Consider in a Demand Forecast

When forecasting demand, multiple variables must be considered to ensure clear and reliable results. To achieve this, it is also essential to consider supply factors.

Here are some key aspects to keep in mind:

Factors that Affect Supply

There are four key aspects of supply that your company should always monitor to achieve sustainable growth:

Price: Price is crucial because if it increases, the supply of the product or service will likely increase, as producers will be more willing to enter the market.
Competitive or Joint Supply: Two particular cases arise. First, if a competitor shifts production from one product to another, the replaced product may become less profitable. In joint supply, an increase in the price of one product can affect another.
Production Costs: If production costs rise, supply tends to decrease, as manufacturing becomes less profitable.
Variation in Resource Availability: When raw materials become scarce, it becomes harder to produce the product, leading to a decrease in supply.

Factors that Affect Demand

When considering factors that impact demand for a product or service, companies must take into account the following:

Purchasing Power of the Target Audience: It’s essential to understand that not everyone has the same income level or the ability to buy all types of products or services. This must be clearly identified by your company.
Preferences and Tastes of Your Audience: A company must understand its consumers’ preferences to better meet their needs and build stronger connections.
Prices of Complementary or Similar Products: Demand for your product will be influenced by the prices of related products, whether they are substitutes or complements. Consumers evaluate the market as a whole.
Volume of Potential Customers: Determine whether your company caters to a niche market or mass consumption.

How to Forecast Demand Step-by-Step

A continuación te compartiremos los pasos que debes seguir para realizar un cálculo de la demanda de un producto o servicio, con el fin de comprender la cantidad de estos que los consumidores estarían dispuestos a adquirir y así, entender mejor el mercado para lograr un crecimiento rentable.

1. Data Collection

Defining your data sources and determining how to access them is the first step when preparing to forecast product demand.

Historical Data:

Examine your historical demand to identify patterns and trends. The time period to consider will depend on the forecasting technique you plan to use.

Remember that demand equals the relationship between orders received and products invoiced. If you’re not tracking stockouts properly, it’s essential to find a way to capture this information to ensure a more accurate forecast.

Typically, the foundation for demand forecasting will come from your ERP (Enterprise Resource Planning).

Other Internal Data Sources:

The acceleration of digital transformation has led companies to adopt new tools for sales and supply chain management, such as WMS (Warehouse Management System), MRP (Material Requirements Planning), TMS (Transportation Management System), CRM (Customer Relationship Management), among others.

You should consider these tools as they can help you achieve a more precise demand forecast, depending on your defined process.

External Data Sources:

It’s becoming increasingly important for companies to identify external data sources that may influence their market behavior, enabling more accurate projections. It’s key to ensure your external data sources are stable and reliable so they can be integrated into the demand forecasting process.

If your industry requires it, you can conduct market research such as surveys, interviews, or secondary data analysis related to your products or services to gather insights on consumer preferences and behaviors.

The more variables and data sources you integrate, the more robust your data processing method must be. This is where machine learning and deep learning models can provide valuable insights from complex data and produce more reliable demand forecasts.

Remember, the quality of your data and sources will directly impact the quality of your forecast results.

2. Identification of Variables

Defining the relevant variables for your business will help establish mechanisms that improve the accuracy of forecasting product or service demand.

Price: Are you aware of whether price fluctuations affect the quantity of products demanded? How volatile is the price behavior of your raw materials, and how does this affect the pricing of your products?
Income: Have you evaluated whether there is a relationship between consumer income and the demand for your products?
Tastes and Preferences: Is your demand sensitive to changes in consumer preferences? How can you gather this information?
External Factors: Do external variables like the economy, politics, demographics, and market trends impact your demand behavior? What are the most significant variables in your industry?

What other variables might affect the demand for your products or services that are considered important in your industry?

3. Select the Demand Forecasting Model

There are multiple models for projecting demand, and selecting the right model is crucial for companies aiming to achieve better results.

Have you noticed that the same product can behave differently depending on how you analyze it? For example, if you examine the performance of the same product across different locations, channels, or combinations, each may show different behavior.

Some will exhibit more stable patterns, while others may be more variable. For this reason, we recommend not relying on a single demand forecasting model for your products or services.

Causal Methods

Causal methods are techniques used by companies to project product or service demand by identifying and analyzing the factors that influence it.

Linear Regression Models: These are statistical techniques used to identify relationships between variables.

y = α + β x

For example, imagine you want to predict ice cream demand (y) based on temperature (x). Assuming you’ve collected data and want to calculate the linear regression:

y = The dependent variable you want to predict, which is “How many ice creams will I sell?”
x = The independent variable, in this case, temperature.
α = The intercept, representing the sales value when the temperature is zero.
β = The slope, which indicates how much ice cream sales change for each unit increase in temperature.

The equation would look like this:

y (Ice Cream Sales) = α (50) + β (2) * x (Temperature)

This means that, according to the model, for every additional degree in temperature, ice cream sales are expected to increase by 2 units.

Multiple Regression Models: These involve multiple independent variables. Multiple regression is useful when you need to predict a dependent variable (like ice cream sales) based on two or more independent variables (such as price, advertising, and temperature).

The multiple regression equation can be expressed as follows:

y = α + β1 x1 + β2 x2 + β3 x3

Based on the previous example, let’s predict ice cream demand (y) using price (x1), advertising (x2), and temperature (x3). Assuming you’ve gathered data and want to calculate the linear regression:

y = α + β1 (Price) + β2 (Advertising) + β3 (Temperature)

The equation would look like this:

y (Ice Cream Demand) = α (1,000) – (50 * Price) + (20 * Advertising) + (2 * Temperature)

This means that, according to our model, sales are expected to:

Decrease by 50 units for each increase in price,
Increase by 20 units for each additional unit of advertising investment,
Increase by 2 units for each additional degree of temperature.

Time Series

A time series represents sequentially collected data points, usually at uniform intervals.

Simple Moving Average

A statistical method used to analyze time series data by smoothing short-term fluctuations to identify long-term trends or patterns. The demand forecast using a simple moving average involves calculating the average demand for a product over successive time intervals using the following equation:

y = (Demand_i) / n

Where:

y = Ice cream demand forecast, which represents the demand estimate for the next period.
n = Number of periods included in the moving average.
Demand_i = Sum of sales in period i.

For example, to project monthly ice cream demand using a simple 3-month average:

y = (Sales in January + Sales in February + Sales in March) / 3

y = (50 + 30 + 60) / 3 = 46.6 → Average of 47 ice creams per month

A simple moving average is useful for smoothing out seasonal fluctuations and highlighting medium-term trends in product demand or sales. The value of n should be adjusted based on the quality of data and the frequency of fluctuations you want to smooth.

Exponential Smoothing

Similar to the moving average, but it assigns weights to historical data, giving more importance to recent information.

Ft+1 = α Yt + (1−α) Ft

Ft+1 = Forecast for the next period.
Yt = Demand in the current period.
Ft = Forecast for the current period.
α = Exponential smoothing factor, which determines how quickly the model reacts to new observations (between 0 and 1).

When α is close to 1, more weight is given to recent data, making the model react quickly to demand changes. When α is closer to 0, the model becomes more stable and less sensitive to recent variations.

For example, if:

Yt = 100 ice creams sold in the current month,
Ft = 90 ice creams forecast for the current month, and
α = 0.2,

Then:

Ft+1 = (0.2 * 100) + (0.8 * 90) = 20 + 72 = 92 ice creams forecast for the next period.

Time Series Decomposition

This method breaks down a time series into its fundamental components, such as trend, seasonality, and error, to obtain more accurate forecasts.

Yt = Tt + St + Et

Yt = Demand in period t.
Tt = Trend direction (increasing, decreasing, or constant).
St = Reflects short-term repetitive patterns (seasonality).
Et = Variation between trend and seasonality (error).

For example:

Yt = 100 ice creams for period t,
Tt = 90 ice creams (long-term trend),
St = 20 ice creams (seasonal pattern),

The error would be: Et = Yt – Tt – St = 100 – 90 – 20 = -10 ice creams (residual error).

Collaborative Forecasting

This approach seeks collaboration among various stakeholders, such as internal departments, business partners, suppliers, or even key customers, to generate more accurate forecasts based on diverse sources of information.

Involving internal departments aligns strategic views across supply chain areas like sales, marketing, operations, and finance. This includes contributions like supplier negotiations, marketing campaigns, and more.

Example:

Marketing predicts a 10% increase in demand due to upcoming promotions.
Sales forecasts a 5% increase based on market trends and the holiday season.
Production identifies supply chain constraints that could reduce sales by 3%.

Using a base forecast of 1,000 ice creams, we calculate:

Collaborative Forecast = Base Forecast + Marketing Contribution + Sales Contribution – Operations Impact

Collaborative Forecast = 1,000 + (1,000 * 0.1) + (1,000 * 0.05) – (1,000 * 0.03) = 1,120 ice creams

Collaborating with external partners, such as suppliers or strategic customers, allows the inclusion of factors like sellout data, marketing campaign participation, and potential supply constraints that could affect service levels.

Artificial Intelligence Models

If you need to analyze large volumes of data and include complex variables, AI models are ideal due to their speed and precision in processing data to project likely demand scenarios.

There are various AI methods for demand forecasting, including supervised and unsupervised learning, such as machine learning, neural networks, deep learning, and generative AI. These enable access to advanced analytics levels like predictive, prescriptive, or cognitive models.

One of the most commonly used models is machine learning, which uses supervised learning to identify patterns from historical data and make future demand forecasts for companies.

4.Scenario Planning

Scenario planning is a crucial step in the demand forecasting process. By analyzing different hypothetical situations or potential changes, companies can prepare for various scenarios that might affect demand.

Price Scenarios: Analyze how different price levels affect demand.
Economic Scenarios: Consider different economic situations and their impact on demand.
Price Elasticity of Demand: Measure the sensitivity of demand to price changes.
Promotion and Event Scenarios: Evaluate the impact of promotions, discounts, or other marketing events on demand to optimize commercial strategies.
Seasonality Scenarios: Consider seasonal patterns to anticipate predictable increases or decreases in demand during holidays or specific events.
Trend Change Scenarios: Analyze and project changes in demand trends due to external factors, consumer preference shifts, or market innovations.
New Product Introduction Scenarios: Project demand when launching new products, considering consumer acceptance and competition.
Supply Shortage Scenarios: Anticipate the impact of potential supply chain disruptions on demand and take preventive actions.
Government Policy Change Scenarios: Evaluate how changes in government policies, such as trade regulations or taxes, might affect demand.

Conclusions

Demand forecasting is a dynamic process that requires a deep understanding of the market and constant adaptation to changes in key variables.

By using analytical tools and mathematical models, companies can forecast demand for a product or service, make data-driven decisions, and respond to changing consumer needs while staying competitive in the market.