Forecasting can be defined as a technique that makes use of historical data to predict future events – an important tool in making informed business decisions. Past data is collected and analyzed so that patterns can be found. In this way, trends such as sales expectations and customer behavior can be anticipated, and by using the information, businesses are able to make future plans and develop business strategies. It goes without saying that accurate forecasting requires specific skills and correct data.
Successful forecasting depends on going beyond common sense and implementing the most advanced analytical techniques to get the most out of your data. In practice, this would mean finding and exploiting correlations.
Forecasting methods aim to exploit these correlations in order to predict what is predictable, in the best possible way, with all available data. Many classical and elementary forecasting methods, like those easily found on spreadsheets, just try to correlate future behavior with past history. This is already a very important step and by no means trivial: look inside the Intuendi demand forecasting software solution and compare it with standard software: you will soon notice that there might be a big difference in the quality of forecasting, even when it is based only on past historical data of a single product.
But of course, things become more and more relevant when multiple correlations come into play: correlations within a family of products, within a region, within different channels; correlations with external data; correlations with predictable events (holidays, planned discounts, promotions). We at Intuendi are committed to deliver to you the most accurate forecast with all of the available analytical tools, in order to deliver value to your business.
There are two top-level inventory demand forecasting models to consider when calculating demand: the quantitative forecasting model and the qualitative forecasting model. It is important to understand the difference between these two models. Qualitative forecasting relies on subjective judgment and qualitative data sources to make predictions based on expert opinions and market insights. On the other hand, quantitative forecasting uses historical numerical data and statistical methods to analyze patterns and make predictions based on numerical trends and relationships between variables. The choice between qualitative and quantitative forecasting depends on the availability of data, the nature of the forecasting task, and the level of precision and accuracy required in the predictions.
The 4 Basic Types of Forecasting
The four basic types of forecasting include the following: qualitative, quantitative, time series-based, and causal.
Qualitative Forecasting is often used when historical data is scarce, unreliable, or unavailable, such as for new products, emerging markets, or disruptive technologies. It emphasizes an understanding of market dynamics, consumer behavior, industry trends, and other qualitative factors that may impact future outcomes.
Quantitative Forecasting, on the other hand, employs mathematical and statistical models to analyze historical patterns, identify trends, seasonality, and relationships between variables, and make predictions based on numerical data. It is suitable for situations where historical data is abundant, reliable, and follows observable patterns, such as in sales forecasting, demand forecasting, financial forecasting, and inventory management.
Time series-based forecasting uses historical data to forecast future values based on past observations and focuses on analyzing data points collected at successive, evenly-spaced time intervals to predict future values or trends. It is commonly used in areas such as sales forecasting, financial forecasting, stock market analysis, and economic forecasting.
Causal forecasting identifies and analyzes the cause-and-effect relationships between variables and requires historical data on both the dependent variable and potential causal factors. It is used when there’s a clear cause-and-effect relationship between variables, such as in economic forecasting, supply chain management, and marketing analytics.
Qualitative Forecasting
Qualitative forecasting is important for helping executives make decisions for a company. With qualitative demand forecasting, predictions are based on expert knowledge of how the market works. These insights could come from one person or multiple people, both internally and externally to the business. There are a number of qualitative forecasting methods.
Delphi Method
The Delphi method is an iterative process that collects and synthesizes forecasts from experts through successive rounds of questionnaires. It was originally conceived in the 1950s by Olaf Helmer and Norman Dalkey of Rand Corp. The name refers to the Oracle of Delphi, a priestess at the temple of Apollo in ancient Greece known for her prophecies. Several rounds of questionnaires are sent out to relevant experts – these can include customers and suppliers.
Intuitive Forecasting
This method relies on personal intuitions and past experiences. Rather than relying on empirical data, statistical models, or formal forecasting techniques, it involves tapping into one’s gut instinct to anticipate future events, trends, or outcomes. As such, it is subjective and prone to bias and errors. Despite its limitations, intuitive forecasting can be useful in situations where data is limited, ambiguous, or uncertain, and where quick decisions are required.
Judgemental Methods
Judgemental methods rely on human subjective judgment and expertise of individuals or groups, rather than statistical analysis to make predictions about future events or outcomes. While judgmental forecasting methods can provide valuable insights and complement quantitative techniques, they are also subject to biases, cognitive errors, and uncertainties. Therefore, it’s important to validate judgmental forecasts whenever possible, document the reasoning behind the forecasts, and consider multiple perspectives to improve the accuracy and reliability of the predictions.
Market Research
Market Research is the process of gathering, analyzing, and interpreting information about consumer preferences and behaviors to make forecasts. It plays a crucial role in helping businesses make informed decisions, develop effective marketing strategies, and identify opportunities for growth. It focuses on understanding the underlying motivations, attitudes, and opinions of consumers through open-ended discussions, interviews, and observations and provides in-depth insights into consumer perceptions, preferences, and behaviors, helping businesses uncover hidden needs and opportunities.
Scenario Planning
This is a strategic planning method that involves creating detailed future scenarios to explore and prepare for various contingencies. It involves creating multiple plausible scenarios or narratives about possible future outcomes based on different assumptions, driving forces, and uncertainties. Scenario planning helps organizations plan flexible strategies by exploring various alternative futures, understanding the potential implications of different scenarios, and developing strategies to navigate uncertainty and mitigate risks.
Download our free eBook on the differences between short-term replenishment and mid-long-term planning:
Quantitative Forecasting
Quantitative forecasting is widely used in various fields, including business, economics, finance, supply chain management, and meteorology. It provides organizations with valuable insights into future trends, helps them make informed decisions, and enables them to plan and allocate resources effectively. By leveraging historical data and statistical techniques, organizations can improve their forecasting accuracy and enhance their competitive advantage in dynamic and uncertain environments. Let us take a closer look at some of the methods employed in this type of forecasting.
Moving Average
Moving averages forecasting is a popular quantitative forecasting technique used to analyze time series data and make predictions about future values. It involves calculating the average of a specified number of past observations and using this average to forecast future values. Moving averages forecasting is a simple yet powerful technique for analyzing time series data and making short to medium-term predictions.
Regression Analysis
Regression analysis forecasting is a statistical technique used to model and analyze the relationship between a dependent variable and one or more independent variables. It is commonly used in forecasting to make predictions about future values of the dependent variable, based on historical data of the independent variables. It provides a flexible and powerful framework for modeling complex relationships between variables and generating forecasts to support decision-making and planning processes.
Exponential Smoothing
Exponential smoothing is a popular and widely used technique for forecasting time series data. Exponential smoothing assigns exponentially decreasing weights to past observations, giving more weight to recent data points and less weight to older ones. It provides a balance between capturing underlying patterns in the data and smoothing out noise, making it a valuable tool for forecasting in various domains, including finance, inventory management, sales forecasting, and demand planning.
Multivariable Analysis Forecasting
Multivariable analysis forecasting, also known as multivariate forecasting, is a statistical technique used to forecast the value of a dependent variable based on multiple independent variables. Unlike univariate forecasting, which involves predicting a single variable, multivariable analysis considers the influence of several variables on the forecasted outcome.
Associative Models
Associative models, also known as association rule mining or association analysis, are statistical techniques used in data mining and machine learning to identify interesting relationships or associations among variables in large datasets. These models aim to discover patterns, correlations, or co-occurrences between different items, events, or attributes without necessarily implying causality. Associative models are particularly useful for uncovering hidden insights and making data-driven decisions in various domains.
Econometrics
Econometrics is a branch of economics that applies statistical methods, mathematical models, and computational techniques to analyze economic data. It aims to empirically test economic theories, estimate economic relationships, and make predictions about economic phenomena. Econometrics plays a crucial role in both theoretical and applied economics, providing a rigorous framework for studying economic behavior and policy implications.
Trend Projection
Trend projection, also known as time series extrapolation, is a quantitative forecasting technique used to predict future values of a variable based on historical data. It assumes that past trends and patterns in the data will continue into the future, allowing analysts to make projections about future values. Trend projection is particularly useful for forecasting time series data with a clear trend component.
ARIMA (AutoRegressive Integrated Moving Average)
ARIMA, which stands for Autoregressive Integrated Moving Average, is a popular and powerful time series forecasting model that incorporates autoregressive (AR), differencing (I), and moving average (MA) components. ARIMA models are flexible and can capture a wide range of time series patterns, including trend, seasonality, and irregular fluctuations. However, they require careful parameter selection and validation to ensure accurate forecasts.
Machine Learning Models
Machine learning models are algorithms that learn patterns and relationships from data to make predictions or decisions without being explicitly programmed. There are four main steps in the machine learning forecasting process: data gathering, data pre-processing, model training, and model evaluation. These models are used across various domains, including healthcare, finance, marketing, and more, to extract insights, automate processes, and solve complex problems.
Time Series-Based Forecasting
Time series forecasting is a technique used to predict future values of a variable based on historical data collected at regular intervals over time. Time series forecasting is particularly useful when the data exhibits temporal patterns, trends, or seasonality. We will investigate the most common approaches in this type of forecasting.
Naive Approach
The naive approach is one of the simplest forecasting methods used to predict future values based solely on the last observed value in a time series. It assumes that the most recent observation will continue into the future without considering any underlying patterns or trends in the data. While this approach may seem overly simplistic, it serves as a baseline or benchmark for evaluating the performance of more sophisticated forecasting methods.
Average Approach
The average approach, also known as the mean method, is a straightforward forecasting technique that predicts future values by taking the average of past observations. It is a simple yet intuitive method commonly used for time series data when there is no clear trend or seasonality present. While the average approach may provide reasonable forecasts for stable or random time series data, it tends to perform poorly for data with trends, seasonality, or irregular patterns.
Drift Method
The drift method, also known as the naive drift method or random walk with drift, is a simple forecasting technique used to predict future values of a time series by assuming that the series will continue to drift in a particular direction over time. Unlike the naive method, which assumes constant values, the drift method incorporates a linear trend or drift component into the forecast.
Historical Forecasting
Historical forecasting involves taking sales data from a specific time period and distilling it down into quantifiable revenue trends. This means looking at your growth between certain months, quarters, and years, and assuming those percentages will carry forward as you continue to grow your profit margins. By examining historical data points, such as sales figures, stock prices, or weather patterns, analysts can identify trends, seasonality, and other patterns that may repeat over time.
Seasonal Decomposition
Seasonal decomposition is a powerful technique used in time series analysis to break down a series into its constituent parts. By separating the trend, seasonality, and noise components, analysts can gain insights into the underlying patterns and make more accurate forecasts. This method helps in understanding the repetitive patterns that occur over specific time intervals, such as daily, weekly, or yearly cycles, which can be crucial for various applications like sales forecasting, demand planning, and financial analysis.
SARIMA (Seasonal ARIMA)
SARIMA, or Seasonal Autoregressive Integrated Moving Average, is an extension of the ARIMA model that explicitly considers seasonality in time series data. ARIMA models are effective for capturing non-seasonal patterns and trends in time series data. However, many real-world datasets exhibit seasonal patterns or periodic fluctuations that ARIMA models alone may not adequately capture. SARIMA models address this limitation by incorporating seasonal components into the ARIMA framework.
Causal Forecasting
Causal forecasting encompasses the use of historical cause-and-effect relationships to predict future outcomes. It takes into consideration the causal relationships between variables and can thus provide a more accurate prediction of future outcomes. Causal forecasting can also help identify which parameters have the greatest impact on the outcome being forecasted, which can be helpful in making decisions about where to focus efforts to maximize results. It can be especially helpful in identifying the impact of new trends, innovations, or competitive pressures on business outcomes and can help businesses anticipate potential future changes in the market or environment. This gives them the opportunity to adapt before a situation changes. Businesses are able to optimize their resources by focusing on the most impactful strategies and tactics. All in all casual forecasting can lead to better decision-making and potentially better financial performance.
Causal Models
Causal models are analytical tools used to identify and leverage cause-and-effect relationships for making forecasts and understanding phenomena. Unlike purely statistical models, which might only reveal correlations, causal models aim to uncover the underlying mechanisms driving changes in variables. Let us examine some different examples of causal models. Granger Causality is a statistical hypothesis test to determine if one time series can predict another.
Vector Autoregression (VAR)
Vector Autoregression (VAR) is a powerful statistical method used to capture the linear interdependencies among multiple time series. This technique is especially useful when dealing with systems where variables influence each other simultaneously over time. VAR models can be applied in a variety of different scenarios. In the field of macroeconomics, it can be used to examine the relationship between GDP, inflation, interest rates, and employment or to assess the impact of monetary and fiscal properties.
Structural Equation Modeling (SEM)
Structural Equation Modeling (SEM) is a sophisticated statistical technique that enables researchers to model and test complex relationships among observed and latent variables. SEM integrates aspects of factor analysis and multiple regression, allowing for the examination of causal relationships in a comprehensive framework. Structural Equation Modeling (SEM) is a versatile and powerful method for understanding and testing complex relationships between observed and latent variables.
Conclusions
Forecasting methods are aimed towards the best possible prediction given the available data. It is no magic, it is science. What forecasting methods do is go beyond common sense and implement the most advanced analytical techniques to get the most out of your data. We at Intuendi are committed to deliver to you the most accurate forecast with all of the available analytical tools in order to deliver value to your business. We do not promise magic. Just accurate predictions.
To choose and validate a forecasting model on your own data, see our guide to forecasting models.
There is no single “most accurate” method. Accuracy depends on data quality, volatility, and whether the model incorporates external demand drivers.
Qualitative forecasting is most useful when historical data is limited, such as for new product launches or emerging markets. It relies on expert judgment and structured insights, but as data matures, quantitative models should replace intuition to reduce bias and protect working capital.
Traditional models rely heavily on past patterns and assume relative stability. In volatile environments with promotions, cannibalization, or channel shifts, those assumptions break down.
Forecast accuracy directly influences how much inventory you purchase and when you commit cash. Poor forecasts lead to overstocks, markdowns, or stockouts that erode margins. At La Casa de las Baterias, improved forecasting with Intuendi reduced stockouts by 25% and increased Inventory ROI by 18%, strengthening both availability and capital efficiency.
Machine learning models require more data and computational resources, but they can capture nonlinear patterns and cross-variable effects that simpler models miss. When forecasting complexity increases, predictive accuracy compounds across purchasing and allocation decisions.
