When forecasting sales to determine how much inventory is needed to meet demand in your eCommerce store, it can be challenging to decide which forecasting method to use—seasonal, year-round, top-down, or bottom-up. With so many forecasting options available, the key factors to consider are your available sales history, the seasonality of your products, and the relevance of recent trends. These elements help guide the most suitable approach for your business.
The methods discussed in this article, such as year-round and seasonal forecasting, are examples of specific demand forecasting models that align with broader forecasting techniques. Demand forecasting, a subcategory of predictive forecasting, focuses specifically on estimating future customer demand to ensure inventory levels meet expected sales. By applying the right combination of these models, businesses can gain accurate insights and optimize their supply chain efficiency.
What Are the Main Forecast Models?
When managing inventory for an e-commerce business, using the right forecast model can make the difference between overstock and stockouts. Forecast models help predict future demand by analysing sales data, trends, and seasonal patterns.
There are many demand forecasting methods, each with a unique use case. The most common include:
- Year-round forecasting for stable, non-seasonal products
- Seasonal forecasting for items with predictable demand spikes
- Bottom-up models focused on SKU-level trends.
- Top-down models based on category-level projections
These models can be used alone or in combination to meet specific business needs. Let’s break them down with examples and use cases.
Year-Round Forecasting Model
Best for products with consistent, steady sales all year.
Year-round forecasting uses sales velocity—the rate at which a product sells when it’s in stock—to estimate future demand. It’s ideal for evergreen products that don’t fluctuate due to seasonality or trends.
Example: Alarm clocks or basic phone chargers sell steadily, making them perfect candidates for year-round forecast models.
Seasonal Demand Forecasting Model
Ideal for inventory that peaks during specific times annually.
This model relies on historical seasonal sales data to anticipate demand spikes tied to holidays or annual events. It compares past performance from the same period over multiple years to project future inventory needs.
Example: Back-to-school items like backpacks or planners typically peak in August and September. Seasonal models will forecast higher demand during those months.
Bottom-Up Forecasting Method
Use for variant-level forecasts based solely on individual sales history.
The bottom-up method starts at the SKU level, looking only at the sales history for a particular item (e.g., size nine silver sandals). It doesn’t rely on broader category trends, making it ideal for accurate variant-level forecasting.
Example: If you sell multiple shoe sizes, this model will generate a forecast for each size and color combination, ignoring sales from related variants.
Top-Down Forecasting Method
Best for seasonal items with limited history or when forecasting by category.
Top-down forecasting takes a broader view by looking at category-level trends, then breaking them down to estimate individual variant demand. It’s ideal when you lack enough sales data for a new product but understand its category’s performance.
Example: A new style of backpack introduced in May can be forecast based on how well backpacks typically perform in August and September.
Combining Forecasting Methods
Commonly, these forecasting methods are combined in particular ways that work for the majority of merchants:
Bottom-up and Year-round
Bottom-up forecasting is often the default method to make forecasts. Bottom-up forecasting can be good to use for non-seasonal products because even just a couple months’ worth of sales history can really inform your forecast.
Top-down and Seasonal Forecasting
Use top-down forecasting for seasonal items with a short sales history. You know that categories such as backpacks peak in August and September. If you introduce a particular backpack style in May, you wouldn’t know how many to forecast to have ready in August and September. However, if you sold a few in May and June and it is trending overall within the backpack category, then you can examine that category trendline. Sales will ramp up in August and September, so you need to have a lot more on hand. You know how that backpack will compare to the category overall because you know how it’s performing within the category.
Automating Forecast Models with AI-Powered Tools
Let Inventory Planner work out your optimal order quantity
Choosing the Right Forecast Model
Choosing the right forecasting model is key to accurately predicting inventory needs for your eCommerce business. By understanding the types of demand forecasting methods—such as year-round, seasonal, bottom-up, and top-down—you can select the most suitable approach for your products. The procedures outlined in this article help you fine-tune your inventory management strategies and ensure you meet customer demand without overstocking or under-stocking.
Each model has strengths, depending on your product type, sales history, and market conditions. Combining these methods or adapting them to your specific needs can improve your forecast accuracy and optimize inventory levels.
As you implement the appropriate forecasting models, you’ll be better equipped to streamline your operations, reduce costs, and stay ahead in a competitive market. With the right tools and insights, you can ensure your inventory is aligned with demand, helping your business thrive.
Frequently asked questions:
1. What are the four types of forecasting models?
– The four main types of forecasting models are:
- Qualitative Forecasting: Based on subjective judgments and expert opinions, used when there is little historical data (e.g., market research, Delphi method).
- Time Series Forecasting: Uses historical data to predict future demand based on patterns such as trends, seasonal variations, and cycles.
- Causal Forecasting: Assumes that demand is influenced by one or more independent variables, such as advertising or economic factors.
- Machine Learning Forecasting involves advanced algorithms and data models that predict future demand based on large datasets and complex patterns.
2. How does bottom-up forecasting differ from top-down forecasting?
– Bottom-up forecasting focuses on forecasting individual product variants (e.g., specific sizes, colors, or SKUs) based on each product’s historical sales data. It is highly granular and specific to each product or variant.
Top-down forecasting, on the other hand, starts with high-level data (e.g., product category or company-wide sales) and allocates the forecast to individual products based on their share of the overall category sales.
3. What is the best forecasting model for seasonal products?
– Seasonal demand forecasting is the best method for products that experience fluctuations at specific times of the year, such as holiday decorations or winter clothing. It relies on analyzing sales data from previous years for the same periods and applying it to forecast future demand, while factoring in any growth trends or changes in consumer behavior over the years.