Demand Forecasting: Models, Benefits, and How to Do It

Qualitative forecasting is a type of demand forecasting that uses descriptive data such as expert opinions, focus groups, competitive analysis, employee input, customer testimonials, and market research rather than numerical or quantitative sources to make demand predictions.

When to use qualitative forecasting: It’s always a good idea to get qualitative input when possible. Typically, you will use it to validate quantitative findings. If you just started a retail business or are opening a specific location or launching a new product in your market, this is the easiest way to get a quick insight into what demand might be like.

In action: At my boutique, on top of looking at historical numbers and economic trends, we would also send out surveys via our Instagram page before buying trips. Here, we would get insights from our most loyal customers about color preferences and trends they liked and didn’t, information about how they’re shopping, and what they enjoy about our brand. These qualitative insights in conjunction with our numbers helped us to make better buying decisions and better understand our customers.

The Delphi Method is a type of qualitative forecasting that gathers expert opinions and insights through a series of questionnaires and uses the responses to draw conclusions and make demand predictions.

To start, retailers send out questionnaires asking experts in their industry targeted questions that will help them to better understand their future demand and industry trends as a whole.

Depending on the answers, retailers will create a summary of the responses and send it out, along with a new set of questions based on the received information. Experts can then revise their initial responses from the previous survey and answer the new questions. This process continues until the expert responses are in consensus.

When to use the Delphi Method: This method is great if you lack internal data and are looking for expert insights. It is also ideal if you are looking for insights from people of various backgrounds and locations, as there are no bounds as to whom you can send your questions.

In action: When Bloom, a single-man flower shop, was set to open, the owner decided to use the Delphi Method to pull expert opinions from flower growers across the US. The owner needed to get a sense of when the stock was going to be available and also wanted to get insights into when these growers saw the greatest and lowest demand for flowers.

After going through several rounds of questioning, he was able to determine the seasons, specific holidays, and growing schedule of his industry and make smart demand predictions.

A trend analysis, or the trend projection method, uses your store’s historical sales and inventory data to identify demand trends. Through trend analysis, you can see cycles in demand, as well as seasonal or time-based trends that increase or decrease demand. You can then apply these findings to the future by predicting demand based on historical trends and ebbs and flows.

When to use trend analysis: Trend analysis forecasting is ideal when you have at least two years of data. You can then see multiple full-year cycles and identify where demand cycles, spikes, or trends repeat annually.

In action: For example, at my store, based on our seasonal trend analysis, we knew that there was always a spike in traffic and sales around the holidays and in the mid-summer months. We then knew to stock up on cold-weather goodies, gift options, and hot summer clothes.

Life cycle modeling is a type of retail forecasting that looks at the typical “life cycle” of a product and anticipates demand based on how frequently customers repurchase. The life cycle of each product type will be different, and what gives them their end date varies.

For example, flowers and face cream both expire; however their timelines are quite different, with flowers lasting days and creams lasting months. There is also the life cycle of something like fast fashion. While inexpensive, trendy clothing items do not technically expire, they do typically require regular replacement based on seasonal fashion trends.

When to use life cycle modeling: Life cycle modeling works for retailers that sell items that either expire, require regular maintenance or a subscription, go out of season, or need periodic replacement.

In action: For example, Small Engine Masters knows that people typically need to replace the mower blade once a year. With this knowledge, it can better anticipate how many blades to stock the store with each year.

Short-term forecasting looks at only a short period of time, between three and 12 months, to make demand predictions. This forecasting strategy allows you to focus on seasonal trends and short-term fluctuations so you can better understand and plan for them.

When to use short-term forecasting: Retailers that operate seasonal businesses with predictable fluctuations in demand can use short-term forecasting. The holiday selling season is another time when short-term forecasting comes into play. You can also use this strategy if you don’t have much data to pull from.

In action: Breeze Ski Rental knows that the majority of its traffic is seasonal and comes in September through April. Thus, it knows that demand for goods will be low from May through August, so it doesn’t make sense to use data from that period to forecast demand for the winter season. Instead, it uses short-term data to look at the winter season trends so it can understand the business’s seasonal cycles and trends.

Long-term forecasting looks at trends over a period of one to four years to predict demand. Rather than looking at things granularly, long-term forecasting looks at the big picture in order to make long-term strategic decisions.

When to use long-term forecasting: Long-term forecasting is helpful in larger business planning—for instance, if you’re thinking about expanding online or to a bigger or additional storefront. Like trend analysis, it’s also helpful for identifying cycles and patterns that repeat time and time again.

In action: Returning to my own boutique experience, during the years I was there, we expanded our brand to another location and opened a “sister store.” This store sold different, slightly more refined, and higher-end pieces, and appealed to a demographic that used to shop at my boutique but had outgrown its style. We knew about this group, their buying power, and loyalty to our brand based on a long-term analysis of consumer trends and shopper profiles.

Active demand forecasting is when you use information outside of your store’s sales data to make demand predictions. Information may include market conditions for your sector of retail, economic conditions, supply chain, and other factors that have concrete numbers and stats that you can use to make demand predictions.

When to use active demand forecasting: This demand forecasting method is ideal for retailers that are starting out and don’t have their own sales data to reference. It is also great for those looking to expand their reach or enter a new industry, but lack internal stats. Many retailers, however, will consider active demand forecasting factors alongside their store’s internal sales and inventory data.

In action: About halfway through my time at my former boutique, the business that I worked for decided to open a new concept store. Whereas its first several stores, Inspyre Boutique, sold clothes aimed at young women at a sub $100 price point, the new concept, April and West, was aimed at an older audience with a bigger budget.

With no internal data on this new market, the company buyers went to work looking at economic factors, industry insights, and local data from the new store’s location. With all these active demand forecasting data points, they decided to go with a much smaller inventory than was used at the other stores and place a greater focus on jeans and accessories.

The barometric technique is a forecasting method that combines three indicators to make predictions about demand over time. These generally include leading, coincidental, and lagging indicators.

• Leading indicators: This is the event that causes demand to go up or down.
• Coincidental indicators: These are the events that are true at the same time as the leading event.
• Lagging indicators: These are events that happen after the leading event that impact future demand.

When to use the barometric technique: This is a great option for retailers that lack sales data from years past. It is also helpful for businesses that see large seasonal fluctuations, as these would be accounted for as lagging indicators.

In action: A local bakery, Pastries by Emma, has put up a stand at the farmers market near the owner’s store. Her leading indicator is that she saw a 30% increase in cupcake sales at her last trip to the market. Coincidentally, this caused Pastries by Emma to run out of lots of flavors, resulting in a 5% loss in sales to stockouts. Additionally, staff had to stay overtime, boosting operating costs by 10%.

If Pastries by Emma were to only account for these factors, they would likely make upwards of 30% more cupcakes for the next market. Emma, however, checked the radar and saw that there was going to be rain the following week, so with this lagging indicator in mind, she predicted a slight decrease in demand.

Want to automate your inventory management and get insights into your demand forecasting with built-in analysis and prediction features? Check out our suggestions for the best inventory management systems.

Inventory Management: The process of having the right products, in ideal quantities, at the right time to sell to customers.

You can choose one demand forecast model or several – whatever will help you create the most complete picture and make accurate forecasts.

There are eight main demand forecasting methods, also known as models, and these include qualitative forecasting, the delphi method, trend analysis, the lifecycle model, short and long term forecasting, active forecasting, and the barometric technique.

Each method looks at specific variables and makes predictions based on those factors. To get a complete and accurate picture of your demand, you should use several models.