Demand Forecasting: Systems, Methods & Tools to Calculate Your Exact Inventory Needs

Fail to forecast accurately, and you're failing your customers. Underestimate demand and you'll be inundated with stockouts and overselling. Overestimate and you've just wasted cash on unnecessary inventory. But how do you carry out demand forecasting without simply... guessing? Answer: Sales data combined with models and systems. And that's exactly what this post is about. We cover everything on demand forecasting in retail - including all the different types and methods, then how to choose one that works for your business.

SEE ALSO: Inventory Management: A Complete Guide for Retailers

Demand Forecasting Cheat Sheet

Want to save a hard copy of this post for later? We put it into a handy demand forecasting cheat sheet PDF to print, read offline or share with co-workers.

Download Cheat Sheet

What is demand forecasting?

Demand forecasting is the systematic method to assess future demand for a particular product. Simply put, it allows you to scientifically estimate sales over upcoming weeks, months and years - so you know exactly how much stock to order and hold at any given time. There are plenty of different options for how to do this. These range from manual calculations, to automatic inventory forecasting systems, to simply opting to rely on 'just in time' re-ordering. But it's generally a combination of using past sales data along with individual market knowledge.

Value of demand forecasting

It might be tempting to ignore forecasting altogether. After all, making sales is a lot more exciting than analysing data. But it's also a lot easier to make those sales with an extra $10,000 to spend on marketing, instead of wasting it on inventory you didn't really need. So here's a few reasons quality demand forecasting can be so valuable to your business: When executed properly, demand forecasting changes the entire landscape of business operations. No longer does a company run blindly, but rather with the tools needed to perform the tasks at hand. Having said all this: It must be noted that there's no one-size-fits-all solution when it comes to demand forecasting. There are several different methods and things to consider before implementing a forecasting strategy. Let's take a look at them.

Research: Quantitative vs qualitative

Conducting adequate research is essential before making any kind of demand forecasting effort. There are two different approaches or styles to this: qualitative and quantitative.

1) Quantitative forecasting research

Quantitative forecasting is all about hard data. It's almost solely numbers and facts that rule the roost. You'll rely on previous sales history as well as knowing your typical peaks and downturns at different times of the year. Here are some key techniques to think about when it comes to forecasting from your data:

• Moving average. This is simply looking at your sales data as running averages over a select time period. E.g. a three-month moving average for the year might plot data from Jan-Mar, then Feb-Apr, then Mar-May - continually updating as new data becomes available. More here.
• Trend analysis. This is exactly what it sounds like - looking at data over given periods of time to spot underlying trends that may impact sales in the near future. More here.
• Exponential smoothing. This prioritises the latest data over a given time period by assigning exponentially decreasing weights for newest to oldest figures. It's useful for making short-term decisions where the latest data can be more relevant. More here.
• Decomposition of time series. This is breaking up a time series of data in order to look at it in greater detail. So rather than looking at an entire year, you'd look at monthly or quarterly fluctuations within that year. More here.

Of course: To make use of any of this, you need to actually have sales data available to you. So a quantitative approach can be highly accurate. But becomes a lot more accessible for businesses that have been around a while - with newer ones maybe not having a big enough data set to draw reliable conclusions from.

2) Qualitative forecasting research

With a qualitative approach, you'll focus on describing a likely outcome rather than measuring it. People's views, opinions and impressions are therefore critical. The method features less structure, and is more about implementing instincts and experience. But that doesn't make the process about guesswork. Instead, businesses use professional experience to interpret their data. There's a heavy emphasis on what drive’s people, their attitudes and thoughts. To get qualitative information for the projection of upcoming sales, you might choose to look at:

• New ad campaign impact.
• Effect of the latest technologies.
• Recent market trends or fads.

You might also consider market/customer research via things like focus groups or surveys. Your customers would discuss and react to new product features you show them.

Combining the two

This doesn't mean you need to pick one or the other types of research. If you have data available, this should always be used in some way. But what if you're launching a brand new product? Or running a big paid advertising campaign unlike anything you've done before? Or an economic recession just hit? It could be that your sales will take an otherwise unexpected uptrend or downtrend. This is where your qualitative research of customer behaviours and market trends can add more depth. Helping to interpret your past sales data in a more accurate way.

Demand forecasting methods

Once you've got your quantitative and/or qualitative research in place, it's time to actually start forecasting demand. There are two things to consider at this stage:

1. Forecasting period. How long are you going to forecast for? Each month, every quarter, annually. Forecasting will then become a repeated task at the end of each period.
2. Level of detailing. How detailed do you need to be in your forecasting? Do you just need a ball park figure, or almost nailed on numbers?

With this in mind, here are the primary methods of forecasting in relation to retail and ecommerce:

1) Passive demand forecasting

In this model, the future is based on the assumption that nothing changes. A company knows that it doesn’t plan any significant changes to its course - like big marketing campaigns, introducing new products, adding sales channels. And so historical data is simply replicated forward continually. Key aspects:

• Ideal for stable businesses.
• Contain conservative growth plans.
• Historical data is used to estimate demand.
• Minimal assumptions made on up or downturns.
• Rarely used – mainly for small or local businesses.

2) Time-series analysis

This is where you use past sales data to draw conclusions about trends and seasonality. It means going back over the past year (or several years if the data is available) to pinpoint where demand seems to rise and fall for different products. Key aspects:

• Ideal for medium-sized businesses with at least a few years' worth of data.
• Helps identify seasonal fluctuations and sales trends.
• Useful for businesses with seasonal products or run sales at specific times of the year.

3) Causal forecasting

The causal model is one of the most advanced options for forecasting demand. And takes into account a whole host of things - both qualitative and quantitative. It uses past sales data, as well as information about competitors, planned marketing activity and external economic forces to predict likely uptrends and downtrends. New marketing campaign going live in three months? Sales will likely go up. New competitor? Sales could reduce. Recession about to hit? Sales will likely go down. Key aspects:

• Ideal for large, scaling or diversifying businesses.
• Requires time and dedicated resource(s) to conduct.
• Utilises both qualitative and quantitative research.
• Considers competitor activity.
• Evaluates economic environment.

Applying this to retail

It's likely that you'll set out demand forecasts for an entire year when it comes to setting sales targets. But inventory planning for retailers should take a slightly different tact. You'll need to strike a balance between:

1. Minimising cost-per-unit when bulk buying from suppliers.
2. How confident you are that the product will sell.
3. Ordering only what you can physically store in your warehouse.

In other words, you don't want to order too little or too much stock at any one time. It's therefore likely that regular time-series analyses will be your best bet for short-term inventory forecasting as a retailer. You'll need to be answering questions like:

• How well did this product sell in the last month, quarter, etc.?
• How well did this product sell over the upcoming period last year?
• Are there any differences in promotional activity compared to the same period last year that could cause a sales upturn or downturn?

Once you have this information, you can make an educated estimate as to how much stock you should be ordering to cover the upcoming time period.

Automated demand & inventory forecasting

There's obviously a lot that goes into accurate and diligent demand forecasting. And doing it all manually is simply not practical for most modern high-growth retail businesses. Especially with how quickly customer expectations and market trends change these days. This is when software becomes invaluable. Veeqo, for example, has a built-in inventory forecast component. This takes the guesswork out of demand forecasting by using past sales history to calculate exactly how much inventory you need:

You can easily work out how much stock is required to cover a variety of upcoming periods. And even offset the figure with an expected sales up or downlift. All putting an end to stock-outs and over-purchasing to keep inventory levels perfectly balanced based on cold, hard data. To see it in action first-hand (as well as all Veeqo's other powerful features), just book a demo with one of our product specialists.

Final thoughts

Demand forecasting is an essential part of managing a growing retail business. You simply need to have some degree of insight into how much you'll sell. And therefore, how much inventory you need to cover those sales. But actually doing this is undoubtedly a complex task. Ultimately, it's up to you to decide on which method to go down: one of the various manual options, or an automated solution.

Demand Forecasting Cheat Sheet

Want to save a hard copy of this post for later? We put it into a handy demand forecasting cheat sheet PDF to print, read offline or share with co-workers.

Download Cheat Sheet