Demand Forecasting: How It Works and How It Can Help Your Business

By |2019-10-25T16:31:08-05:00October 21st, 2019|eCommerce Best Practices|

The world of e-commerce is becoming increasingly complex as more and more people begin to buy and sell goods online. The more avenues there are to sell on, the harder it becomes to track all of that data. As the channels companies can sell through continues to diversify, the ability to parse through sales data and predict trends will quickly become an invaluable resource.

Demand forecasting is a data analytics technique that analyzes historical sales data and uses algorithms to estimate future sales. Forecasting helps to inform how many units need to be ordered, preventing companies from spending money on inventory that won’t sell. It is a powerful tool for businesses that are expanding into new territories with fluctuating demand, but many companies still struggle to correctly factor demand into their replenishment process.

Automating demand planning is particularly important as it is tied to the replenishment process; 61 percent of respondents in a survey say that if their revenue doubled next month, their replenishment process would be the first point of failure.

Download Here & Now: A Guide for the Modern D2C Brand to find out how today’s brands are addressing the challenges of omnichannel ecommerce.

We’ll take a look at how it all works, and what tools and information you need to apply it to your business.

How does Demand Forecasting Work?

Demand forecasting applies algorithms to your sales data to make assumptions on supply chain and inventory activity, allowing a business to define service levels for the future accurately. Generating this data and referencing it can make all aspects of inventory planning a lot more informed.

The process can be broken down into qualitative and quantitative forecasting, which each use different resources and data sets. Quantitative forecasting involves looking at the existing data for a particular company, like financial reports, sales, revenue figures, and website analytics. A company can then apply this data using statistical modeling and trend analysis to gauge future activity.

On the other hand, qualitative forecasting focuses more on the wider economic climate, relying on estimates and expert opinions that are supported by hard data. Qualitative forecasting takes into account emerging technologies and innovations that may affect future sales, as well as pricing and availability changes, product lifecycle, product upgrades, and more. All of this information is viewed holistically to predict demand for consumer goods.

Demand forecasting can be valuable for all kinds of businesses. Projections on profit margins, expenditure, turnover, risk assessment, and other key elements of a business model can be very useful for business planning. Forecasting allows executive management to plan for multiple business functions, including future labor needs, inventory costs, advertisement campaigns, and individual regional sales projections.

The method is particularly useful in the retail and ecommerce sectors where accurate forecasting can improve inventory management. Demand forecasters can model optimal stock levels for individual products at different times of the year, as well as determine when to reorder stock, making this an excellent technique to develop and master. Particularly in the retail industry, implementing demand forecasting practices becomes easier if you have the right tools. Skubana’s order and inventory management platform includes features that support demand forecasting. Users can track in-depth inventory based on stock levels and sales velocity, allowing you to plan for the future and allocate the right resources to the right channels and warehouses.

Demand Forecasting For Consumer Goods

Within the sphere of qualitative and quantitative forecasting, there are several different methods you can use to predict demand. Choosing the right one depends on your business needs, and the first step is to evaluate each method.

1.   Collective Opinion

The collective opinion method of data forecasting leverages the knowledge and experience of a company’s sales team to aggregate data on customer demand. Members of the sales team report on the sales performance of individual products in their respective regions, allowing you to take a broader view of overall demand from granular data. If you look at sales data from an established base demand, you can compare highs, lows, and peaks, as well as seasonal trends. Inventory management platforms tend to have built-in features allowing sales executives to gather and analyze this data.

Factors like the prices of a product, related marketing campaigns, employment opportunities, customer economic data, and competitors in the area are all taken into account. It’s important to keep in mind that demand forecasting relies on both the team’s judgment and the data, so use a systematic approach to gathering and prioritizing this information.

One of the best ways to get the kind of customer data your sales team will need is through customer surveys, another valuable forecasting method.

2.   Customer Surveys

Customer surveys can provide key information on customer expectations, desires, and needs. This form of market research also takes a look at individual customer demographics and economic data to gain a better understanding of the customer base.

There are many ways you can conduct these surveys.

  • Sample survey: A select sample of potential buyers are scientifically selected and interviewed to determine their buying habits.
  • Complete enumeration survey: The largest sample of potential buyers possible is targeted for interviews to gather a more expansive data set.
  • End-use survey: This survey gathers data from companies in related industries to determine their view on end-use demand.

Sales teams can use tools like Typeform or Qualtrics to gather the data they need to analyze customer desires and behaviors.

3.   Barometric Method

The barometric method involves using economic indicators to predict trends.

  • Leading indicator: A performance indicator that might predict future events.
  • Lagging indicator: An indicator of past performance that analyzes the impact of past events.
  • Coincidental indicator: A measure of current events happening in real-time, or in a short time frame.

There are many ways we can use these indicators to our advantage.  For example, an increase in customer complaints is a leading indicator of emerging problems in production, distribution, or customer service. Meanwhile, a sharp spike in sales might be a leading indicator that business is picking up. The spike could either be an anomaly or the beginning of a trend – businesses can use lagging indicators like growth and retention to provide more context.  Finally, a good example of a coincidental indicator would be turnover, which can be monitored in real-time to demonstrates current sales activity. All of these indicators can be used to measure current, past, and future activity when it comes to your business, allowing for better inventory management and supply chain management.

4. Expert Opinion Method

You can solicit expert advice from external contractors to determine future activity. Market experts use different methods, such as the Delphi technique, which involves a series of questionnaires designed to solicit the information needed to make predictions.

Following a brainstorming session with the relevant experts, assumptions can be made that can inform your business on what to expect in the coming weeks, months, or even years. This method can be cost-effective as well as time-efficient, allowing companies to implement it quickly.

5.   Market Experiment Method

Market experiments can be carried out under controlled conditions to inform retailers on consumer behavior. A/B testing of different discounts, special offers, fonts, site features, and imagery can inform a company of what appeals to customers.

Adore Me, a lingerie company, found that by testing two similar images of the same model in a slightly different pose that one image was wildly more popular. By using the popular image, the company was able to double its sales.  Other experiments have shown that companies experience more sales when offering prices ending in odd numbers.

6.   Statistical Method

Using statistics can also be very informative when it comes to demand forecasting. Statistical methods are detailed, reliable, and often cost-effective. Some versions of statistical methods include:

  • Regression Analysis: This method allows a company to identify and analyze the relationships between different variables.
  • Trend projection: This method relies on larger data sets of historical data to establish performance history over time, identify trends, and extrapolate potential future trends.

Regression analysis variables include sales, conversions, and email signups, among others. Taking a holistic view of how each is affected by the other can help a company allocate resources to the right area and boost sales.

Trend projection is also a useful tool that allows for more efficient inventory management by predicting demand for different times of the year.

Of course, gathering your data is only half the battle. As McKinsey Digital points out, you also need to act on that data by incorporating it into your business model. You need to foster a data-driven culture within your organization. If demand forecasting is new to you, brainstorm among your staff how best to use the data you’ve gained from your research.

Staff training or hiring a consultant can pay dividends when first implementing demand forecasting, although sometimes it’s as simple as carefully strategizing among your internal staff.

How to Measure Forecast Errors in Intermittent Demand Forecasting

Intermittent demand, or sporadic demand, is seen in data where a high proportion of values are at zero.

Intermittent demand forecasting relates to predicting sales for a product with sporadic sales, and there are several errors that businesses can encounter here. It’s important to be aware of the common pitfalls and forecasting errors that companies often run into that impact accuracy. Forecasting pitfalls include:

  • Failing to account for seasonal demand
  • Using one identical forecasting calculation for products with different sales behavior
  • Failing to filter out promotional sales activity which can skew actual demand trends

It’s common to see ecommerce retailers selling camping gear at a reduced rate during the winter to compensate for an over anticipation of sales. This miscalculation results in an inventory surplus because seasonal demand wasn’t properly calculated.  You can read through a more comprehensive list of forecasting pitfalls here.

An estimated 50% of products experience intermittent demand. Sometimes the data on these products will have gaps in sales that relate to supply chain management or common sales trends. Other times, the patterns are more random and difficult to predict.

In these cases, traditional methods of demand forecasting do not always apply and can lead to errors.

Proper inventory management restocking models can prevent this. Based on a tradeoff between time, revenue, and profit, retailers use several different restocking models to meet their goals:

  • Continuous inventory system: This model constantly and automatically restocks inventory below a certain cutoff point, regardless of other factors.
  • Periodic inventory system: This system requires that a retailer periodically assess the stock levels and decide how much stock is needed based on current demand.
  • Economic Order Quantity (EOQ): This model calculates the number of SKUs a retailer should restock to minimize costs and maximize value.

Beyond the use of models, errors can be avoided by taking a granular approach to KPIs when it comes to individual products. There is no one-size-fits-all model for demand forecasting – finding the right models and methods can be a matter of trial and error.

Of course, using the right software and tools also goes a long way when it comes to getting the best performance from your demand forecasting process.

Tools and Resources For Demand Forecasting

Many tools and software programs facilitate demand forecasting of some kind. To find the right software for your business, consider your needs, specific features you want, and if you have the internal resources for demand forecasting.

1.   What You Need

Take time to carefully document your sales and demand trends in a detailed specification document.

When it comes to future performance, are there areas of your business that you feel you lack insight? What constraints in terms of budget and project lifecycle might emerge when considering the right tools for forecasting?

Consider the past and current state of your inventory management and supply chain management – what were some recent issues, and what do you need to mitigate them in the future?

2.   Features of Forecasting Software

Once you have an idea of what you need, consider the features of demand forecasting software and which ones would address these needs.

  • Complex event/hypothetical situation modeling
  • Price modeling
  • Granular forecasting metrics like econometric and cluster analysis
  • Performance measurements to gauge the value of forecasts
  • Ease of use and ease of business integration
  • Measurement of forecasting errors
  • Reliable technical support
  • Multi-tier planning for separate outlets, regions, or franchises
  • Machine learning capabilities

Choosing an ecommerce operations software is an important decision, as the software can make a massive difference in your ability to manage your inventory and optimize sales. Focus more on getting a solution with the demand forecasting features that best suit your needs.

3.   Staff Evaluation

To get the most out of your demand planning solution, make sure that your staff has the technical capabilities to operate your software. Some solutions are very attuned to seamless business integration, while others have a steep learning curve that may end up impacting productivity and consuming resources.

Consider whether you want to have a delegated team specifically for demand forecasting and, if so, what roles are required. If you’re using multiple, separate tools to meet your needs, determine if this is the right approach to take. If you stick with this approach, you may need certain staff with specific technical skill levels to operate them. Whether you choose a standalone option or a series of individual tools will depend upon your needs.

Skubana is a cloud-based order and inventory management platform for multichannel brands that provides end-to-end analytics and forecasting solutions along with demand planning capabilities. Using historical sales data, Skubana is able to calculate how many units you need to reorder within a certain time period and the exact date a purchase order should be issued. Projected growth can also be factored into these calculations on a SKU-by-SKU basis. Skubana then automatically creates purchase orders with recommended reorder quantities, streamlining the replenishment process by factoring in forecasted growth, demand, and vendor data into these calculations.

Optimize Your Inventory With Demand Forecasting

By improving demand forecasting and optimizing your supply chain, you can effectively increase profits and mitigate unnecessary costs. Take a look at your business and determine which methods best fit your industry, then select the tools that will help you accurately calculate demand.  Establishing this system improve your inventory flow and give you the resources needed to bring your ecommerce business to the next level.

Leave A Comment