Share on facebook
Share on twitter
Share on linkedin
These five steps will help you increase accuracy of your demand forecasting process, resulting in less inventory waste.

How to Improve Your Demand Forecasting Accuracy

Even small improvements to your demand forecasting process can have material impact to your business. Based on my experience, I identify some of the most common mistakes with forecasting demand, and provide the following tips to improve forecasting at your business.

Step #1: Well-Defined Sales Pipeline

Improve Demand forecasting accuracy with sales pipelines The sales pipeline is a critical input into your demand forecasting, as it informs the business of new customer wins. To be an effective input for forecasting future demand, it is important that all members of the sales team document their pipeline in a consistent manner. For instance, it’s not uncommon for some sales associates to classify their sales prospects at a later stage versus the corporate policy. If this occurs, the demand forecasting team will predict higher product demand than is appropriate. 

As example, I once worked with a business in which the sales team routinely classified prospects as ‘wins’, when in fact they had not yet won the account. While this action started with just a couple sales leaders, it became widespread practice among the remaining sales team, who felt compelled to reach the level of their peers. As a result of this practice, the business’s demand forecasting process nearly always overstated customer demand. By the time the management became aware of this practice, their supply chain had already built an enormous amount of excess inventory in anticipation of future demand. The inventory eventually aged, and the business was forced to write off and dispose of the inventory as it expired. In short, this made inventory management especially difficult. 

Because a reliable sales pipeline is so important, you may choose to incentivize the sales team based on whether their pipeline reporting proves to be accurate. In my example above, the sales leaders were encouraged to report customer wins, as it affected their status on a leader board. However, this resulted in over-reporting of new wins and generated an inaccurate forecast of future demand. You should instead motivate the sales team to help you accurately forecast demand for your products, even making it part of the bonus structure. 

The pipeline should reflect both the likelihood of customer wins, as well as the expected timing based on stage of each prospect. I encourage businesses to adopt consistent practices for quantifying the pipeline. You can then use risk-adjusted probabilities when you begin to forecast demand resulting from new wins. Sales prospects in the early stage should have very little or zero impact in the demand forecast. Once the prospect has engaged with active contract negotiation, you may begin incorporate the likely outcome in your demand forecasting process.  

How this would work in practice: You have a sales prospect has indicated an interest in purchasing 1,000 quantity of a product per month, and have just begun negotiating pricing and payment terms with your sales team. You typically realize 60% success rate for customers at this stage of the sales cycle. For this reason, you could forecast demand for 600 units for the next month. Once the prospect has made a purchase order, you then include the full 1,000 units in your demand forecast. 

Step #2: Ask for Customer Forecasts

sales forecasts from large customers aid in demand planningYour customers have the best information on the growth prospects for their own businesses. While you can always perform a time series analysis of historical order patterns, you will still be guessing on the likely growth of these customers. If your largest customers are willing to share their forecast of expected purchases, you can incorporate this knowledge into your demand forecasting. 

Customers are frequently reluctant to provide a forecast of their upcoming demand, as they don’t want to be held accountable when actual demand is different from their forecast.. No one wants to give estimates if they aren’t required. As a consumer, have you called a contractor and asked for an estimate for a major house remodeling project? They generally will decline, and tell you they need much more details before even thinking of giving you a quote. The same holds true for your customers – They will hesitate to give you an estimate of product demand before they actually need the product. 

So, how do you encourage your customers to begin forecasting demand for their own businesses? 

First, you could offer a price discount to customers who provide you with forecasts of future purchases. You can likely afford to give some price concession for this, as more accurate demand forecasting results in lower inventory costs for your business. You likely have the ability to sell at a slightly lower price, while still saving money on warehouse costs. Everybody wins!

Alternatively, you could offer a guaranteed service level (or fill rate) to each consumer who is willing to forecast their product demand. For instance, you may have a typical fill rate of 94%. For those customers that assist you with forecasting demand, you may agree to guarantee a fill rate of 96%. With better forecasting, you should naturally have higher fill rates anyhow. Thus, this promise likely costs you very little.

Step #3: Add Statistics to Your Forecasting

Use Statistics in your demand forecastingOnly 20% of demand planners use statistics in their demand forecasting. In my experience, statistics is a little intimidating for the typical person. Your demand planner would most certainly be more effective at demand forecasting if they understood key statistical principles. 

This is one of those cases where it is absolutely worth it to invest in education for your key employees. Find an online course, or even make plans for your employees to attend a course at the local university. Whatever route you take, it is important that the demand planner understand basics of statistics prior to deploying statistics-based software in your business. They should be comfortable with performing regression analysis, which is essentially a tool to calculate correlation between different sets of historical data. 

With this knowledge established, you can now employ basic statistical methods in your demand forecasting.  As an example, if your business makes plumbing supplies, you may find that your business sales are highly correlated to the level of new construction starts. Or, if you make components for the automobile industry, you may find that your business growth is highly correlated with consumer purchases of new vehicles. 

There is nearly always an industry or economic indicator that can be used as a predictor for your sales. You just have to find it, and develop methods for applying it to your own demand forecasting.

By incorporating industry and economic indicators, it is quite possible you can increase the accuracy of your demand forecasting by 10% or more. This action alone can dramatically decrease the amount of stock and make your existing inventory management practices even mor effective. 

You can further refine your statistical analysis by implementing a software package designed specifically for demand forecasting. Software packages aren’t able to identify economic/industry predictors as described above, since that requires real humans that know your business. It requires human judgement to make decisions as to which industry data is most relevant. However, these packages do excel at performing time series analysis with historical data. Specifically, software is ideal at quantifying seasonal factors and order trends, both of which can further improve your forecasting.

Step #4: Collaborate With Other Functions

Collaborate with other functionsFrequently, demand forecasting is viewed as a ‘supply chain activity’, and is managed in isolation. This perception prevents other functions from contributing valuable input into this critical forecasting process. 

The marketing leaders typically have the best knowledge about long-term trends in the marketplace. Further, they are aware of upcoming product innovations, and can provide insight into upcoming changes in the company’s product portfolio. 

Sales leadership should also be included – Remember, the sales pipeline is a key component of demand forecasting, and you should ensure sales leaders understand how you are including the pipeline data in both short-term and long-range forecasts.

Step #5: Eliminate Unprofitable SKUs

Improve demand forecasting by removing unprofitable productsVirtually all businesses have an excessive number of product variants, many of which actually generate losses to the business. On average, about 20% of a company’s product portfolio generates three quarters of its sales. These are the high-volume products that are the mainstays of the portfolio, with highly predictable demand.

Stepping down a level, the next 30-40% of products are still profitable, but generate lower volumes with slightly higher volatility. The bottom 20-30% of the product portfolio generates little sales. This is often referred to as the ‘product tail’ and demand for these products is highly volatile. This bottom level of products frequently lose money, as it’s hard to predict demand for these products.

Businesses often don’t realize that these are actually money losers, because they may in fact generate sales and gross profit from these products. However, when you factor in the real cost of the product tail, such as warehouse cost and inventory scrap costs, they inevitably lose money. 

By eliminating unprofitable product variants, you may find that instead begin purchasing your most popular products instead – They may have bought a specific product variant simply because they were unaware of other options.  Even if they don’t, you are still better off by eliminating products that cost you profit. 

Let’s be honest – It is hard to make decisions to eliminate products. It is virtually always easier to make investment decisions that result in increased capacity or added breadth to your product line. It almost seems like a failure of the business when you take steps to eliminate any product. This is largely why businesses often fail to make the necessary decisions to prune the product line. 

Summary: Demand Forecasting is a Journey

Remember, demand forecasting is equal parts art and science: You will never reach 100% accuracy; by nature forecasting is never perfect. However, a good goal is to continuously deploy and refine methods that yield the most accurate forecasting of demand. Like any other process, you won’t always see results in the short-term. However, if you make plans for steady forecast improvement over time, you will gradually see the benefit through lower inventory and higher customer fill rate. 

Bryce B.

Bryce B.

I will help you build an effective demand planning process, resulting in lower inventory waste and higher order fulfillment.  
Learn More About Me >>

About Me

I possess nearly 20 years of experience in senior finance roles. I can help you improve your demand planning process, resulting in less inventory waste. 

My Story >>

My Services

Rapid Response Analysis

Do you have warehouses full of inventory?
Are you constantly missing out on orders due to lack of product?

With this quick-turn engagement, I will rapidly identify the root cause of planning issues and provide expert guidance on resolution.
Learn More >>

Assist in simplifying demand planning process
Deploy Planning Process

I design and deploy a simple and repeatable planning cycle that ensures every stakeholder has a voice:

  • I develop a repeatable process by which all your functions collaborate on the forecast. 
  • I build a prediction model based on your historical sales trends. 
  • I provide training and hands-on guidance throughout this process

Learn More >>

Recent Articles