Demand Forecasting: Look for the Next Big Wave of Innovation

The art and science of demand forecasting have both come a long way in the past 20 years. Yet as much as demand forecasting has improved, we believe there’s still plenty of opportunity to take it much further. We’re especially enthusiastic about the business results that future improvements will bring.

Today we share some of the remaining challenges of demand forecasting and the reasons for our optimism about overcoming them soon.

Although the topic of demand forecasting can be as dull as a rock, we trust you’ll share our interest when you consider how much demand forecasting can influence the amount of money your company makes.

Why care about demand forecasting?

When you forecast demand for finished goods, the financial consequences of inaccuracy can be huge. Forecast errors cause you to run out of stock. The result is missed sales and unhappy customers.Forecast errors can also cause you to carry too much inventory. You may have to borrow money to pay for the excess. You may also have to mark inventory down, donate it or even junk it just to get rid of it.

Forecast error reduces profit. It can also reduce cash flow and increase the need for capital. Excess inventories yield a lower return on assets.

Conversely, more accurate demand forecasts can reverse all of these problems. They can improve in-stock performance, increase revenue, improve customer satisfaction, reduce inventory investment, improve cash flow and improve return on assets.

What is demand?

Demand is the number of units your customers will buy if you have the product in stock. It’s often the same as sales, but not always. If your customers want to buy 23 units but you have only 20 in stock, your sales will be 20 units, but your demand is 23.

Some companies have a hard time forecasting demand simply because they can’t distinguish between sales and demand. In other words, they don’t know how much they could have sold if they’d had an item in stock.

(Fortunately, this is fairly easy to fix. Many demand-forecasting and inventory-planning systems have long since reduced or eliminated the problems that arise from the difference between demand and sales.)

Forecast accuracy is a measure of the difference between your demand forecast and your actual demand, usually expressed as a percentage.

What are the challenges of demand forecasting?

When demand forecasting works well, it can work really well. Modern demand-forecasting systems can deliver forecast accuracy of 99.99%. But such high levels of forecast accuracy are stubbornly elusive for many kinds of items.

Here are four primary causes of forecast error:

• Statistical forecasting methods have inherent limitations because they use the history of demand to project forward. If history teaches us one thing, it’s that the future is often different from the past.
• Statistical forecasting systems don’t understand the contextual or situational factors that influenced demand in the past or that will influence it in the future.
• Demand for some kinds of items in some situations is inherently harder to forecast than for others.
• Human judgment and memory are unreliable.

What are the best hopes for improved forecast accuracy?

Despite these challenges, we’re optimistic about the future of forecasting because of the convergence of these important trends:

• New technologies have made it cost-effective for organizations to collect and manage huge amounts of data. Data storage is cheap. Even small computers have enormously more processing power than they did 10 years ago. The Cloud provides virtually limitless capacity for storing and processing data inexpensively.
• You can easily tap into many sources of detailed contextual data, including information about weather, insights from social media, consumer interactions with websites and other digital media, POS market-basket data, information from customer-loyalty programs, and so on.
• The growth of software as a service (SaaS) or Cloud-based demand-forecasting systems is changing the way you can generate and manage forecasts.
• The fields of neuroscience and behavioral economics have taught us a great deal about how the human mind works. We know a lot more about decision making than we did just 10 years ago. The new insights can improve the way we design forecasting systems.

Collectively, these trends mean you’re likely to continue seeing big improvements in forecast accuracy. And we think the improvements are especially likely for items that are still hard to forecast. Stay tuned here for details.

Do you see any other signs that demand forecasting is likely to improve soon? If so, what are your hopes and expectations?