Using Business Intelligence to Predict Harry Potter DVD Sales

Movie companies do about half of their new DVD sales in the first five days after launch, so picking the right week can be the difference between profits and losses. Overestimate demand and you’re stuck with inventory. Underestimate and not only do you irritate customers, but on a hot property like the latest Harry Potter film, Harry Potter and the Deathly Hallows, Part I (which is scheduled to open Nov. 19), you could forego lots of revenue.

Since 2001, when Harry, Ron Weasley and Hermione Granger debuted on the big screen, the franchise has generated an estimated $1.7 billion in gross U.S. box office sales for distributor Warner Bros. Entertainment. But worldwide sales of Harry Potter DVDs have brought in almost double that—$3 billion—according to Box Office Spy. That’s business worth getting right.

In the past decade, as Harry and his friends grew up, so did the business intelligence capability at Warner Home Video. The company has fine-tuned its sales forecasting models to ensure retailers will have enough stock to satisfy every consumer willing to plunk down $20 or so for a disc or tape, says Thomas Tileston, the company’s vice president of business decision support.

Support for Process Changes

Previously, Warner Home Video could estimate sales only to within 40 percent, Tileston says, resulting either in out-of-stocks or unwelcome overstocks. The company wanted to improve the sales and operations planning process, starting with forecasting. A key first step was standardizing data definitions so various business groups used the same statistics and terms to produce forecasts. IT is also moving some of that common data to a central Teradata warehouse, rather than having business analysts store it in their own desktop SAS analytics tools, he says.

Better data has allowed Warner Home Video to improve its manufacturing and distribution process by, for example, more accurately estimating how many discs it needs to make and how many to provide to each retailer. Business analysts also used SAS development tools to write custom scripts that analyze each movie according to criteria such as genre, cast, plot synopsis, box office trends, and price to generate a basic sales forecast.

In addition, Warner Bros. can now fold in sales data from its own similar movies and public information about how movies from competitors such as Sony and Universal have performed in the past, Tileston says.

Feeding many consistent variables into a prediction model returns more accurate results than models that chew on fewer variables, says David White, a senior research analyst at Aberdeen Group. Many types of data allow one to identify more patterns, he adds.

As Warner Home Video readies the home release of the new Harry Potter, Tileston expects its forecast to be within 10 percent of actual sales. “You would love to be dead-on from the start,” he says, “but it’s the history that helps you refine and refine.”

Follow Senior Editor Kim S. Nash on Twitter: @knash99.