Lionsgate has connected SAP BusinessObjects to SAP Business Information Warehouse to analyze how movies perform in theaters, what viewers liked about the movies they saw and why they chose to see a movie instead of seeking another form of entertainment. "By using this data, we can fashion the marketing for home entertainment and other distribution channels. We make sure we get the right product to the right consumers," Collins says.
The next step is to extend BI capabilities to mobile users. Collins is eagerly awaiting the day when SAP will work on tablets and smartphones, so employees can have real-time access to data even if they're using mobile devices.
Nihad Aytaman, director of business applications at fashion designer and retailer Elie Tahari, says a bigger challenge for him than mobile access is how to handle the volume of data, which is sure to grow exponentially as BI scales. Aytaman, an IBM Cognos user, points to load methodologies and load windows as performance-related areas of concern.
Currently, the New York-based company has opted not to use cubes -- data structures used to manipulate and analyze data from myriad perspectives -- and instead directly uses real-time data to generate the more than 20,000 reports requested each month.
Nihad explains that cubes allow the BI presentation tools to slice and dice the data along various dimensions like date, region, account and so forth. But cubes take time to build, and therefore are usually created overnight when more resources are available. The data in them can't be updated during the day; the only way to refresh a cube is to rebuild it. Therefore the data becomes stale as the day progresses.
Nihad says that by using a database instead of cubes, "We are able to keep the process time for reports at acceptable levels -- within one to five minutes -- by fine-tuning our data warehouse database and loading only the data that is necessary."
But as data volumes continue to grow, and as reports require more time to run, there may come a point at which Aytaman will need to rethink that strategy and contemplate using cubes.
To account for this increased querying and reporting, Aytaman and his team have built a generic operational data store (ODS) to support the company's 800 employees and 22 stores around the world. The ODS serves as a buffer between the transactional systems and the data warehouse, protecting the warehouse from having to be changed in the event that any of the transactional systems have to be replaced or new ones have to added as a result of acquisitions.
The company recently added the ability for its department-store users to analyze selling trends and ensure that the right inventory, including styles and sizes, is on the floor. Predictions are made based on past sales patterns. All this information is fed into IBM's TM1 planning software, which calculates the company's needed weekly inventory levels.
In addition, corporate executives, warehouse managers and others regularly conduct queries to ensure that any activity that impacts revenue -- from shipping to the design of store displays -- is handled in the best way possible. If store managers detect a shift in styles purchased, for example, they can make sure the most popular items are displayed prominently.
The investments of time and money needed to support all of that have paid off: Aytaman says that the use of BI has resulted in a 10% to 20% increase in retail sales.