Data warehouses and holistic business intelligence
Business intelligence is the process of analyzing company data to better understand that data, spot anomalies or trends and make predictions.
It’s possible to do analytics on individual systems — such as the sales system or the inventory system — but the resulting business intelligence is then significantly limited in value.
When the analytics are performed across many — or all — of a company’s systems, the resulting intelligence is more comprehensive. Executives can see a genuine, holistic, bird’s-eye view of company performance. They can spot correlations and relationships between sales and inventory levels to help make informed decisions, for example.
To get that holistic view, data warehouses are used to pull all the data together.
“Corporate operations like finance, labor and sales are presented through business intelligence dashboards using data warehouses,” said Chida Sadayappan, lead specialist for data cloud and machine learning at Deloitte Consulting.
In recent years, these platforms have been using machine learning to drive deep insights and predictions, he said.
Analytics platforms are available that work with traditional on-premises data warehouses. As enterprises shift to cloud-based warehouses, the analytics capabilities are increasingly built in and often include the latest AI and machine learning functionality.
“The data warehouse is the most powerful source of data to drive business intelligence and strategy,” said Avneet Dugal, vice president of global insights and data at Capgemini.
Choosing the right platform depends on the business use case and companies don’t necessarily need a single warehouse to handle everything. For example, when a company needs to do real-time analytics.
“If the data first has to go to this warehouse, and 48 hours later you can get to it — you need to change your mindset,” he said.
There are new technologies that make analytics faster and cheaper. Dugal said it can help see what’s happening in the market in a more responsible fashion.
There are also analytics applications that don’t need an up-to-minute stream of the latest transactional data, but instead need long-term historic data.
“A lot of business use cases are about consumer behavior trends,” he said. “A coffee company, for example, could look at how coffee sales are doing at certain times of the year. Or you can look at economic cycles, to see which products react favorably to up markets and down markets.”
Segmentation, where customers are put into similar groups, is another application of business analytics that benefits from large amounts of historic data.
“You need to listen to what folks are asking for, and not put everything in one giant system,” he said.
