Corporate managers will never be able to make good, informed decisions based solely on data. But — if formatted properly to provide knowledge and insight — data can facilitate complex business decision-making.
What you might need, says Blake Sellers, president and CEO of Avvantica Consulting, LLC, is business-intelligence-for-decision-support (BI-for-DS) capability, which will allow easier access and analysis of existing business data.
“When managed properly, a business intelligence implementation can deliver continuous benefits through a series of relatively short projects,” Sellers says. “With this approach, the company is better able to manage the risk of an individual project while working to obtain real business benefits in a relatively short period of time.”
Smart Business asked Sellers to further describe the concept of BI for DS.
What is business intelligence (BI)?
It goes by many names, depending on how it’s used. Some of the names are data warehousing, data marts, executive information systems (EIS), executive dashboards, multi-dimensional analysis and reporting, corporate performance management (CPM), business performance management (BPM), decision support systems, and online analytical processing (OLAP).
How does business intelligence differ from traditional information systems reporting?
Most business applications are designed to capture and/or manage transactions. Examples might include entering a customer order, receiving material against a purchase order, or recording a journal entry. Some level of reporting is usually available from most transaction systems, but they generally are not that useful for decision-making — especially strategic decision-making.
On the other hand, business intelligence develops and displays information that can be used to make better management decisions. Two things are unique about business intelligence. One is that you are bringing multiple ‘views’ of your data together in the same place. The other is that the database itself is designed for query and analysis, which is difficult to do in typical transaction systems.
Why are standard reporting systems insufficient?
First, transaction management systems and their associated databases are typically designed and then optimized to support many simultaneous users updating the database. A key design objective of standard reporting/transaction management systems is sub-second response time for a large number of users. Unfortunately, databases designed with that objective are not optimal for analysis. In some cases, attempting to use a transaction system for substantial analysis and reporting can bring the system to its knees. So you need a database structure that’s designed for analysis and reporting.
Second, typical business applications only have access to a limited set of data files. With a business intelligence application, you can combine information from multiple data sources, which allows for multi-dimensional analysis ‘at the speed of thought.’
Can you provide an example?
Say a company needs improved sales reporting. Key data might come from five separate systems: an order management system, a payroll system, a customer relationship management (CRM) system, a planning and budgeting system, and the general ledger.
For this particular analysis and reporting need, management wants to understand things like sales year-to-date, sales versus prior years, sales versus goals, sales trending, commission calculations, what products are selling, forecast accuracy, and much more. A BI-for-DS system represents an effective approach for bringing all of this information together.
What are the key components of a business intelligence application?
Start with various source databases. Extract the source data from the original systems, transform it, and load it into the BI databases. Part of this extract/transform/load (or ETL) process might be aggregating daily transaction data into weekly or monthly totals, or mapping unique codes from separate systems that actually mean the same thing.
The results of the ETLs are first stored in a relational database, because certain reporting may not require multi-dimensional access.
Next, create the multi-dimensional database which is often referred to as a ‘cube.’ This is typically the key element of a business intelligence application.
Finally, design a user interface to extract information from the cube that typically includes reports, graphs, gauges, tables, query/analysis and so on.
What’s the best way for a company to get started?
Start by identifying and building a specific prototype or ‘proof-of-concept’ application. This will allow the organization to get started with minimal risk, and will also help to build momentum and support for the overall concept of BI for DS.
In parallel, we recommend that companies develop a high-level strategy for business intelligence. For a medium-sized company this can generally be done fairly quickly; six to eight weeks is typical. The strategy helps a company to define an overall technical architecture for BI, identify its priorities, determine the level of resources required, and estimate the overall time frames that are likely to be involved.
BLAKE SELLERS is president and CEO of Avvantica Consulting, LLC. Reach him at (214) 379-7920 or [email protected].
