Before operational business intelligence (BI) came on the scene, classifying BI applications was simple: enterprise applications and BI/DW activities were clearly separate domains with their own IT staffs, budgets and expertise. There was no overlap between the tools, the data, the expertise or the people working in the different camps. The closest these groups came to working together was the scheduled feeds from the enterprise resource planning (ERP) system to the data warehouse (DW), but even these took place before any BI work was done and were generally handoffs to the DW team.
It was also easier to categorize software vendors by ERP and DW/BI domains. Nowadays, however, ERP systems can be deployed on multiple relational databases, their schemas and application programming interfaces (APIs) are published and DW/BI vendors partner with ERP vendors. You can use an independent BI vendor’s products to query, report and analyze data from ERP applications. In addition, ERP vendors are selling their own DW/BI solutions as extensions to their product lines.
The upside of operational BI is that it is simpler for both business users and IT to use one suite of BI tools to access, report and analyze business data, regardless of where it is stored. Business users need only learn one suite of tools. IT only has to learn, design, deploy and support one set of tools. The enterprise cost is lower with reduced licensing, maintenance and infrastructure costs.
The downside to operational BI is that vendors, IT and business users oversell the concept and confuse the BI tool and the data it accesses. Because operational BI systems let you use the BI tool to access an ERP system directly, users start thinking that they should always go to that data directly and in real time. Just because you can do something does not mean you should or that it makes sense in a business context. Somehow people start thinking they do not need those data warehouses and data marts anymore. After all, they think, if you can get to the data directly, then why spend all that time building and maintaining a data warehouse?
Don’t give up on the data warehouse. Operational BI shines when it is part of an overall portfolio of business information solutions. Data integration that involves data cleansing, data integrity and data transformation is vital to creating useful business information out of raw data. Many companies have multiple customer, product, part, services and employee lists. Health care has multiple doctors, patients, providers, insurers, procedures and prescription lists. They all need a data warehouse to handle the creation of consistent reference lists (or conforming dimensions in data modeling lingo). You are not going to do this on the fly when accessing ERP data from multiple systems across the world in real time.
The danger is considering operational BI as a panacea for your business information needs. Too often, the latest technology is seen as a solution that avoids all that tough, time-consuming, data integration stuff we have been doing for years. It is not a shortcut. IT and business still need to communicate and agree on the data definitions and data transformations for business information.
With the qualification that operational BI is part of the overall solution and not the solution, here are my recommendations for putting it to good use:
With a little forethought and planning, operational BI will only blur distinctions between categories, not the quality of your BI initiatives.