Integration Software

Metadata as a pilar of a Self-Service BI foundation

The high cost of traditional BI

Today, most of the large companies have a solid corporate Data Warehouse and BI front end in place. The (historical) data we have been gathering into those data silos has been through classical BI development cycles, involving IT, BICC and business teams. But the BI delivery as we know it, has a price, a high price and return on investment of the BI output, being the reports, analysis and dashboards, is difficult to measure, ironically enough.

The BI lifecycle is costly, for changes and new requests the same parties as mentioned above are involved. In that story, IT does not come for free due to its procedures and regulations, its structure. Add to that the difficulties we keep facing when gathering requirements and translating them into valuable and actionable output (first time right), we cannot argue the fact that the whole BI process is costly and time-consuming. With business eager to get answers and insights asap, the classic BI offering is starting to get questioned.

But the need for BI remains, more than ever. With new data to be discovered through Big Data and Semantic technologies and the growing Open Data resources we find on the web, we can also argue that BI still is very crucial for our companies to remain competitive and to support growth. No doubt about that.

Self-service BI as a shortcut

A self-service BI strategy can help in solving some of the issues traditional BI environments are struggling with. First of all self-service BI moves from a centralized and controlled environment to decentralized and autonomous data islands (1). This requires new governance rules, different user roles to be defined. BI competence centers will need to find a balance between BI centralization and decentralization. The BI software stack needs to be extended with the right tools for the right purpose to be used by the right user profile within new contexts and governance agreements.

The BICC needs to know what its key users are capable of and where they can assist in helping them to do data discovery or in facilitating them to gain fast access to the data they require. In a lot of cases the data might already be there, available in the corporate DWH, but the BI consumer or business just doesn’t know it. Investing in a better understanding of the already available semantic layers, will also help business users to independently and quickly find what they are looking for.

One way to do this is by putting IT BI staff together with business users, so they can work closely together. Another way to achieve better insight and knowledge of the already available data is to provide access and insight into the metadata.

Visibility to metadata

BI teams have been delivering reporting, dashboard and analysis interfaces for many years now. In most cases, these BI outputs are based on semantic layers, which provide a business view on top of the DWH (database), hiding query logic and join complexity and translating technical labels into meaningful business names. Good knowledge of these semantic layers starts with knowing where the data provided is coming from, mapped from and what calculations or transformations are being used to come to a specific measure, ratio, etc.

This is where data lineage reporting provides answers. By providing access to data lineage to your business users, you can eliminate going back and forth between IT, BICC and Business to gain insight in data mappings, calculations or transformations being applied by the DWH (ETL). In addition, providing a browsable metadata layer allows BI consumers to look for specific data and to find it there where they might not expect it, or have forgotten about it.

An open metadata layer on top of your DWH removes already one IT bottleneck: the knowledge of where your data is coming from and how it is transformed is no longer in the head of IT BI developers only. It creates data confidence and gives back part of the control to the end user.

You can find out here how CoMetReporter provides such a metadata layer on top of an IBM Cognos BI environment:

Leave a Reply



Twitter Feed

Contact us

three × = 9