Letting the Data Speak for Itself
Once you extend your thinking beyond the data warehouse and "free" the data to speak for itself, the potential applications of the data un-warehouse concept are virtually unlimited. Let me suggest three powerful possible applications that would offer substantial benefit:
- As a “Trust but Verify” engine for senior management to check the validity of conclusions resulting from analysis based on data warehouse information
- As a “Sanity Check” component of a larger IT/Business Alignment process, an exercise now recommended by many IT consultants to ensure the synchronization of business objectives and IT actions
- As a “Project De-risking” device applied to IT development projects, particularly Business Intelligence projects.
Trust but Verify
The notion of a “Trust but Verify” application derives from a concept promoted by President Ronald Reagan in the 1980’s as the US and the USSR undertook a nuclear disarmament process. The idea was to develop a process in which neither party would assume that the other party was lying, creating a sense of professional good will. Yet should a question arise, either party had the right to examine the details – the data of record – to verify the word of the other.
This practice is now applicable to issues of corporate governance and to situations where considerable investment in mergers and acquisitions is based upon analysis from data warehouse applications. Senior managers would now have the means (they have always had the right) to examine the data in the data store of record to verify the factual basis of proposals, to prevent a misreading of the situation and avoid the costs of mistaken action.
Sanity Check
The “Sanity Check” application relies on the ability of the original data to enable insight into the alignment of IT practice objectives with the business goals of the organization. The high cost of information technology and staffing has prompted firms like Gartner Group to develop an IT/Business Alignment process for determining to what extent, if at all, the process of data warehousing is leading an organization down a path that diverges from its corporate objectives, and a data store of record that supports full analytical access would provide an ideal tool for performing such an assessment.
For example, undue influence of the IT department on product choices, operational procedures and the technical organization of the data warehouse may be inadvertently masking the company’s vulnerability to changes in the market such as pricing increases, decreased demand or a shift in competitive practices. Using the data store of record as a source of information in “compare and contrast” exercises could help identify errors and suggest ways to correct the procedures and restore alignment with business objectives.
Project De-Risking
The “Project De-Risking” application would offer corporate management teams the opportunity to control one of the most volatile cost centers in business. For a variety of reasons, new Business Intelligence (BI) projects incur unique risks. Compared to a physical construction project, BI projects are built on an uncertain foundation, using untested materials, by workers without relevant experience, and no clear architectural plan. It is no wonder that failure rates are as high as they are. Yet, while de-risking BI projects is a critical goal, no standard industry best practice has been developed to address the issue. The data un-warehouse could answer this need by providing the data of record, untainted by assumptions of other projects that have come before. The data of record can be used as a test bed for the assumptions of the project – the equivalent of a control group in clinical trials.
Delivering Real Understanding
The idea of building a reference data repository alongside the data warehouse may sound radical, but the notion of a data warehouse was itself radical not that long ago. The concept of a parallel system that maintains a historical data store of record is as time-honored as the techniques of double-entry bookkeeping to maintain auditability and separation of responsibilities (e.g. Accounts Receivable versus Accounts Payable, General Ledger accounting versus internal audit versus external audit) in accounting.
New technologies have reduced costs, new thinking has extended the mandate of business intelligence, and new business requirements are driving new infrastructure developments. Perhaps the time has come to look towards a parallel data universe, one that offers the potential for a degree of business insight and project control that is beyond the scope of the traditional data warehouse.
Arthur Ritchie
October 10, 2007
