Data quality management – first steps

How should a company begin to put in place a data quality management strategy? First, as with all disciplines, IT executives must assess the current state of data in the enterprise.

Once they’ve completed the assessment, they should begin to spell out policies for data quality management along four key parameters.

* Data classification: Determining which data need to be kept accurate and complete, to what degree, and according to what timeframe (real-time, daily, monthly).

* Organizational structure: Determining who in the organization has ultimate responsibility for data quality (the applications team, the business owners of the data, the information steward). “[We need to] put a greater emphasis on bottom-up efforts as opposed to top-down efforts,” says a technical consultant at a carrier that has recently embarked on a program to merge all mission-critical data into a single data store. “The company generates an enormous amount of requirements and technical documents and architectures – none of which has anything to do with reality. Efforts that are successful are bottom-up: People start trying to understand what the data is and how it’s used, and make it easy to record that information.”

* User classification: Determining who is responsible for maintaining data quality on the user side, especially at point of entry and transition.

* Technologies: Data profiling, data standardization, data enrichment, data integration, and data monitoring tools.

Deciding how much emphasis to place on DQM – and what kind of resources to put behind that effort – is a job that should involve more than just IT, and include the business owners and users who will be key to making any quality initiative a success. Specifically, it should include:

* The data entry personnel, to ensure that IT’s policies, processes and technologies are easy to use and are embraced

* The data owners, who should have plenty to say about what matters when it comes to quality (and, perhaps, what doesn’t)

* Customer service representatives, particularly when it comes to demographic data, and often data entry

* Sales and marketing, which can have clear requirements for customer data and its definition

* Product development teams, who are often responsible for creating and developing some of the most critical data a company owns, on its own products and services (including pricing and specs)

* The finance department, which may have unique requirements for data accuracy and compliance

To get real buy-in from these groups, it’s important to do more than just pay lip service to their input. Some companies may find value in creating a DQM czar separate from the information steward, especially those that rely heavily on sensitive, and increasingly regulated, customer information. In these cases, money for DQM technology and personnel may also come from the business units with the most interest in preserving the quality of their data. Other companies will make the same person responsible for DQM and information stewardship overall, in which case the DQM budget is likely to come from a general IT resource pool.

Either way, he or she should create a task force that includes representatives from all the relevant business units, and which meets regularly to discuss and update DQM policies, definitions and procedures. The group should also be prepared to evaluate any technology in use, and monitor the existing systems for gaps and success.