The causal link between poor data quality and business losses has been well documented. Organisations worldwide lose billions in revenue and missed opportunities annually due to the inadequate state of their data.
Part of the problem is that many executives seem to think that improving data quality is a once-off exercise, and if they have conducted a data-cleansing project as part of launching a data warehouse or other business intelligence initiative, the job is done. Nothing could be further from the truth.
Data quality is a lifelong process, once begun, never ended. Its resolution involves an enduring corporate commitment to improving data on the one hand, and the application of technology on the other.
It is regrettable, therefore, that many executives equate data cleansing with the far broader process of data quality: this includes business areas such as data profiling, standardisation, validation and enrichment, while data cleansing tends to form part of the data preparation process in anticipation of submitting it to the data warehouse or data mart.
What is regrettable is that no matter how hard IT tries, it cannot get the boardroom to involve itself directly in overall ownership of data as an issue. The ideal would be for the executive to make data quality part of its overall mandate, but the record shows that just does not happen. The executive is simply too pressed with too many other issues to make data a priority.
So, realistically, the onus falls back onto IT to manage data correctly, diligently, and consistently. Quite honestly, no one else is going to do it.
Only IT understands the importance of data. Only IT folk have the capacity to change issues, direction and events.
They need the buy-in and the commitment of the business and especially executives to liberate ongoing funding and break through logjams if they are to succeed in the long-term, but ultimately the responsibility rests on the shoulders of the IT team to make it work.
Here are the steps to make it happen:
* Create the ability to measure data quality. If you can't measure it, you can't manage it. This is true of any business imperative. Through measuring data quality, companies can gain an idea of their data pros and cons. To give effect to this, design data quality rules, fine-tune the rules and construct a data quality scorecard.
* Designing data quality rules includes analysing data relationships, assembling expert knowledge and data profiling.
* To analyse the data, companies need to know how the data elements relate to each other.
* Expert knowledge requires communication with business users and identifying what data means to them.
* Data profiling provides an automated means to measure data quality. It extracts data from the current system and builds a knowledge base, allowing data analysts to examine inaccuracies and align data according to business rules.
Only IT understands the importance of data. Only IT folk have the capacity to change issues, direction and events.
Mervyn Mooi, director of Knowledge Integration Dynamics (KID)
It is advisable to trap data errors at source, as far as possible. As examples, data error can be introduced through data capture, legacy system conversion, or inconsistent conversions and field mapping. To offset this, business and IT should work together to define common conventions and routines. These should be reviewed on a regular basis.
Correcting data quality is a life-long process, as noted earlier. To obtain corporate and executive buy-in, it is advisable to begin with an area of extreme pain in the business - perhaps a department or division which is struggling with its data, and experiencing consequential business losses.
Sort out the data problems here, then let word of mouth spread regarding the success. Obtain further departmental buy-in, and let the virtuous feedback loop drive the success of the project. Technical analysts must work with the business, which understands the data and the objective it serves. Together, they need to build a good, high-quality database to store their data.
The benefits will easily outweigh the costs: IT does not waste its time trying to reconcile data; the business makes fewer customer-facing errors and makes better decisions.
The record shows that organisations which create a task group comprised of both business and IT folk, monitor their data regularly, and take proactive steps to improve its quality, enjoy a significant return on investment. Those which don't, continue to struggle with their data, with consequential losses.
Share