DS0509_Opinions_p12-p13_Trillium:Layout 1 29/4/09 17:25 Page 13
DATA QUALITY
Ed Wrazen
Vice president of product marketing,
Harte-Hanks Trillium Software
self-sufficiently access, understand and with business users on sample data you should Migration to production is typically scheduled
investigate the data and offer insight. now be able to conduct a volume data stress for an off-peak time. Other decisions to be made
Advanced data profiling and discovery tools test. Rules can now be validated and fine-tuned, might include whether to first run old and new
eliminate this difficulty by providing an through an iterative analytic process. This systems in tandem.
intuitive interface and insight into all types of requires understanding the intended meaning The first step in going live often involves an
data issues. Data and business teams are able of the data and business users and data initial load or cleanse process. Errors are likely,
to work together, prioritise and resolve issues - analysts should work together here. This is also but if you have followed a proven methodology
and determine appropriate remedies. your opportunity to review and add your such as that outlined, any problems now are
Having a good insight into the source data, specific terminology; industry terms, company likely to be minor. You’ll have the experience and
check it supports the anticipated data model. definitions and regional colloquialisms not tools already in place to swiftly assess and
Again a good data profiling tool can display the initially part of the standardisation terminology. resolve issues.
schema that represents the state of the native This is a chance also to determine terms
data and these can be cross-referenced with the requiring geography-specific standardisation. PHASE 5: GO LIVE
intended model. Spotting problems now will With validation and tuning underway, You should ensure that a cross-functional team of
save potentially weeks of manual effort later. processes are ready for integration into business analysts, performance engineers, data
applications and services. If architected for the architects, field technicians and vendor contacts
PHASE 3: IMPLEMENT enterprise and with an eye for the future (as is are ready for emergencies. They can intervene
Planning completed, it’s time to improve the best practice), these processes and the rules after a problem has been clearly identified.
data by implementing the data quality developed, should be reusable across multiple An ongoing process for problem resolution
processes and rules, using automation systems, platforms and applications – and an escalation tier hierarchy is important.
wherever possible. The first step is to create a including expanding to a multi-use, real-time Help Desk: IT: Data Stewards: Project
User Acceptance Test (UAT) plan to confirm process if needed. Managers. Note that there are alert functions
that the data quality processes are producing within better data quality tools bringing
desirable results. PHASE 4: ROLLOUT PREPARATION attention quickly to any critical issues or
Implementation can then begin, typically One of the first steps in preparing for rollout is to degradations below thresholds.
including cleansing, standardisation, complete UAT with user sign-off. Business and
enrichment and matching/linking (See technical users should again collaborate aided by PHASE 6: MAINTAIN
“Trillium Software Data Quality Methodology” data profiling and discovery tools to quickly You ought now track data quality levels over time
available at www.trilliumsoftware.com). As you address any concerns. – and take actions to ensure your data assets are
implement the new data quality process Part of your preparation should also include maintained as a trusted source. Reporting trends
designs, have business users review sample briefing the Help Desk so they know how to regularly to top management will aid executive
data results to ensure it meets their escalate any technical difficulties. User training is support for action in the event of any decline.
expectations. They should be able to compare essential too and will facilitate end-user
and report using the same data profiling and appreciation, approval and adoption of any new Download the 22-page white paper,
discovery tool they’ve used all along. processes affecting them, such as new fields in "Data Quality Essentials" from
Once results are verified, test more data entry, screens or pop-ups requesting www.trilliumsoftware.com
thoroughly; having already performed Q&A validation of automated cleansing and matching.
OPINIONS ON DATA | MAY 2009 13
Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17