This week, I was honoured to be invited chair the DataQuality Asia Pacific Congress, this year held in Melbourne. (Special thank you to Sangita Rai and Katelyn McGee from ArkGroup for preparing such a well-run event.)
For me, the event is one of the highlights of the Information Management calendar and this year’s Congress featured two days of conference proceedings, together with an additional day of in-depth workshops (including my own half-day session on Managing for Effective Data Governance, presentation materials for which are available here.)
Participants in the conference came from commercial, public and not-for-profit sectors and featured contributions included: Melbourne Water, Telstra, National Australia Bank, Australia Post, Insurance Australia Group, ANZICS, Mater Healthcare Services, Australian Bureau of Statistics. Contributions for the vendor and service provider community included presentations from ExperienceMatters, Accenture and BigData-Startups.com.
|Telstra: Winners of the DQAPAC Enterprise Award for 2014|
In addition to the main conference, the Asia Pacific chapter of the InternationalAssociation for Information and Data Quality (IAIDQ) presented the 2014 Data Quality awards, with honours going to:
Project Award: Winners: Mater Healthcare Services; Runner-up Telstra.
Enterprise Award: Winner: Telstra. Runner-up; SBI Insurance.
There was so much to take in throughout the two days that it’s pretty tough to try and summarise everything! (I will probably revisit some topics in more depth in future posts…) But for now, here are my “top 10” highlights:
1. Data Quality is boring!
Or at least, it can be perceived to be boring by people who aren’t actively engaged. Passionate people are needed if the data is doing to be cared for, so you need to engage in a manner that is relevant to the business community, not concentrate on the practitioner processes. Use language and examples that are business-oriented. Make it fun!
Branding is also a crucial element of the communication process – it raises the visibility of Data Quality & Governance, and makes it more engaging. (e.g. meet Deekew, the mascot for data quality at SBI Insurance.)
2. Learn from you mistakes
Don’t repeat the failures of before. Projects should incorporate a “Long Hard Look” exercise at the start of a project phase as part of project initiation and planning, not at the end of a project (when the lessons will quickly get forgotten).
3. Plan for change, and expect to change your plans.
Having a good plan will enable you to adapt later. But don’t adhere rigidly to the plan, because reality never turns out the way you though it might. Be flexible, be adaptable, be responsive, and be available.
4. Identity Management and Federated MDM
The concepts of “Party” and “Roles” are critical to enabling a single consolidated understanding and unique identification of each “customer.” At some point, an individual could play any number of roles in interacting with the organisation (e.g. customer, member of staff, supplier, broker, agent, benefactor, consultant…. IAG currently recognise 37 valid roles that a Party may play.) The concepts of Party and Role require constant ongoing education to the business.
5. Customer-Centric = Information-Centric
To enable a customer-centric approach, an information-centric enterprise view is foundational; it enables both cross-application integration and analytic insight. System- and Process-centric architectures drive compromises (because of silo thinking being built-in by design). As a result, data quality suffers. We need to have an information-centric architecture to overcome this.
6. Data Quality Declaration
The Australian Bureau of Statistics (ABS) publishes a Data Quality Declaration statement (DQD) with each data set it issues. The DQD provides contextual and narrative guidance to data consumers as to the relative suitability of the data set within a given context. The consumer can then adjudge whether or not the data set is suitable for their purpose.
7. “The data is always right”
A data quality error indicates a failure in the process, the system or the people. Use the data to inform and drive process change.
8. Data Quality by Design
In manufacturing, the production line would stop if there was a failure in the quality of products. Why is it not the same with failures in data entry?
Note that the just because we have the ability to profile a feature doesn’t mean we should! 100% data quality is almost never necessary (at least for analytic decision-making). Pragmatism should be applied to prioritise profiling and remedial efforts.
9. Useful Reference Frameworks
Several public domain reference frameworks were identified as being useful to the Data Quality Practitioner community, including:
10. Data Quality with “Big Data”:
In “Big Data” environments, the velocity, volume and variety of data processing create issues for timely measurement of veracity. Compromises will be necessary and we therefore need to be very clear and pragmatic about applying our data quality checks:
- When to check.
- What to check.
- What rules apply.
- How frequently to check.
- Where to check (which steps in the data processing flow).
Please do leave a note to let me know your thoughts, or to share any similar experiences.