Damn those definitions!

Communication is the key to establishing a common lexicon.

I recently responded to a enquiry on LinkedIn relating to the problem of achieving consistent data definitions:

http://www.linkedin.com/groupItem?view=&gid=748817&type=member&item=236805368&commentID=134724817&report%2Esuccess=8ULbKyXO6NDvmoK7o030UNOYGZKrvdhBhypZ_w8EpQrrQI-BBjkmxwkEOwBjLE28YyDIxcyEO7_TA_giuRN#commentID_134724817

While the online discussion was very worthwhile and raised several important points, I thought I’d take the opportunity to elaborate a bit further on the topic of establishing a common organisational lexicon. The issue of common data definitions (or more accurately, the lack of them) is one that’s currently exercising a lot of attention here at UNSW, and it is by no means the first time that I’ve been through this process.

In all the situations that I’ve encountered this “common definitions” problem, there are three characteristics that are always evident:

1. Each party thinks their definition is the “correct” one, and it’s everyone else who’s got the problem.
2. There is never “no definition” for a data item. There are always too many definitions (which are incomplete, contradictory, poorly documented uncommunicated etc.)
3. The definition(s) in use don’t actually correlate with the data set that is available, leading to interpolation, extrapolation, assumption and caveat.

In the simplest terms, this all boils down to a communication gap. People just don’t talk to each other. And when they do, it usually ends up being driven by an initial anomaly being identified, which quickly leads to outright disagreement between factions, which invariably then results in battle lines being drawn up.

The role of the Data Governance agent in all this is to mediate, facilitate and communicate – proactively and before any trouble arises if possible. (In this case, the “DG agent” is whoever has picked up the task of resolving the definitional problem. It doesn’t need to be someone who is in a DG “job”; in many cases it will be a project Analyst, ETL programmer, Business Intelligence developer or other solutions person who has inherited the issue by default because they need to resolve the definitional issue in order to deliver their part of the solution).

Whatever the scenario, the agent will need to get the interested parties together and find out whether they're actually talking about the same thing or not. You'll get to one of two states:

1. They may be using different contextual language to refer to the same root content - in which case you've only got one definition (It’s then ok if they want to use their own terminologies within their own context; that's just maintaining a lexicon of synonyms);
2. They're using the same language to talk about two different things that don't agree - in which case you need to give each a specific and separate definition to each that also encapsulates the context.

In the forum discussion noted above, I discussed the example of "Average Revenue Per User" (ARPU) from when I worked for a major UK-based retailer of mobile phones.  For the purposes of this blog, I’ll offer a couple of other examples:

1. Within an Australian Federal Agency that deals with bio-sciences, two researchers were having a significant disagreement about whether a particular organism was to be classified as a “pest” or a “parasite”. Because they both had their particular perspectives (one being from Biosecurity, the other being from Veterinary Science), they could not accept the other person’s definition. By bringing them together and highlighting that the bug in question could be classified in either manner, dependent on the context, the definitional contention was solved.
2. Here at UNSW (and in common with every other university on the plan, I should imagine), the number of students we have within the institution is a pretty important measure! However, not all “students” are created equal, and the trick is to work out the wider context of why the question is being asked:

• Are they full-time or part-time?
• Are they enrolled in degree-award programs or non-award courses of study?
• Are they studying for a single degree or dual award?
• Are they a continuing student or a new enrolment?
• Are they undergraduate or higher-degree students?

 These sort of definitional conflicts occur all the time, and it can take quite a bit of analytical skill, facilitation and communication (as well as patience!) to reach clarity. Good luck!

Higher Education Information Management Strategy Forum 2013

Liquid Learning will be hosting their 2nd Annual Higher Education Information Management Strategy Forum at The Grace Hotel in Sydney from 14-16 May.

I'll be first presenter up on the first day (no pressure...),  exploring The Information Agenda - Why Data Governance is a key to evidence-based decision-making and informed action.

The range, diversity and sophistication of information assets that educational institutions require to operate successfully is becoming increasingly complex. In the meantime, the organisational capabilities necessary to support these information needs are often fragmented, may not evolve quickly enough, or are even absent altogether. The University of New South Wales (UNSW) has identified the need for a more co-ordinated and integrated approach to the collection, collation and use of data and information.

My session will explore the rationale behind UNSW's decision to invest in Data Governance, and will examine what is required to develop and maintain a holistic approach in order to support key organisational objectives, including:

• The information services and competencies required to manage complexity and produce actionable insight;
• Improving Data Governance, data quality and metadata management processes to meet the changing needs of internal and external clients;
• Focussing on information value through assigning formal accountability and decision rights.

More info from the conference web site, here.

2013 Data Quality Congress AsiaPacific: Top 10 takeaways

I was delighted to attend in Ark Group’s annual Data Quality Asia Pacific Congress last week, and privileged to be able to contribute on all three days of the conference (expert panel discussion on Day 1, a presentation on MIS Strategy on Day 2, and leading an interactive workshop on Information Asset Management on Day 3).

The event featured participants from a diverse range of organisations such as National Australia Bank (NAB), Telstra, Walgreens, SA Water, Sydney Airport, State Bank of India (SBI) and the United Nations Conference on Trade and Development (UNCTAD). Their though-provoking content provided stimulus for a highly interactive forum with some great debate about data quality, data governance, master data management and cultural change.

A wide range of topics and themes were touched upon during the event, however some concepts were cropping up consistently throughout the various sessions. I’ve captured my own personal “Top 10 Takeaways” from the conference (not in any particular order):

1.     Branding. Engaging with a whole range of communities requires visibility of the Data Governance group. You’ve got to be actively marketing on a daily basis. Mike Jennings of Walgreens stressed the need to ensure that the message keeps hitting home by having an easily identifiable logo, pervasively used on all output from the Data Quality/Data Governance Unit.

This needs to be reinforced by a clear Vision and Mission statement, as well as persistent and consistent communication of the Data Governance agenda.

2.     It doesn’t matter where in the organisational hierarchy the Data Quality function sits. According to Nonna Milmeister of Telstra, the debate about “does Data Quality live in business or I.T.” is a red herring – it just needs to exist somewhere!

While sponsorship is vital, the function lives within whichever part of the organisation has the most ability to sponsor and influence the uptake of data quality management.

3.     Crowd-source your data quality. Open government initiatives are leading the way in getting end-users of data sets to provide feedback and enrichment.

Tim Moon of RXC encouraged thinking about opening up access to the raw data, new and innovative uses can be found which deliver additional value – to the organisation itself, and to the wider community. (For further view on this, see this excellent TEDTalk by Tim Berners-Lee.

4.     Metadata needs to be accessible if it is to mean anything. Common definitions aren’t of any value unless they’re accessible, in context, and directly relatable to the data set being viewed.

Mark Bands from ANZ suggested that the “holy grail” is to have the business meaning, calculation rules and lineage of a data element available in real-time and in context with a single “Right Click”. The market-leading Metadata Management tools can enable this capability.

5.     Standards aren’t always so standard. Well-defined standards, in conjunction with the right Governance models, can be a force for continuous improvement. Consistency of approach drives quality of outcome.

Unfortunately, the current application of standards & frameworks within the Geospatial Information Systems (GIS) industry mean that the data managed by vendors of different solutions are not necessarily interoperable. According to Jonathan Roach, there is no co-ordinating body to enforce & certify GIS solutions, and the vendors can interpret what frameworks do exist in their own way. A particular problem is variance in data processing granularity, which means that transfer of geospatial data between systems is lossy (64 bit processing for CADCorp, 32 bit for Esri, only 16 bit for MapInfo).

6.     CRM Integration through SOA. Ram Kumar from IAG showed how having a robust MDM architecture helps drive integrity of the customer record, even in federated environment.

For each attribute, one system is designated the “master”, then all other transactional systems call that value on request via the ESB. At IAG, the CRM system doesn’t even master any customer details – its purpose is to manage interaction events.

7.     Set your Information Principles first. Ram Kumar again. For IAG, the Information Principles were first set of principles endorsed by the SBI Board, even before the Business Strategy was endorsed.

IAG have gone beyond the standard platitude of “right data, at the right time, to the right person”. Their approach requires clear definition of:

a.     The right data

b.     In the right place

c.     For the right person,

d.     At the right time

e.     In the right format

f.      Of the right quality

g.     In the right context

h.     With the right security

i.      With the right governance.

Any information set is only under governance when all of these have been explicitly addressed.

8.     Legal & Compliance. Mike Jennings from Walgreens proposed involving Legal & Compliance as an active part of the Data Governance structure, to ensure appropriate guidance and assessment of legislative impacts (e.g. privacy, security, integrity).

9.     Data Governance needs benefit opportunities. Scott Jerome of SKM highlighted that Data Governance can support management of risk, improved efficiency and better effectiveness, while Clint Morrell from AGL identified specific monitised benefits that have resulted from their Data Quality initiative.

Identifying the explicit links from the Data Governance actions to the resulting business outcomes promotes the value and establishes a platform for further investment in improving the organisation’s data assets.

10.  Map your Data Quality workflow. Andrew McAlindon from SA Water asked the following though-provoking questions: what process and resolution path do you go through when an issue is raised? What are the decision rules, who has authority to act, approve & signoff?
Mapping out the activities and responsibilities will help ensure the decision making process is clearly thought through, fit for purpose and visible.

Data Governance SNAFU: "So, why do we need all this Data Governance schtick anyway?!"

"We're all too busy doing real work for all this Data Governance stuff. Thanks for your time, but I really don't think it's for me. Have a nice life." Familiar?!

Well, my esteemed colleague at UNSW, Chris Will, just put me onto a short series of videos by New York University Health Science Libraries, which make the point so much better than any long-winded Blog post that I could come up with...

Data Sharing 1 - Request 

Data Sharing 2 - File Formats 

Data Sharing 3 - Documentation 

(They've also consolidated the whole story into one complete mini-movie here...)

Even more from NYU Health Science Libraries on their YouTube Channel....

Enjoy! (And then die inside just a little when you think of the current reality within your organisation...)

What’s what, where’s where and who’s who?

Developing an Information Asset Register for effective Data Governance.

(Note: this post also published by Image & Data Manager Magazine).

A significant aspect of effective Data Governance is about orchestrating the exchange of information between multiple parties; to facilitate (and arbitrate) a robust, repeatable approach to delivering content in context, in support of more effective and efficient business outcomes.

But how do you do this if you don’t know what data you’ve got, what state it’s in, or who is responsible for it?

For over 10 years now, I’ve been advocating the idea of maintaining an Information Asset Register, as part of an enterprise-wide approach to managing Information as an Asset. (this factsheet from the UK National Archives is a useful working definition of the term).

This approach goes beyond the systems and applications auditing process that takes place within IT departments. Rather, the Information Asset Register is about building up and maintaining a complete, reliable inventory of data holdings within the organisation, the different contexts within which the information is (or could be) used, and identification of the various interested parties  – if you will, it’s an index of “what’s what, where’s where and who’s who”.

The Register is then used as a tool for enabling more explicit and productive discussions about data between respective creator, collector and consumer parties. Crucially, by acting as a catalyst for discussion between information stakeholders, this approach encourages more collaboration across functional boundaries, establishes points of contact for more proactive information sharing and breaks down any existing protectionism within information silos (an approach that a Public Sector colleague of mine refers to as “POIM” – as in “P*ss Off, It’s Mine…”).

It can be seen that this is therefore not a project  - maintaining and publishing the Information Asset Register quickly becomes a key ongoing service provided by the Data Governance function. This requires some incremental level of investment in your Data Governance capability, if only to provide the resources and skills needed to enable the proactive brokering and facilitation of a data-oriented discussion. (Some organisations will require higher levels of investment if basic Information Management practices and capabilities are not yet in place).

The approach isn’t widespread yet, but some progressive government organisations have been taking the lead (notably the UK National Archives, Australian Commonwealth National Archives and Queensland GovernmentCIO Office). One challenge as I see it is that these organisations are taking an approach that is driven out of compliance and records-keeping requirements, rather than seeing Information Asset Management as a value-adding opportunity. I’d argue that if the drivers were based on business improvement and outcome benefits (rather than “meet the basic minimum”), organisations adopting an Information Asset Management approach would start to see real transformational change, almost by stealth. (See also my prior post on the concept of Information as a Service.)

Anyway, the Information Asset Register is certainly an approach that I’m adopting within my Data Governance role at UNSW – time will tell whether it proves to be successful!

Some specific online resources that you may find useful to help get you started:

Office of the Australian Information Commissioner - Information Principles

UK Office of Government Commerce: White Paper on Digital Assets

SAS Institute White Paper on Information Assets

Teradata White Paper on Information as an Asset

Note also that, further to my blog post of earlier this month, I will be running a workshop on this topic at the Ark Group's Data Quality Asia Pacific Congress in March.