Demystifying the roles of Data Owners and Data Stewards
Conversations about Data Governance inevitably turn to the
question of responsibility and accountability.
As a practitioner community, we
seem to have settled on the terms “Owner” and “Steward” to represent two
important roles within the Data Governance process. (Sometimes I still
encounter the word “custodian” being used, though I’m always wary when anyone
describes themself using the “C” word, as it usually indicated someone who is a
withholder rather than a sharer…)
However, the specific expectations for such roles are not
always fully articulated, can be defined in unclear terms, and may well vary
from organisation to organisation.
I suggest that it’s important to encourage clarity of
expectations, while also keeping things as simple to understand and practice. A
rules-heavy, task-intensive and bureaucratic process is unlikely to gain much
traction.
I therefore prefer to focus on behavioural and cultural
aspects, with the aim of encouraging a general shift in approach. I offer the
following summary key features and expectations of Data Owners and Data
Stewards:
Data Owners are:
- Accountable for the effective management of Information Assets
- Must understand the business value of the asset and the way it creates or realises value within all business processes
- Responsible for cross-organisational value of the asset
- Responsible for implementing and maintaining an Information Asset to ensure it is fit for the operational purpose(s) for which it is required
- Responsible for ensuring that an Information Asset has proper quality, security, integrity, correctness, consistency, privacy, confidentiality and accessibility
- Owners should reside within the operational functions of the business.
Data Stewards:
- Operation of the information asset is delegated from the Data Owner to one or more Data Stewards
- Appointed in cases where Data Owner are not in a position to manage an Information Asset directly. (eg. due to workload or complexity of the Asset).
- Assist the Data Owner in the day to day management of information asset
- Ensure that relevant protocols, principles, methods, processes and standards are applied (In conjunction with the Data Owner and other Data Governance functions)
- Extract maximum value from asset over full lifecycle
- Maintain and enhance value of asset, where appropriate
- Manage the asset within owners expectations and requirements
- Acquisition, creation, maintenance, exploitation, enhancement and disposal (on behalf of the owner)
Note
that these responsibilities are often implicit within people’s existing job
functions.
By identifying the roles of Data Owner and Data Steward we are
really just clarifying expectations and commitment to be mindful in our
dealings with data.
What
do you think? Have I accurately and fairly identified a salient set of
expectation for these key Data Governance roles? Or do you disagree with these
concepts? I’d love to hear your views.
Hi Alan
ReplyDeleteI would agree with all of these, but possibly add the following to Owners:
* Ensure the management of the information asset is in line with policy and organisational objectives
* Ensure that the appropriate education/induction/training for the information asset is in place
To the point about acquisition, creation, maintenance, exploitation, enhancement and disposal I would add planning, although this possibly more for the owner than the steward.
Cheers
Adrian
good observations Adrian.
DeleteI didn't state this explicitly in my blog post, but in putting together any set of high-level expectations, I try to focus on identifying the _purpose_ of a role by answering the "What?" question. My aim is to stay away from asking "How?", which should normally by answered as part of implementation plans rather than being statements of objective. ("Why?" is answered by the overall business context).
Your suggestions possibly imply methods by which the stated expectations are fulfilled - in which case, I'd argue that they are "How?" and are addressed as part of strategic planning.
Alan,
ReplyDeleteI am new to this space, but very interested in the dialog. While the definitions above provide a general frame work for determining roles and responsibilities, from a pragmatic stand point, they fall short. I spent the last six years working directly in support of two cycles of SAP ECC 6.0. Both implementations, despite having big four consulting help and an organic COE in place, were marred by data governance and stewardship related issues. We discovered after going live that fields in our legacy data bases were being used in niche manufacturing groups in ways that differed from their use in the main. The failure to detect this during 4 integrated test cycles led to 3 weeks of scrambling and a complete shut down of manufacturing of theses small but critical factories. The impact of DG related issues for the two full cycles was measured in $10Ms.
The question I have is in an EIM framework who is the person and the discipline responsible for understanding the nuance of data uses in these old data bases. Who drives the change management and documentation of issues as business processes evolve? My gut is the Steward is more than you define above. It is person who understands at a data table level how data supports process with an understand of the process at an execution level .... the Master Planner who evolves into an IT professional.
Your thoughts?
Respectfully,
john
Thanks John - a great case study and it doen't surprise me (unfortunately). Time and again, major ERP implementations fall short in this area and to my mind, it's because almost all programmes take the well worn "people, process, technology" approach to planning implementation and change. There's a fourth dimension which ties all these together - Data (Information).
DeleteSadly, too many people in our industry are able to adequately operate in only one of these dimensions (and Management Consultants are no exception!), while the "Data/Information" dimension is often not even thought about at all. The (wrong) presumption seems to be that like-for-like migration of data will work, because if the data was ok in the legacy system, then it'll be ok to just take it over lock-stock-and-barrel into the new solution. FAIL.
In my opinion, until the Information Agenda is also explicitly planned for, resourced for and delivered to, these failures will still continue. All system implementations should incorporate data mapping, data definition and data quality streams as an explicit and important part of project delivery. The down-stream costs of not doing so are significant, as your experience clearly illustrates.
In terms of your "EIM Framework/who is the person" question, it's not the role of the Data Owners & Stewards - their briefs are specific to assuring the operational fitness-for-purpose of the assigned data sets, but they would look to an expert "Information Strategist" supported by an Information Management/Data Governance practitioner group to provide the methods, capabilities and resources necessary to deliver against the requirement.
Does that IM/DG capability exist as a clearly identified and resources capability? Many organisations now have it, or are putting it in place, many still don't have it at all, or still see it as "IT's job".
For reference, I've proposed and example Charter for such an IM/DG capability ( http://informationaction.blogspot.com.au/p/example-data-governance-charter.html ). Also check out my backlog of blog posts, several of which examine requirements for this specialist IM capability. (I'd suggest http://informationaction.blogspot.com.au/2013/06/what-do-simple-folk-do.html and http://informationaction.blogspot.com.au/2013/01/who-is-highlander.html as being particularly relevant to your question).
Also, watch out for a new book from IBM on the role of the Information Strategist which I'm contributing to, which is currently proposed for release some time towards the end of the year.
Good luck!
ADD
Alan,
ReplyDeleteStumbled across your blog as I was trying to find someplace that suggested where a Data Governance group should reside in an organization.
Anyways, I have read a few of your posts and I have to say that while I agree with most of what you are saying, there are 2 things I disagree with. 1 thing in this post and a consistent theme in all of your posts.
First, this post. While it all sounds good in theory, I don't believe this can actually work in real life. If systems are implemented correctly, there is no need to define a Data Steward other than for metadata. If employees work in a system that meets their needs, they will do an excellent job maintaining their data. If a system does not add value to their day to day, any employee will not have a vested interest in what the company says their job should be.
Second, Data Owner. Why do you use this term? In my experience, once you tell someone that they own data, they want to do whatever they want with it. Whether it fits into the company's policies or not. So, over time, the "data owner" does what they want and turns the field from a company field to a department field. In my old job, I had so many fights with employees because the company had said that they owned the data. They would then try to do what they wanted, and as Data Governor I had to step in and stop them. Then I would get a rant about how they own the data and I couldn't stop them. The majority of the time I won and the change did not happen, but it would be followed up with the employee saying why do I own the data if I can't do what I want with it.
Our difference in opinion could just be because of differing industries. I work in consumer goods (manufacturing) and have to deal with Marketing who is creating and driving their product
Hi George. Thanks for your comments. I don't think we've got a difference of opinion - just a subtle difference of perspective (and possibly of semantics).
ReplyDeleteWRT your comment on the role of "Dara Stewards", in an environment like a manufacturing plant with high data automation and low levels of human data entry, I would agree with you - once the system configuration has taken place, then because the data is almost all systems-generated, then the role of stewardship is "light touch" and focuses on maintaining the metadata definitions. In many cases, however, this is not so - many businesses still have significant dependencies on human data input (e.g. bank teller transactions, telco product/service setup and billing configuration, university student enrolments). You also have the problem of correlating and co-ordinating across multiple systems of record. This may not be such an issue in a business such as an FMCG (where at a simple level, most key business processes are administered within either the ERP, or by the manufacturing plant control system); however, in more complex environments such as a Telco supply chain, you may have multiple component systems each dealing with a specific part of the end-to-end value chain.
In such more complex cases, the role of the Steward needs to oversee the ongoing veracity of data creation and maintenance.
As far as the role of the "Data Owner" goes - this is all to do with choice of language in setting of expectations for the role, and then holding the identified owner to account for meeting those expectations. You will see that I identify two key expectations which are crucial to this (I've capitalised for additional emphasis):
* Must understand the business value of the asset and the way it CREATES OR REALISES VALUE IN ALL business processes
* Responsible for CROSS-ORGANISATIONAL VALUE of the asset
Championing a sharing culture and re-use of data within other contexts is therefore a critical duty of the role, and if an individual cannot meet these criteria, then by definition, then they cannot be the Data Owner.
Regardless of the industry, the mandate is for a more conscious consideration of the information needs within the business and improved communication, collaboration and co-operation between departments - only by breaking down silos and looking at the whole-of-business perspective do we drive a culture of information-enablement and increasing value. (See also my post from last year on Information as a Service: http://www.informationaction.blogspot.com.au/2012/07/information-as-service-what-is-it-and.html )
Cheers
ADD
Thanks for replying.
DeleteTo date, I have only worked in data governance at consumer goods companies. So when I think of data governance, I am thinking only in my small world. But, you are right, there are a lot of industries that do rely upon manual input data. Thanks for making me think outside of my box!
And I definately agree with you on the need to break down the silos.
Thanks
Hi Alan
ReplyDeleteHope you are keeping well. Although I quite liked what you said in our discussion (oh more than a year ago) that data steward was a better term than owner becasue "owner" implied - "its mine and nobody else can see it". So I have adopted executive steward and steward instead of the 2 above. More so because I work for government and so, with the exception of private or sensitive data, all data is "owned" by the public. So, I quite like the idea of no public servant or elected representative being "owner" of government data.
Amar
Hi Amar - it's good to hear that you're making progress. If you can encourage a sharing culture through different terminology, that's great - it is the expectations and accountabilities that need to be addressed, so I think it's most important to choose the language that resonates within your own organisation. Good luck ADD
ReplyDeleteHi Alan,
ReplyDeleteI agree with you on that we need to adapt structures and terminology to suit the culture and governance of the environment in which we are working. I do not believe you should try to adopt approaches the from else where. Every company is different and what works for one will be a complete failure in another. In essence, one size does not fit all. GuyH