When reading a recent excellent blog post called “How to Assign a Data Owner” by Rayk Fenske I once again came to think about how I dislike the word owner in “Data Owner” and “Data Ownership”.
I am not alone. Recently Milan Kucera expressed the same feelings on DataQualityPro. I also remember that Paul Woodward from British Airways on MDM Summit Europe 2009 said: Data is owned by the entire company – not any individuals.
My thoughts are:
- Owner is a good word where we strive for fit for a single purpose of use in one silo
- Owner may be a word of choice where we strive for fit for single purposes of use in several silos
- Owner is a bad word where we strive for fit for multiple purposes of use in several silos
Well, I of course don’t expect all the issues raised by Rayk will disappear if we are able to find a better term than “Data Owner”.
Nevertheless I will welcome better suggestions for coining what is really meant with “Data Ownership”.
I’m with you, Henrik, in that I hate the term “data owner”, but for different reasons. As is usually the case, companies dealing with their data quality woes have a tendency to have a blind spot when it comes to their customers and their needs, and this is an example.
Who owns my data? Me. It’s MY address, MY name, MY credit card number, MY order information, MY purchase history. And whenever I give my information to any organisation I do so on trust, that they will treat it as a valuable asset which should be treated with decency and care.
In effect, organisations rarely take this view and try to take ownership of the data, and that is reflected in this sort of semantic use. I’d love to see terms used which reflected the true values with which a customer’s data should be treated.
You may be opening a can of worms here !
My perception of “Data Owner” and “Data Ownership” is very different to Rayk’s.
I have worked on too many “end of food chain” projects – projects that depend on existing data – that suffered due to “no clear ownership of data”.
Too often, I have had to reinvent the wheel, from first principles. This incurs unnecessary cost, and is error prone.
I discuss “no clear ownership of data” as common Data Governance issue #4 in the “Data Governance Assessment Process” on my blog at http://wp.me/pzxUA-F
Your post highlights the need we data quality professionals have for a glossary of terms, in which we can develop a shared understanding of terms like “Data Owner” and “Data Ownership”.
Thanks for raising this important topic,
Good thoughts Henrik.
I like to think that when many organisations talk about ‘data owners’ they infact mean something more aligned to the role of a ‘data steward’ or ‘data custodian’. Someone who is the primary contact for the data, understands how it is created, understands which downstream systems/processes make use of this data, and can assist with any queries/issues relating to it.
The term ‘data owner’ suggests firm accountability & responsibility, and that’s probably why it is widely adopted term.
I have however met many ‘data owners’ who were simply key consumers of said data, and probably only became ‘data owners’ because no one else would put their hands up to accept accountability.
If you think in terms of treating data as an asset to the company or organisation then ownership is an accurate descriptor. For example, a retailer has brick and mortar storefronts and distribution centers that are managed by a properties division, a fleet of trucks that are “owned” by the logistics division, an the servers that are the backbone of the IT infrastructure are “owned” by IT. Each of these physical assets are owned and managed accordingly by their respective divisions in the organisation.
Trouble with data is that it isnt something tangible. It is a more difficult sell to management that they need to treat the organisations data the same way they treat their properties, fleets, and IT infrastructure. This is where data governance comes in to play and helps provide the structure for which accountability and ultimately ownership can be assigned to the organisation’s data.
Ah, this old chestnut.
first off, my experience with “data ownership” projects run by IT Strategy teams is that they usually wound up either with IT owning everything (“you think it’s so important, you do it”) or an army of Information Gollums running around the business stopping people from doing anything with their Precious.
Owner implies possession, power and control, not necessarily accountability or responsibility for outcomes arising from the use of the thing (I own my car, am I responsible or accountable if my mother-in-law crashes it?)
From a bluntly legalistic perspective, the owners of all assets are the shareholders, who appoint Directors to take care of their assets for them. A good number of court cases have made it clear that, in the context of corporate governance, the buck stops with the Directors of the business and they simply delegate functions (but cannot absolve themselves of responsibility)
So, if the shareholders own it, what do the Directors delegate to staff? Ultimately it is accountability or responsibility for the information asset, in the same way as they delegate responsibility for the use of a company car or a company laptop.
In the context of Data Protection (which has a strong overlap with Information Quality), it is the DIRECTORS who are liable for prosecution should the breach be big enough. Likewise, in other cases where a duty of care around the quality of information exists it is the Directors who may find themselves being sued.
Graham’s view of data ownership is very much in line with the principles for Data Protection – ultimately it is not the company’s data but the data subject’s data that the company holds on trust.
I agree with Ken that projects suffer when there isn’t a clear line of sight as to who is the person who is responsible or accountable for information in a given context. Someone needs to be the sign off, someone needs to be in charge of caring for and feeding the asset after the contractors and consultants have gone.
However, I would disagree that “ownership” is the correct term, and I would probably disagree across any of the scenarios Henrik has identified.
Perhaps we have approached this from the wrong perspective over the years as a profession, looking from the data upwards to find an “owner”. Perhaps we should simply take a lawyer’s eye view and tell our CEO that they are responsible and accountable for the information in the business (which they certainly are legally) as the information is a key asset that drives the business and can affect customers or regulators if not handled with care. Then see where the Boss delegates things to.
I must agree – it’s about stewardship, not ownership, when it comes to corporate data – responsibility to the shareholders or owners for the proper care of the assets of the organization. At our workplaces, we aren’t considered the “owners” of our desks, or our PCs – but we’re responsible for them. And data carries not just asset value, but also risks of a much higher order if not properly cared for. “Ownership” is an unfortunate term unless one thinks of the old rule: “treat it as if you owned it.” In that context, the value of the idea is clear. Does the word get in the way? Sometimes. But when it does, that’s more of a cultural problem than a semantic one. Candidly, I think there are more useful things to talk about.
I struggle with “Data Ownership” on a daily basis and believe that a data ownership position at the enterprise level is required especially with the movement of a single source of MDM governance. The ownership would ensure a holistic management of all data to meet the business needs whether it is financial, engineering, purchasing, maintenance, etc.
thank you for picking this issue.
I’m very suprised by the sometime harsh responses here, as I thought that “Data Owner” is a common term. I must admit that I was totally wrong.
For me a Data Owner defines a single data, a single attribut, not sure, how to call it. From my perspective a stewardship is a weaker role, guiding users to the data.
But beside the different names (and I’m sorry to say that: I still have no clue, how to call such a person) my issue was, how to assign the role (whatever it is named) to a certain data field. Who is the the one defining descriptive meta data? Who is the one saying that DQ is good or poor?
I’m with you, Henrik. The term “owner” bothers me but I can see your point that stewardship may fail to isolate accountability. On the other hand, I don’t want to get caught up in semantics.
@rayk Apologies if I came across a little harsh with my response. You say that your definition of a “Data Owner” is a person who defines what a piece of data is.
But the company that built my car defined what it is and its attributes. But they don’t own it.
If we consider Information to be an asset then we can compare to other assets. Do you have a “Worker Owner”? What do you call the people in your business who define the attributes and features of the desks you work on or the PCs you buy or the staff you hire?
You go on to ask who the person is who judges if the quality of data is good or bad. That can only ever be the person who is seeking to use the data in their processes (or the external customer) who finds that they can’t do what they wanted because of the quality of the data. In which case, in some cases that person is the data owner (back to Graham’s point again).
As for a steward being someone who simply guides someone to the data, I think this is a limited and overly restrictive view of that role (and again.. I’m not 100% happy with Stewardship as label either). Looking at the origin of the actual word, a steward was responsible for the care and maintenance of an estate or the assets of an estate (i.e. the land around a nobleman’s castle). Such a role goes well beyond simply guiding people to something.
As for the challenge of finding who needs to put a definition around a data field, my experience has been that we need to step back from the data and look at the process or processes that will use the fact that sits in that field. Find out what those processes are, find out who is accountable or responsible for the processes (there will often be >1 process and >1 person with responsibility) and get them in a room to get a definition together.
What this brings to light in most cases is the other special characteristics of data as an asset, especially the fact that it can be used in different processes without being destroyed and that it spans the organisation.
These special characteristics (as described by Tom Redman in his Data Driven book) mean that the actual way in which businesses need to tackle the management and governance of this mega-asset might not be (heck – in my view it certainly is not) suited to the traditional ‘silo’ single point of ownership approach that often exists.
Ultimately, if you need a single name then you need to go to the highest common point on the management hierarchy to find the person who can cascade those meanings and objectives downwards.
Thanks everyone for all the comments. It’s a pleasure to host such debates.
Rayk, you sure have prompted evidence that there apparently is no common perception on what actually is meant by a “Data Owner” and “Data Ownership”.
I think this goes beyond semantics.
Data Governance is an emerging discipline with an evolving definition, so are the terms related to Data Governance.
Wow, great debate going on here Henrik. I often talk about letting the ‘people’ own the data. When I use the term owner, I am talking about having some level of definition and control i.e. every resource that touches data at some point during it’s lifecycle owns the definition and control at that point.
Data ownership can not be seen as an individual owning a specific ‘piece’ of data, it needs to be seen as an holistic ownership of the business over all the data in an organization. People need to understand that they have joint ownership of that data, and that when they interact with it they need to understand the impact of their interactions.
The term ownership is simply semantics when it comes to data governance. However by using that term you instil a sense of urgency, commitment and pride in the business over their data.
“It’s not your data, it’s our data, we are all responsible for the quality of our data, we all collectively own our data”
Don’t get me wrong here, I totally or in most cases agree with what is said here. And a special thanks from my side goes to Daragh.
I really like the definition, which could be derived from the origin of stewardship. But I don not want to go deeper here.
To explain my view onto the DQ world: I have data in my company (or the company I used to work for). Many people are using this data. So we simple labeled them Business User. They do much more, but they also use data. Whom do they call in case of some sort of alignment to the definition of specific data? Do their coordinate on their own? This will be in line with the definition of DQ as company’s asset (which I 100% agree with).
My “problem” is that my personal experience has shown that exactly one responsible person is needed! This could be of course top management. But even if the importance of DQ can not be underestimated, top management sometimes has other things to do. We called this person Data Owner.
This maybe the wrong term within DQ community, but this is what is established in our company. I’m not sure who mentioned it, but on this level it is semantics for me too.
For the comparisons with other assets. For desks and the people working on them we have the HR department and sometime sort of a union. For cars I have at the end a product manager. Not exactly MY care, but the cars of a certain type. Thus the data owner is not responsible for every customer, but for a high quality definition of the separation of Corporate Clients and Retail Clients (bank background on my side).
And: I love this discussion!
On a recent consulting project I assumed the role of “Data Governor”. I’m not sure if that is better than “Owner” but the intentions were good. The intention was to have a grass-roots data governance program and the client had to start somewhere. As part of a committee, I defined rules for data management. We tried to keep our eyes on the “big picture” and prevent the data quality issues we had just resolved on the data quality piece of the project from occurring again. In the end I ran a lot of SQL to rectify the issues caused by those who broke the rules we set. MDM and DQ remind me a lot of the legislators and policemen; two disgruntled step-brothers!
Excellent post and choice of topic. You have again prompted much debate and interest, congratulations.
I agree with many of the above views, but would like to suggest a refinement of the approach which may be easier to ‘sell’ to business stakeholders.
If we take the view that information is an asset to an organisation, typically a high value asset, then ‘ownership’ roles need to be established in order to ensure that the condition/value of this asset is not degraded.
The term ‘owner’ is, I believe, a generic term for a number of activities which are required to safeguard your information asset. The terms responsibility and accountability have been used above in slightly different contexts, however, these terms are key to the overall debate.
Accountability is where the buck stops – typically the person who is accountable for data is not in an active data management role, but may control the staff/budgets of those who do. Responsibility covers the active, day-to-day data management tasks required to maintain and improve the information asset.
If a number of generic data/information management roles are mapped against organisation specific posts/roles as part of a RACI analysis (Responsible, Accountable, Consulted, Informed) then an organisation specific framework of data governance activities is established. This should avoid the confusion about the terms ‘owner’ and should ensure responsibility and accountability are clearly stated.
What is needed for this to work are a clear and consistent generic set of data management/governance activities to form one leg of the RACI analysis.
Does this refinement provide a way to reduce user/stakeholder confusion and ensure that data is better ‘owned’??
Well, what a conversation thread.
My initial reaction was WHAT? with all that talk about corporate responsibility and liability. I agree that when we are sat outside on the pavement its the organisation that “owns” the data; but when I am in the building and looking for an answer I want to know who to ask. And here what I need to know is that the person whose name is on the ticket is the one who will be able to tell me “its 86” – or whatever – and in the comfort of knowing that they have their finger on the pulse – they are responsible, or “own” the data. You can always argue that the term “own” is abused, and I agree. We only have to scan the comments above to see the range of the use/abuse of this simple word.
But lets get back to basics. We need a word or phrase to describe a function. Why not “own” as long as everyone in the organisation understands what we mean?
One of the big issues with data quality is that many in the organisation see it as “someone else’s job”. If we start trying to badge it up then we risk becoming the self-fulfilling prophecy.
We can talk about the person who is responsible or accountable for the data but try talking in the office about the data responsible person…
So, lets continue the debate!
Thanks for the continued stream of comments.
In the LinkedIn group called IAIDQ Information/Data Quality Professional Open Community there are some additional comments.
Also here the roles called “Data Steward” and “Data Custodian” are taken into consideration.
I dislike the term “data ownership” …
From my accounts, data is an asset, but its fluid. how can you label someone or a group of people that “owns” the data.
Application Process, Business Logic and user input change the data. Would you call the Data Architect of an enterprise the owner of the data ? of course not, because he cares about how the data is presented and accessed. Would you call the user of the data, or the person who updates the data the owner, of course not, because they would not know anything about how it is shared, used or stored.
I am with you Henrik. Like many of the folks on this thread state, “accountability is important” SO how to you define accountability.
My suggestion, look at both ends of the candle and all of those in between, that is your “group of accountability” Who designed the model, who manages the process / access, who uses / updates the data, and who reports on it. That is a high level of how I define the accountability group…
You know some of our friends in DQuality (via blogs / twitter and such) all believe that collaboration is key, so how do you turn collaboration into accountability, and will the project, or initiative mandate participation from both ends of the candle and all points in between ?
Great thread, and glad you have lots of input… good stuff… thanks a bunch…
Information “person” or Data “person”.
When discussing semantics I would like to push it a little but further in order to clarify the case.
There is a HUGE difference between Information and Data that has a big impact on the “ownership” or “stewardship” being the subtle difference between the meaning and the representation.
The Information needs to be defined and governed by someone in the organization that takes ownership of the semantics of the business concept. This should be the information owner and is in most cases found close to the CEO (C)level person as this semantics reflect the way the enterprise is defining and managing the information object that needs to be inline with the process view and business strategy.
In this case the customer is never the owner as we a simply addressing the structure and not the instantiation of the information object. Regardless of the point of view we take this information object and corresponding RACI roles are the linchpin of any governance/quality project.
When we talk about Data and ownership I like to look at this from the point of responsibility for the correct representation of the real world object in the managed information holder (IT system, spreadsheet, piece of paper, …) anything that a company uses to execute a business process/take a decision.
One can argue that the true owner is indeed the customer but we need to define an Accountable and Responsible to the instantiation of the data in the information holder and that needs to be consistent with the definitions and semantics of the information object as defined by the Information owner.
Garnie and Jan, thanks for adding to the debate – really appreciated.
I too dislike the term ‘data owner’ – It is too abstract failing to connect with articulate the context purpose and value of information captured, maintained and used via IT systems and therefore fails to connect those responsible at the information conjunction between business process design & mgt and IT systems design & mgt i.e. administrators!!
We need to raise the bar to ownership of electronically managed ‘information’ collections e.g. databases as it is at this level an organisation can define the scope of data to be collected, the methods and limits for its further processing and quality metrics to monitor its reliability and determine the scope and duration of management control obligations required for its protection and disposal.
In InfoSec and IT, “owner” is not a true ownership in a sense, since it can be assign in the corporate world. “Owner” express “due care and due diligence” of a prudent person(s) very well. A prudent person would give anything of materiality to someone who would not take great care of it.
I mean, a prudent person would NOT give anything of materiality to someone who would not take great care of it.