Making business needs drive the data ownership model
A short while ago I was working with a regular client on data quality and our conversation turned to the difficulty of assigning ownership for the data. Whilst we talked about their issues a light went on in my head to indicate the arrival of a new idea. I say “new” but of course I really mean “new to me”. However, subsequent internet searches did not unearth any references to similar approaches.
In the end I thought “if this is new to me then maybe it will be new for others and I should write something down for those who have a similar problem”. My apologies to all those for whom this is a well-trodden path and who I imagine are shaking their heads in disbelief at my previous ignorance.
Let’s set the scene
First a few words regarding my client’s landscape. They are a large, multi-national, food manufacturer in a long-term relationship with SAP. They have a data governance department that manages the data quality for master data across the enterprise, primarily materials, customers and vendors. The governance team manage the quality rules regarding master data and are very active in identifying non-compliance. However, the governance process suffered due to a lack of involvement by the operational line management.
One of the fundamental contributions to governance is the alignment between the data rules and the business direction. Do the existing rules support our business needs? Are there areas in the business that will change in the future and need a different set of rules? What is the action to be taken within the business process to reduce non-compliance?
These are the duties that would I would normally assign to the role of “data owner.” This data owner is also held accountable by the senior management for improving the level of compliance to the rules. It sounds straight forward but assigning ownership for master data can be tricky.
Customer data is usually straight forward, as it is almost exclusively in the domain of sales or the O2C value chain. Likewise, the Vendor data is closely related to procurement or P2P. These business functions are usually managed by specific business managers and they would be the first candidates to be data owners for the relevant master data.
The difficulty is in finding a data owner for material master data and true to expectation this was the area that gave my client the most issues. The complexity of material types and life-cycle mean that it is hard to find a “natural” owner for the master data.
In my client’s manufacturing environment materials that were sold might be manufactured or bought-in, a material could be a raw material or an ingredient or packaging for manufacturing, it might be a reusable container for the logistics team, it could be a spare machine part for maintenance, or supplies for the office.
Every one of these types would be managed by a different internal business team. Even within the manufactured materials the lifecycle might touch many business functions – R&D, product development, manufacturing, procurement, tax and finance – and all of these have their own, unique master data relating to the same product.
All this complexity means that there are no obvious candidates for ownership of the material data. A common solution is to create ownership by committee but this usually leads to a slow process for issue resolution and still requires a strong leader figure to act as chair to the committee.
The governance team at my client had developed automated, monthly quality reports but without clear data ownership there was no business forum to discuss the suitability of existing rules or agree on new ones. Most importantly the governance team were seen as responsible for failures in compliance while they had very little business influence to address these failures.
Finding a new governance approach
We were discussing the difficulties of allocating data ownership, and especially ownership of materials, when I saw this new (for me) solution.
My favorite analogy when discussing governance is to liken it to the governance of the country. Most countries have a set of rules that govern how its population should behave. We call these rules “The law.” These rules are analogous with our data quality rules. The rules are managed and compliance is enforced by the judges and the police.
These roles are comparable with the master data governance team in my client’s operations. However, while the judiciary may manage the collected laws and the police monitor compliance, it is not they that set the laws. These rules come from the government who make the decisions on which laws are required. It is the government role that is comparable to that of data owner.
It was while thinking about this analogy that I saw a potential solution to my client’s issues. If we took the analogy further we could see that the national government is not one person but a collection of “managers” and decision-makers that make rules within their specific areas of interest, housing, agriculture, transport etc., and that “The Law” is actually a collection of all of these rules.
When a specific law requires correction it is the department that raised the law who propose adjustments and that same department that is held responsible if the law does not deliver the expected benefits. Why couldn’t this same system be applied to governing our master data quality?
In this new compliance model the governance team would continue to own and manage the “book of rules.” They would be the gatekeepers for new rules being created and would perform impact analysis, stakeholder analysis, implementation, communication, etc. for new rules.
However, before any new rule could be implemented there must be an identified owner associated with the rule who had been onboarded to the role. This would usually be a person related to the area where quality issues had greatest impact and who therefore had the greatest interest in maintaining good quality. The governance team would also continue to produce monthly compliance reports but these would now be sent to the rule owner for assessment and action. Where corrections to the data were required these would be actioned by the governance data experts but only at the request of the rule owner.
Finally, approval for changes to the rules could only come from the rule owner.
To implement this new ownership model we agreed that the rules would be created with the governance team named as rule owners to reflect the current reality. Their first priority was then to onboard business-based rule owners and to update the rule accordingly. In this way the implementation of rule ownership could be done immediately but with ownership being transferred over a period of time as the business owners were identified.
What’s the benefit?
As I discussed this approach with my client we could see a lot of benefits: -
• A better understanding of ownership by the business. Often I find that the business users struggle to understand the concept of “Data Ownership”. How can they “own” the whole data object when some attributes are outside of their business domain? Owning a rule is simple to understand, and more tangible, especially when it was created at your request to manage business issues that you have identified.
• Improved performance by business owners. By refining business ownership from a whole data object to a group of data rules we reduce impact and workload on a single data owner. This should improve the quality of the work that the owner is able to deliver.
• Improved business involvement. The ownership responsibility is spread across more resources and by assigning owners closer to the pain-points we get a high level of involvement with the issues and more drive to improve.
• Clearer responsibility for the business. It is very difficult for other business departments, when finding a data quality issue, to find the correct contact person, especially when a data object is owned by committee. It is also easy for members of a committee to avoid responsibility when faced with issues. Rule ownership gives a clear, single point of contact and responsibility.
• Flexibility. The rule ownership model makes it simple to expand the governance landscape when new data rules are identified and even allows for regional/geographic rules to be developed with regional owners.
Let’s wrap this up
My mantra when designing a data governance organisation has always been “An organisation that fits the needs of governance but also fits the ability of our organisation to execute and sustain it.”
I feel that “rule ownership” not only fits the aims of governance but makes a much more direct connection between business pain-points and those who should resolve that pain; it can be easier to understand and implement; and finally, by making the governance less “monolithic” and more flexible, I feel that the client has a better chance to execute and sustain it.
There is no single solution to governance but “Rule Ownership is certainly an approach that I will consider again if the situation dictates it.