With one data governance project delivering business value and a broad deployment underway, Canada Post has a good idea of what works and what doesn't work when it comes to enterprise data governance. I recently caught up with James Smith, the lead of data governance at the Canadian postal service, and we chatted about how the organization is approaching this critical process (see Keeping It Private and Canada Post Delivers on Data Governance). Here, he discusses his five top best-practices.
Executive buy-in. "The model starts with having support from senior management to suggest that this is the way they want to go and to align with the strategic plan," Smith says. "That's No. 1."
A data-governance office. Once executive support is in place, the organization needs to create a data governance office and give it responsibility for working with all parties -- including audit, privacy, and security -- in implementing data policies and then communicating them to the enterprise. "That's the fundamental first layer that needs to happen -- and then, once that's in place, you can assign accountabilities with data stewards."
Engagement of data stewards. As we learned yesterday, data stewards at Canada Post live within the business and "own" the data. They're responsible for its upkeep, meaning the onus for quality and compliance falls on their shoulders. They must have a vested interest in keeping engaged with the data, Smith says.
Communication. "It's pivotal," Smith says. This means he himself is "consistently communicating with all the data stewards on the rules, policies, calls, data issues, data solutions, and big projects," and that the data stewards need to communicate among themselves, too. "This really is an enterprise approach, and communications is the key."
Technology. "You can't have technology without data or the data without the technology, both of which need to be aligned with each other," Smith says. Since a new application or system could transform the data and thereby affect the schema or the model associated with the data, following best-practices around data architecture is critical, he added. "That's why IT is represented in the committees." Canada Post relies on data management software from SAS, this site's sponsor, for "de facto execution of the best rules of data governance."
For companies providing 360-degree views of data, the governance requirement is never-ending, Smith said. Do you agree? Share your thoughts on data governance below.
I can see the concern raised about leeway below management, but I also believe the data stewarts, positioned with a communication layer like webRTC, can be a means to share governance issues and coordinate the right solutions.
@kq4ym -- so are you suggested the data stewards would be leery of giving opinions, or those "below" them? If anything I'd think the data stewards, who "own" the data, would feel empowered by this structure, knowing the formal processes and having support from the data council.
As the plan described seems top down management with the data "stewards" in the middle, I wonder if things would work out smoother giving a bit more leeway to those at the bottom of the list. The communication factor seems of course politically and practically correct, but there's always the problem of those at the bottom of the stack a bit leery of giving opinions as the program progresses.
Ugh - I don't like 'information advocacy." That sounds like information is an under-represented entity that needs support. Governance is a responsibility -- which I think is truer to the task at hand. If you don't govern your data responsibly and thoroughly, you're going to be in serious trouble.
That's a lot of meetings -- but I do think it does speak to 1) the importance of data governance, 2) the need for support at all levels of an organization, and 3) the challenges and scope of the work.
Smaller organizations should have simllar processes in place, I would imagine, tailored for their size. Maybe smaller companies would need both the executive-level steering committee and a data council, for example, but rather key top executives who participate in the data council.
Funny indeed. One thing Zurell shared that I found interesting was the structure of her governance task force, as it were.
She said she has a steering committee of VP-level executives that meets monthly, a 'data council' made up of director/manager-types that meets every 2 weeks, and a strategy group of lower-level workers. Each has a "cross section" of LOB and IT representatives.
I wonder how this kind of large-scale initiative would translate in a smaller organization.
Funny, Michael -- I just posted basically the same thought in another post, not having read this comment! It seems these companies are following best-practices for enterprise data governance. Perhaps we'll have to revisit each of them a year from now and see how those efforts have come along and if any of their experiences diverge from the best-practices.