The Data Governance Journey
How would you answer the following? I trust the accuracy of data I use in my job. Yes or no? The data I use and the results I publish are consistent with the rest of the organization’s metrics. Yes or no? If you answered ’No’ to either one, here’s a follow-on question – is data […]
How would you answer the following?
- I trust the accuracy of data I use in my job. Yes or no?
- The data I use and the results I publish are consistent with the rest of the organization’s metrics. Yes or no?
If you answered ’No’ to either one, here’s a follow-on question – is data inconsistency a pain point for your organization? If so, enter Data Governance and its proven frameworks, processes and tools.
There are many online sources to assist teams with building a business case for Data Governance and let’s assume your organization has cleared that hurdle. Now you’ve got the buy-in you need to start work, how should you proceed?
Think of Data Governance as a journey.
1. Agree on the Destination
- Having ongoing executive involvement is strongly recommended for any data governance effort. Identify those executives (or their representatives) whose ongoing support you will need to champion department or business unit level changes – these should be the members of a Data Governance Steering Committee (SC).
- Now, have each of these executives define what data governance success looks like for them – how will they know the program is successful? Use this input to craft a vision statement for the Data Governance Program and then work through revisions to get a final, approved version. The vision statement defines your journey’s destination and will be used to measure of progress.
- Set expectations with the SC that they will be required to meet on a regular basis to review progress and may be asked to help remove obstacles that hinder program progress.
2. Determine Who’s Driving
- Next, time to delegate! Have each executive on the SC identify data Subject Matter Experts (SMEs) from their department or business unit. This group becomes the foundation of your Data Governance Working Group (WG).
- Recognize that the majority of the research, discussion and documentation required to align on a set of enterprise-wide data standards will be conducted by the Working Group.
- WG members should have in-depth knowledge of core data subject areas (e.g. data analysts, data scientists) or have technical solutions expertise in (e.g. data architects, data modelers, data integration leads).
- Membership in the WG is often a part-time role assigned in addition to existing job responsibilities. Where possible, work with team members and their leaders to formally carve out time for WG tasks.
- Decide who leads the Working Group (aka the Data Governance Lead), this person will keep the WG on task, act as the primary point of contact for the SC and usually manages priorities and budget for all governance activities.
3. Get On The Road
- A good place to start is to have the Working Group identify logical data groupings across the enterprise – top down is an easier place to start to map the core data subject areas. For example, at Excella the highest level logical data groupings are Financial, Employee, Customer and Internal Operations – all data can be rolled up into one of these groups.
- For each data grouping, assign an owner (the Data Steward) who will manage the governance activities here.
- Create an initial description of the types of data that fall into each group to guide future discussions and work.
- Agree on a method (and template) for every Data Steward to begin decomposing in their data group – I recommend starting with the highest visibility elements, for example, what are the metrics used by Executives? Don’t try to boil the ocean at this stage – if it feels too detailed and becomes overwhelming, refine the approach to limit the detail in scope.
- The Data Stewards can work with contacts in the relevant business areas to collect additional information they need.
- Set a timeframe for all Data Stewards to gather an initial decomposition of key data, then convene the Working Group to review the results and create a high level picture of the enterprise data landscape.
- Now decide as a group, where you want to start establishing governance standards – to make a positive impact, swiftly look for the metrics that cause team confusion (because there are multiple versions of the same metric) and stress (because they are considered enterprise critical). Pick a few of these, or even just one, as a starting point.
4. Maintain Your Speed
- Let’s assume the WG has picked one metric to start with, the process now focuses on establishing a baseline.
- For each metric or data element agree upon and document the following:
- Definition – what information does it provide?
- Data Type – what is the data format?
- Valid Values – are there are a set of limited values or a range the metric should always be within, for example, a finite set of category codes, a time range etc.?
- Source/System of Record – where should users go to obtain it? (If it’s a calculation, state the trusted source for all components.)
- Calculations – if this metric is derived via calculations or business rules, what are they?
- Quality Controls – what quality controls are applied to ensure accuracy and consistency?
- Data Grouping – which of the logical groupings does this belong in?
- Data Owner – identify the Data Steward responsible.
- Other Information – link to other relevant and useful documentation. (e.g. data lineage across systems, relevant technical documentation etc.)
- Now repeat for the next metric and so on. The idea is to build a repository of standards over time. (and maintain them as the enterprise changes)
The next part of the governance journey is getting the rest of the organization to use the standards you are creating – a forthcoming blog post.
Share your Data Governance journey with us at excella.com.
You Might Also Like
Leading organizations realize how critical it is to prioritize data-informed decision-making. When quality data work...
What is data literacy? Data literacy isn’t all that different from literacy in any other...