What is Data Governance?

There’s one fundamental issue that underpins your company’s mission. Teams go back and forth on ways to acquire more customers, generate more revenue, and lower their costs. But there’s a much bigger question that should come before all of these:

Do you trust your data?

You can’t address your goals with any measure of confidence until you trust your data. And you can only trust your data once you map out and implement a strong data governance strategy.

Data governance is the practice of creating policies, procedures, and standards to manage and protect data across an organization. Every organization, regardless of their size or industry, can benefit from a comprehensive data governance plan. Because without a strong data governance plan in place, your team could face a variety of challenges to growth.

Since everyone on your team touches your data in one way or another, the most effective data governance plans are collaborative, inclusive, and accommodate every role in your company. And while an effective data governance strategy is a huge step towards making faster, more accurate decisions, it’s only possible through careful planning and a commitment to change management:

Be clear about why you’re pursuing this new data governance strategy, including your goals for the program.
Get buy-in from stakeholders in leadership roles. This will become especially important as you form a data governance council (much more on this below).
Engage all data users in the conversation, and show them why a strong data governance culture is so valuable.
Provide training resources so all stakeholders - from new hires to long-time employees and executives - feel confident exploring your organization’s data policies, especially the company’s data taxonomy and data tracking plan.

Just as an effective change management strategy can guide your team through any type of shift in how your company works, a good data governance plan can improve the results from every other business function. So it might be a bit surprising that many companies only begin to address their data governance issues after homing in on a higher-level challenge. Whether your concerns are related to product analytics, lifecycle marketing, CDP, or any other function, your real issues are most likely rooted in data governance.

In this article we’ll examine the nature of data governance, the elements of an effective data governance strategy, and how to design a plan that works for you and your team.

‍

Table of Contents‍

Why Should You Care About Data Governance?

Before we share how to build an effective data governance strategy, it’s worth discussing why data governance is such a big deal in the first place. Poor data governance is an issue that affects every system and every person in your company. And because your company’s approach to data governance will shape every decision you make on the job, it’s an issue that’s too vast to be addressed by any one tool. To put it more bluntly, there is no such thing as a data governance technology solution for all the data your team deals with every day.

If you aren’t managing and protecting your data with a strong data governance effort - then you are likely exhibiting some of these symptoms:

Wasted Time and Effort - Team members spend more and more of their time correcting data errors, instead of using data to advance the goals of the business.
Impaired Decision Making - If employees rely on inaccurate data, they’re more likely to make decisions that diverge from your business goals and the market as a whole.
Escalated Expenses - Correcting errors, addressing compliance issues, and hacking your way through operational inefficiencies caused by poor data management can drive up costs and hurt profitability.

According to Gartner, poor data quality costs organizations an average of $15 million USD every year. Imagine how much time and money your team could save if all your data was accurate and secure. Imagine how much easier it would be for everyone in your organization to access the data they needed to make timely, well-informed decisions. The good news: you don’t have to imagine, since Harvard Business Review already performed this research. The bad news: they found it costs ten times as much to complete a unit of work when the data is flawed vs when it’s accurate. Worse, only 3% of companies meet basic quality standards for data governance.

So, how do you break into that 3% and build an effective data governance plan?

Elements of an Effective Data Governance Strategy

There are three main characteristics of every good data governance strategy. Data collection efforts must be accurate, events must be organized in accessible formats that follow rigid guidelines, and teams must ensure the security of their data by granting appropriate levels of access to employees and third-party partners.

These three factors come into play in some form or another every time a company digs into their data governance challenges. And they don’t appear out of thin air, but rather from a careful exploration of questions like these:

How do you collect data, and how do you manage the data you already have?
How can team members be confident all your data is accurate and consistent?
Who will create guidelines for how to classify your data?
Who manages these guidelines?
Are these guidelines centralized and accessible to everyone?
Have you established processes for who can access data, run reports, and suggest new types of data to capture?

Questions like these reveal a simple truth: data that’s not actively managed, is mismanaged. And navigating all these questions can feel like walking a tightrope with little room for error. But if this is the first time you’ve considered questions like these, you’re not alone. We’ve seen many different companies adopt the following standards with great success.

How To Design a Commonsense Plan for Data Governance

Every SaaS tool in existence was created to address specific pain points within an organization. But none of these tools can solve those problems in a vacuum. Every stakeholder needs to understand how the tool works, how to engage with the tool to answer key questions, and how to manage the output from that tool to address vital business needs.

As we mentioned earlier, data governance is the foundation of a company’s success, because without accurate data, none of your tools can deliver trustworthy results. And since there’s no such thing as a single data governance platform for all the data in your company, it’s even more important to frame your data questions carefully. We just shared several of these questions above, and a few more specific examples include the following:

Who is responsible for performing quality assurance auditing/testing when new events are created?
When it comes to implementing data events in your organization, who has the final say?
How are event definitions documented at your organization?

The first step to building an effective data governance strategy is to form a data governance council. The data governance council can answer all these questions and many more, so that everyone on your team can work with high-quality data. From there, we’ll cover two important use cases: data governance for new employees and data governance for new product releases.

The Importance of the Data Governance Council

Before we dig into this topic, we must first address the elephant (or mammoth) in the room: why create a data governance council? Are there any alternatives?

We’ve worked on data governance projects with over 850 clients, and during that time we’ve seen two alternatives to the data governance council come up again and again:

The Wild West approach, where data access and accountability is distributed individually, and all users are given the ability to implement events. The problem? Data democratization without guardrails and guidelines just makes it easier for everyone to tap into faulty data. If you don’t implement policies, procedures, and standards to manage and protect data across your organization, the Wild West approach will only create confusion.

Then there’s the Single Point of Contact method, in which one person reviews every newly created event for accuracy and consistency. In this situation, one single person has the final say for all data governance decisions. Elevating a gatekeeper to correct bad data might seem like a logical first step. But in practice, it creates an instant bottleneck and prevents further innovation. That, and it’s slow as hell.

For all these reasons, we recommend creating a data governance council which will be responsible for setting guidelines around data schema implementations, alterations, and deletions. This approach incorporates several different roles that clear a path for clean and useful data.

The data governance council is responsible for managing the entire data governance process, delegating responsibility and ownership, and enforcing standards for the entire company. The data governance council is also responsible for documenting all these standards in a transparent, accessible tracking plan.

Next, we have data owners, who oversee engineers and product teams as implementation occurs, and who also monitor the tracking plan for issues that might come up.

Finally, there are data users. It is their responsibility to understand data naming conventions and event structure, so they can work with the data in a self-serve capacity and answer business questions as quickly and efficiently as possible.

After you establish the structure of the data governance council, create the following policy documents that can be referenced, circulated, and shared with all employees:

Data integrity policy defines how you will maintain accurate and trustworthy data
Data classification policy defines whether certain datasets should have hierarchical levels of access or classifications.
Data access policy clarifies which employees and third party users have access to certain types or levels of data.
Data usage policy defines what actions approved users may perform with different types of data.
Data security policy to ensure that there are clear points of escalation when the data does not meet guidelines. The data security policy will also dictate which data sets will have a higher barrier to entry.

Once your data governance council publishes these policy documents, the foundation has been set. At this point, the data governance council must consider three interconnected motions: 1) define & document events, 2) audit & validate data, and 3) adapt & iterate on that data to serve the company’s needs into the future.

This clip illustrates the progression from a Wild West scenario, to a gatekeeper as the Single Point of Contact, and then finally to a more formalized data governance plan. And while this clip casts the data governance solution Avo as, “the ultimate self-serve utopia for data governance,” this refers only to Avo’s use in product analytics. As we mentioned earlier, there is no such thing as a data governance technology solution for all the data your team deals with every day. For additional context, please watch the full podcast episode here.

Defining and Documenting Events

The data governance council has several core responsibilities in defining and documenting events. They must standardize all customer data by identifying the events to be tracked, and enforce strict object-action naming conventions in accordance with the policies previously mentioned. The data governance council must also create and document a data taxonomy and data tracking plan (see a brief example of a product analytics data tracking plan here). This holds true whether your tech stack is woven together by a series of direct integrations, or if all your tools are connected through a CDP. And lastly, the data governance council is responsible for approving and disapproving any new events to be added to a company’s data schema. In this way, a data governance council is the antidote to the Wild West approach, since event creation can quickly get out of hand in the latter.

Any tracking plan developed by the data governance council should exhibit the following characteristics:

Serve as a roadmap for the implementation of any new technology solutions that rely on that data
Organize and describe event-based data
Enforce naming convention standards

Remember that the tracking plan is a living document that should act as a single source of truth for all the event data at your organization. One of the primary goals of the data governance council is to promote consistency across the company. This is embodied in the tracking plan, which should incorporate a clear, intuitive data dictionary. The data governance council must answer the following three questions for every event definition in the tracking plan’s data dictionary:

“Are we utilizing object-action naming conventions?” Object-action naming conventions begin with the object in question and conclude with the action that a user can perform on that object. Examples include, “Page Viewed,” “Account Created,” and “Button Clicked.”
“Are our event properties dynamic enough for the event to be triggered in multiple locations?” What you don’t want are different events that describe the same action in different locations; this can quickly lead to confusion.
“Do all of our events have clear and succinct definitions?” This is the litmus test for every entry in the data dictionary.

Once the data governance council has defined and documented all events in the company’s taxonomy, it’s time for the next stage.

Auditing, Validating, and Adapting Data for Your Organization

Once the data governance council defines and documents the events necessary for a company’s business goals, they must ensure those events remain accurate. To do this, the data governance council must continually audit and test events for reliability across all destinations. Furthermore, we recommend the results of these audits be preserved in the data taxonomy and data tracking plan. It’s important to note, these types of ongoing data check-ups are usually much easier if all your customer data is organized within a CDP. Since a holistic data governance strategy underlies all of a company’s decisions, and a CDP can unite customer data from across a company’s tech stack, discussions of data governance and CDP should go hand in hand.

When mapping out a data audit, test core use cases to make sure events and respective properties are passed through to the various tools in your tech stack as planned:

“Do we have our data types set correctly?” While every type of data has an optimal format, we've seen many organizations capture all their data as a string. In these situations, they can’t filter their data the way they want, and they can’t analyze it for meaningful insights. Be careful about this error, especially as it relates to your product analytics tool.
“Are the correct values passing through?”
“How do we handle unplanned events, properties, and values?”

As you make a plan to perform regular data audits, keep in mind that some CDP platforms like Segment let you automatically monitor and flag unwanted data. This can streamline the auditing step and make it easier to exclude bad data from your reporting. It is always preferable to automate the auditing process, as manual work at this stage can introduce unintended errors.

Adapt and Iterate on Your Data Taxonomy and Data Tracking Plan

Every bit of data has a lifecycle, and you can improve data trust by confirming data integrity at every stage of that lifecycle. And as your products evolve, the data governance council should examine each event in the data taxonomy and data tracking plan to make sure it’s still aligned with its intended purpose:

How have the original business question and description changed for each event?
Any changes should be documented in the data taxonomy and data tracking plan. How will you note these updates?
Do you need an entirely new event? Or could you add an additional property to an existing event?
Does the event or property you are trying to change have downstream implications?

Recall that one of the data governance council’s primary goals is to promote consistency across the organization. However, keep in mind that members of the data governance council might not be as familiar with certain types of data compared to team members who work with those events every day. For these reasons, we recommend the data governance council create a process to capture new event needs along with a backlog of requested taxonomy changes. An event request change form like the following example can be an effective guardrail for your team’s approach to data governance, and prevent them from devolving into a Wild West situation:

Role & Team
Details of the request
What business question are you trying to solve with the new data?
Is this request a tweak or net-new?
Did you notice this gap within a specific tool?

Organizing event changes in this way can help the data governance team set clear priorities to update the data taxonomy and data tracking plan. The alternative - fixing every request on a first come / first serve basis - can quickly lead to logjams.

In the preceding sections we’ve shared how the data governance council can 1) define & document events, 2) audit & validate data, and 3) adapt & iterate on that data to serve the company’s needs into the future. With these three interconnected motions in place, how should your team provide new hires with the best possible introduction to your data governance standards?

Data Governance for New Employees

There are some elements of new hire onboarding that all companies share. Every company would like new hires to understand the team’s culture and how they can be effective in their specific role. In addition to these, we would like to suggest another topic that all companies should add to their new hire onboarding. Every organization, regardless of their size or industry, should train new employees on their data taxonomy and data tracking plan, as well as their best practices for data governance.

We have seen many companies successfully introduce data governance principles to new hires with the following steps:

Require new employees to read the data policies and data taxonomy & data tracking plan, so they can familiarize themselves with existing events and core business questions addressed by those events.
Encourage employees to review these documents once every 3 - 6 months, depending on their involvement with customer data.
Create a transition plan where every outgoing employee is required to review role-specific data governance practices and any nuances with the new hire.

One of the best ways to create a data-centric culture is to instill the right values around data governance in your team from the very start.

Build a data governance plan with help from the experts

Data Governance During Product Releases

When your team follows clear data governance principles, it’s much more likely your data will be accurate, organized, and secure on a day-to-day basis. But product releases offer a special challenge for many companies’ data governance efforts. Indeed, for many teams, frequent product releases are part of their day-to-day work. For these reasons, the data governance council must address a few additional questions:

How will this product release affect the deployment of events, properties, and user traits to downstream tools / our CDP?
If this product release relies on new data sources, how will we accommodate them?
What steps should we take to avoid polluting production environments with testing data?
How will we share all these changes to the data taxonomy and data tracking plan with our team?

If it’s true for the day-to-day motion of your data governance council, it’s doubly true for product releases: more data does not lead to more insights. Greater insights are only possible if you make a plan to smoothly incorporate all these new events into your existing schema.

How To Set Up Your Data Governance Plan Correctly The First Time

In this article we’ve provided a high-level guide to organize and secure your data while ensuring its accuracy. Once you advance beyond the chaos of a Wild West scenario and the challenges of a Single Point of Contact, you can build a data governance plan that aligns with your company’s goals at any scale. Along the way, you’ll be able to address comments like these:

“My engineering / data team won’t prioritize getting me the data I need.”
“My data team is becoming a bottleneck in business processes.”
“The Exec Team keeps asking me for data about things that don’t make any sense.”
“The Exec Team tells me they don’t trust our data, but no one can pinpoint which data is incorrect.”
“We have conflicting numbers for a single metric.”
“Without solid data to work with, I’m unable to project future performance accurately.”
“I know I need to be data first, but I don’t have buy-in from the rest of the team.”

And yet.

While every company must face issues like these as they enact the policies of their data governance council, it will look different for every organization. At Mammoth Growth, we’ve found it valuable to frame any data governance strategy within these four core pillars:

Data collection efforts follow a prescribed process & incorporate quality assurance tactics, so the data can be trusted
Data is documented and organized in accessible formats that follow clear guidelines, and team members are educated on their proper application
Tool and data access is granted at the right level to all employees and partners
Ownership and accountability for all facets of data governance are clearly defined and followed

In order to address companies’ ultimate goals for data governance, Mammoth Growth developed our Analytics Architecture program. The objectives of this 8-10 week project are a complete audit of a company’s tech stack & data infrastructure, and a roadmap for improvements to their data governance strategy and any corresponding initiatives. The Analytics Architecture program allows Mammoth Growth to develop a deeper understanding of a company’s existing data governance methods and pain points.

When approaching data governance projects, Mammoth Growth follows these steps within our Analytics Architecture project:

Investigate how a company’s data flows through all the tools in their tech stack.
Clarify how teams currently manage different tools used for product analytics, lifecycle marketing, CDP, and beyond. We take special note of data taxonomy and employee access in addition to overall data governance.
Combine these findings with insights into how the organization uses all the other tools in their tech stack to find more accurate insights faster.
Prioritize data governance gaps that could deliver major results with improvements.

The outcome of this process is a roadmap for the client’s data governance strategy: how to execute it, what results you’re aiming for, and benchmarks for success. As Mammoth Growth moves through each of these steps with our clients, we take an agile approach to deliver business value as quickly as possible. Mammoth Growth focuses on the quality of each client’s user data and their data governance policies as this enables faster, more accurate analysis. In addition, this makes the client’s preferred technology solutions more accessible to a wide range of stakeholders on their team.

Data governance can be challenging and complex, which might also explain why it’s so often de-prioritized. Yet while many companies cling to the status quo, other organizations recognize the value of data governance as they pursue their plans for digital transformation. As we mentioned earlier, you’re not in this alone. Contact one of our experts today, and let’s talk about your data governance goals.