April 10, 2023

How to Enable Governed Self-Service Analytics

By Spencer Hamilton

I’ve never had small children in my house, but my parents can in fact confirm that I was once a small child. They can also confirm that as a small child, emboldened by my creative ambitions, I once took a big blue crayon to the largest blank canvas my young eyes had ever seen—my bedroom wall. 

While I may not have technically known it was against the rules, the wall was absolutely not what my parents had in mind when they gave me that big blue crayon.

Moral of the story? My parents, no doubt, wanted me to express my creativity in a productive way, ideally without property damage. 

They needed to create parameters in which I could do just that; in a sense, they needed to govern my expression. 

In almost every modern organization, data and its respective analytics tools serve to be that big blue crayon. Users across the organization need that big blue crayon to make decisions every day, answer questions about the business, or drive changes based on data. 

In a perfect world, IT doesn’t need to worry about users ‘writing all over the pristine walls’ of their data warehousing solutions. Users could have unfettered access to anything they need from a data standpoint whenever they need it. 

The whole basis of self-service analytics is to enable the business and remove technical roadblocks, but this is easier said than done. 

In this blog, we’ll dive into the concept of governed self-service analytics and how to go about enabling it in your organization.

What is Governed Self-Service Analytics?

To define the concept of governed self-service analytics, it’s best to take a step back and think about the different analytics operating models.

If you’re reading this blog, my guess is that you’re on either end of the spectrum: completely decentralized and the business is allowed to draw all over the walls, or centralized and the business is just begging for a blue crayon instead of the same red and yellow ones they keep getting. 

In practice, it’s probably more realistic that your organization is sitting somewhere in between, either looking to give more autonomy to the business or taking a little more control over what data the business has access to. 

This decision-making process is exactly what governed self-service analytics is all about. A truly governed self-service analytics model puts data modeling responsibilities in the hands of IT and report generation and analysis in the hands of business users who will actually be doing the analysis.

Business users build reports on an IT-owned and IT-created data model that is focused on reporting solutions. Users can request new attributes or new columns, and IT owns the process of enhancing the dataset. 

This model makes sense in a number of ways, the first being that IT is typically much more technically capable of ensuring modeling standards are met and consistent, as well as making sure security considerations are met. 

The second, and in my opinion most important, is that the business is able to build exactly what they want the way they want it. Try telling a toddler who wants to draw all over the walls that you will just draw on the walls for them instead. My guess is it will be met with a tantrum of monumental proportions. 

While this may be the first time you think of your business users as mature, well-dressed toddlers, I assure you the quickest way for your self-service analytics initiative to fail is to make the business feel like they aren’t in control of their data products.

This is why the fully centralized analytics operating model often falls flat and fails to deliver a tangible return on investment in the short term. 

These organizations often struggle with the adoption and ownership of data products, which leads them to start working towards a federated model. How do they do that, exactly? Let’s dive in.

How to Enable Governed Self-Service Analytics

Now that we’ve defined the analytics operating models and the concept of governed self-service analytics, let’s look at the implementation.

Like any other analytics initiative, there are several layers to doing this successfully. Similar to creating an analytics strategy, enabling governance in self-service is going to be extremely people and process driven. 

The right framework has to be able to support data discovery, security, quality, and change management in a streamlined format where business users can get what they need quickly and accomplish their tasks effectively. Your entire organization needs to understand how to work with data, including those who haven’t traditionally incorporated it into their day-to-day processes.

We can handle how they will work with data in our training programs, but building a data-literate culture should inform users of what is possible. Knowing that shifting mindsets and behaviors across the organization needs to be a parallel component of your data governance framework will make these subsequent steps successful.

Step 1: Identify and Define User Types

The first step in enabling this governance framework should be to identify and define the different users in your organization. This should include the IT resources that will be responsible for building out standardized data models from the raw sources as well as all personas of business users: viewers, casual developers, and power users. 

In addition to the developers on the IT team, you should also consider including support analysts that can handle access provisioning or ticket escalations. These folks often hold ‘IT Business Analyst’ roles and can assist with backlog generation from their understanding of the processes and questions of their respective business units. 

Lastly, a reporting center of excellence (CoE) or team of design experts should be created to standardize the format of all corporate reporting solutions as well as consumption methods for the data models. This team likely will consist of experienced IT analysts and/or more technical business users. 

Looking back at our users and personas, we should see a pretty even split between business and IT users. Almost like we meant to do that! Now let’s talk about the process.

Step 2: Define Governed Self-Service Process

Ultimately, the process we create now should put the pieces of our team we generated in step one all together. 

We need a process that captures the people involved in each step of self-service, from reporting requirements all the way to report deployment. Depending on the level of maturity of your organization, you may or may not have a training program already defined for your power users, or a Center of Excellence focused on establishing a data culture. 

If not, this would be a logical prerequisite and something we love to help with at phData. To learn more about our capability offerings, check out our enablement services! 

Let’s assume we have the capacity and capability to facilitate power user training and start there. 

Step 3: Facilitate Content Creator Training

Invest in training that pairs your analytics tools with how to perform real analytics challenges relevant to your organization. To do this, find an evangelist or several who can help coach employees and act as product owners for the technology. 

Leveraging Data Coach, which combines expert tool training with conceptual Data Fluency training, is a great place to start!

In this training, all of our content creators should be able to define their use cases and associate a value with them. This value could come from time savings from an efficiency gained through the reporting solution or from action taken based on new KPIs being reported. 

Our users should be able to define the value of their use case, but not so that we can gatekeep the solutions that actually get created.

The practice of defining the business value forces content creators to think about what action will be taken when using the report they look to create. Reporting sales for the last three years is great, but what action am I looking to influence with that information?

Knowing the answer to this question solves two problems for us: we can accurately capture ROI for our analytics program, and our users are prepared to create more effective reports that actually influence that next step. 

This training should also familiarize our users with the subsequent steps in our self-service process, so they leave with action items to provide everything IT needs to ensure the data model is configured to meet their needs.

Step 4: Standardize Report Requirements

Okay, so the business has started defining use cases and their value, and now it’s about time for IT to deliver the big blue crayon. In the same way, we don’t want IT to be a roadblock in the self-service process; they’re also an important catalyst in our governed model. 

If IT is slow to produce data models, then the business will grow frustrated with this approach and drift back to the world of Excel exports and bootstrapping data together that way.

In order to make sure IT can quickly and efficiently address new data enhancements, we need a standardized format in which the business provides its requirements. This standardized requirements template should capture things like:

  1. Report Update Frequency
  2. Report audience and permissions considerations
  3. Summary of report page(s) and data
  4. Additional data for enrichment or comparison
  5. Any outlier definitions (business logic)

This template should require sign-off from the creator’s manager or team members to ensure the business logic is defined correctly and that both the creator and audience are on the same page. 

The signed template also provides the MVP parameters for IT to include on their first delivery of the data model. Anything not captured in this document would be considered a future enhancement and not in scope for the current request. 

Step 5: Provide Access to Metadata and Existing Reporting

If you aren’t already familiar with the concept of metadata and a data catalog, this is an important consideration to make. While I’ll argue in the early stages of your self-service journey, it isn’t mission-critical; as your organization becomes more and more data-driven, the need for a data catalog will grow exponentially. 

The more mature and complex a data platform gets, the higher emphasis and detail are needed for data governance. As an alternative opinion, Airbnb actually built an end-to-end solution that allowed users to search all data elements to address self-service adoption woes. 

Regardless of the stage in which you choose to implement a data catalog solution, these types of solutions allow your users to determine if the report they want to create already exists or, better yet, that the data is already modeled and ready to go thanks to another similar use case. 

Access to metadata in any ‘formalized’ model goes a long way in giving the business the transparency they’re often looking for. This model also allows IT to manage what is accessible by the business and creates collaboration between both teams to ensure data quality. 

Step 6: Establish Data-Driven Culture with a Center of Excellence

While this is listed as the sixth step on our list, creating a data-driven culture is really a fundamental piece of your entire analytics initiative’s success. Subsequently, your center of excellence will drive and sustain the principles of making a data-driven culture a reality.

Ideally, the CoE owns training, best practices, communication, and change management, as well as showcasing existing use cases within the organization. It drives the data-literate culture and is arguably the most important and difficult part of your framework. 

This group can kickstart the growth of your analytics initiatives and organically help identify data needs or uncover new use cases. In the context of this process flow, your CoE provides another avenue for content creators to ask questions about their own use cases, existing reports, and collaborate with other users. 

Having executive leadership involved in the center of excellence, even on an informal basis and cadence, is crucial to the success and adoption of both a data governance framework and the data products it fuels.

A data-driven culture has to start all the way at the top, and the behaviors propagate downward. Leaders cultivate the idea that data is no longer an accessory to decision-making or that monthly metric meeting where we throw numbers on a PowerPoint slide, but rather a requirement in everyday decision-making.

The expectation that data is used at every level of the organization drives both adoption and the need for a data governance framework, and one can’t succeed without the other.

Oftentimes we see analytics initiatives lose traction due to IT not having the capacity to fulfill requests that span from not just data modeling but also training or support on the tools themselves.

While IT should still have the capability to assist in this manner, the CoE helps make sure there is enough support for business users to grow and succeed with their use cases.

Step 7: Create a Transparent UAT Process

The last step to ensuring your self-service process is both governed and successful is to create transparent boundaries around the process to migrate finished data products to production. In the first part of user acceptance testing, the CoE should be responsible for signing off on the overall design standards of the solution. 

In addition to design, they review the solution to ensure all best practices are followed from calculations to permissions. Once the product has passed the CoE review, each solution should be reviewed by the IT team to make sure any and all queries are structured for performance and efficiency.

This kind of strict governance often plays an influential role in curbing broad data mapping issues, incorrect data-related decisions, and redundant evaluations. 

It was an opportunity for effective self-serve analytics as the design team approved (and initiated) database disk space additions for forecasted volumes.

Once the UAT process is completed, IT should have a formal, ticketed process to promote the report to Production. IT’s involvement in content promotion helps verify a few things:

  1. The report columns do not have personal identification information
  2. The report satisfies expected performance indicators/timelines 
  3. New report metadata is registered into the system
  4. UAT approval from all business teams who tested the report cube along with the business need and test case documents

In many cases, if the reporting changes are minor or not involving business-critical applications, IT may not need to be included in the promotion to Production.

Like every other step in the process, use your best judgment to strike the right balance between autonomy and control.

Key Components of Self-Service Governance

The primary objective of creating a data governance structure for your self-service initiatives truly is to add value. While it may at first be discouraging to learn you can’t, in fact, draw all over the walls however you see fit, the controls and parameters put in place by data governance ensure several key parameters that will affect every data product in the organization. 

They help make sure that a single source of truth is easily found, used, and corroborated among all users. Then ensure the suitability of processes and access controls, and satisfy compliance requirements.

Lastly, they make sure that the data warehouses are structured in a way that can support the analysis and workload required by the data products. 

The true value of any self-service capability is not just based on a single tool or product but rather the framework built to support the creation, evolution, and consumption of the products.

Your framework should accomplish five key components:

  • Focus on the people
  • Prioritize data quality
  • Confirm data security and compliance
  • Create a community around governance
  • Provide a clear process

When implemented correctly with these components, true governance can drive the adoption of self-service platforms.

When the business has a positive experience with governance, the results speak for themselves and engage new business units and audiences. Through an effective framework, organizations can enable freedom in a controlled environment without compromising data integrity or the end-user experience.

Closing

Whether your organization is keeping all of the big blue crayons in the childproof, locked cabinet, or the walls are nearing a point of irreparable damage, recognizing the need for a robust governance framework is an inevitable step on the road to constructive and productive expression.

Just like when I was a toddler, letting the business know that there will be some structure and process to the way they develop data products won’t completely thwart their desire to have them if done correctly.

When done correctly, it will actually help them create more meaningful and scalable solutions.

Mastering the balance between control and freedom in your organization will ideally be a continuously evolving process rather than set-in-stone steps and objectives.

Like your analytics strategy, it should be agile enough to adapt to the changing needs of the business and IT. 

As mentioned, the more mature your organization’s data platform gets, the more complex data governance will be. 

Data sources that were defined and developed change. Rules don’t stay the same, and it’s important not to meet these requests with a sense of frustration. Instead, it’s better to acknowledge that the requests are a sign of the process working, as business units are submitting requests to your IT teams.

At phData, we know the allure of the big blue crayon firsthand. Quite literally drawing from years of experience, learning, iterating, and doing this for customers at every stage of their life cycle, we’ve built a set of tools, processes, reference architectures, and a team that is ready to help you get on the right path towards enabling governed self-service analytics within your organization.

Data Coach is our premium analytics training program with one-on-one coaching from renowned experts.

Accelerate and automate your data projects with the phData Toolkit