This is part 1 of a multi-part series on how to succeed with self-service analytics. 

Self-Service Pitfalls

Self-service analytics seems like a win-win proposition for business and IT, but it can quickly go awry without sufficient governance.

Here’s what happens: newly empowered data analysts create a multitude of conflicting reports and dashboards with unique business logic and data. As a result, the data doesn’t add up and business users no longer know which report to use. Executives get frustrated when they ask simple questions, like “How many customers do we have?” and receive conflicting answers.

As an organization becomes awash in conflicting reports, data analysts refuse to trust any except those they create themselves. In addition, business users revert to asking IT to create custom reports for them, or worse yet, they stop using data altogether to make decisions. Instead of liberating business users, ungoverned self-service analytics creates data chaos—a proverbial Tower of Babel where people talk, but few communicate.

Rather than use the term “self-service analytics”, we need to start talking about “governed self-service”. Proper governance balances user empowerment with standards to ensure data and business alignment—something that every organization (and ultimately every individual) wants and needs.

Self-Service Paradoxes 

Self-service analytics involves reconciling polar forces pitting freedom against control. This dynamic is played out in many areas of business and technology: speed versus standards; innovation versus operations; agility versus architecture; and departmental needs versus corporate interests.

Organizations need to balance these dual and dueling forces to deliver an optimal self-service environment, one that balances user empowerment (self-service) against business controls (governance). You can’t have one side of the equation without the other; you need both at the same time to counterbalance the excesses of each. Just as self-service without governance creates data chaos, governance without self-service creates an overly bureaucratic and oppressive operating environment.

The tension between freedom and control creates a lot of interesting paradoxes that companies experience when trying to implement self-service analytics:

Paradox #1: Self-Service Requires a Lot of Standards 

At face value, one might think that self-service allows users to fulfill their own data analytics needs any way they want. But the opposite is true. Without standards, business users have to work harder and longer to get what they want with greater risk of failure.

For example, we all take for granted that we can hop in our cars, drive to the grocery store, and return safely, taking the most direct and efficient route possible. We think we have autonomy and total freedom, but we don’t. The illusion of freedom is created by our conscious (or unconscious) adherence to a multitude of rules, regulations, and standards that govern how we drive: we drive on the right side of the road; we stop at red lights and slow down at yellow ones; we obey the speed limit (mostly); we get a driver’s license and register our cars annually. Without these and hundreds of other rules and regulations, the roads would be chaotic and dangerous, and we’d risk our lives every trip to the grocery store.

Standards are the foundation upon which self-service analytics is built.

In the same way, self-service analytics requires a multitude of standards or it devolves into data chaos, described above. These standards include how to define and calculate metrics, how to store and access data, how to query and publish reports, and how to govern individual privacy and security. Standards are the foundation upon which self-service analytics is built.

Paradox #2. Self-Service Requires a Lot of Tailoring 

Self-service means different things to different people. An old school executive thinks self-service involves viewing a report online instead of on paper; a business manager believes it is drilling into a dashboard and saving the view as a “report” that they can view next week with fresh data; a data analyst thinks it’s the ability to find, combine, analyze, and visualize data from any source that makes sense to answer a business question. And each is right.

Self-service places a premium on classifying business users based on their information appetites and then providing relevant business views….

For business intelligence (BI) teams, this places a premium on classifying business users based on their information appetites and then providing relevant business views, analytic functionality, training, support, and facilitation. It also means tailoring permissions so users only see the data, metrics, and analytic functionality relevant to their jobs. Permissions clean the visual and analytical palette, giving users only what they need to do their jobs quickly and effectively.

Paradox #3. Self-Service Requires a Lot of Hand Holding 

Self-service implies that a task is easy enough for almost anybody to do without assistance, like pumping your own gas at the service station. This is true, but becoming self-sufficient takes time, especially for things that are more complex than pumping gas, such as creating custom dashboards and analytic views.

Most BI leaders schedule one-on-one meetings with business executives to help them become proficient with analytic tools and applications. This is expected for top executives and most rank-and-file business managers, but it’s equally true for hard-core data analysts. Just because someone has a “data analyst” title doesn’t mean they know how to use a self-service tool to explore the organization’s data and generate insights that matter. Most business people, data analysts included, need a helping hand to understand the nuances of the company’s data, its visual design standards, and how to use a new tool.

Since a BI leader can’t provide one-on-one assistance to everyone in the organization, other BI team members need to chip in. BI relationship managers should schedule one-on-one time with business unit leaders and managers to get them comfortable with analytic tools and output. BI developers, if not embedded in the business, should make themselves available to data analysts and data scientists via office hours or data labs, where analysts can work through problems with specific tools and data sets.

Paradox #4. Self-Service Requires More Collaboration with IT, not Less. 

Many business people love the idea of self-service analytics because it promises to liberate them from the shackles of the IT department and its perpetual backlog. With self-service, business users can create their own reports and dashboards without having to wait for the IT department. But the reality is that self-service requires the business to spend more time with IT, not less.

Self-service analytics doesn’t work unless the business steps up and owns the data analytics program.

Self-service analytics doesn’t work unless the business steps up and owns the data analytics program. Ownership takes many forms: a data analytics council that reviews and approves strategy and tactics; a data governance process that defines, documents, and manages key data elements; a data quality process that monitors core data elements for defects; a report governance process that certifies reports and the analysts who create them; a permission structure that defines who can access which data; and an agile process that prioritizes projects and capabilities for BI developers.

In essence, the business sets the rules and standards that the IT team implements.

In essence, the business sets the rules and standards that the IT team implements. If the business doesn’t assume ownership and responsibility for data analytics resources, processes, and strategies, then it will never enjoy a robust data analytics infrastructure with easy access to clean, integrated, timely, and relevant data. The business needs to work hand-in-hand with the IT department to make self-service analytics work.

Paradox #5. The Need for Self-Service is Inversely Proportional to the Quality of Core Business Dashboards

Although many business people clamor for self-service analytics, in reality, most don’t want it.

Although many business people clamor for self-service analytics, in reality, most don’t want it. How can that be? Data is so locked up in many organizations that some business people spend nights and weekends finding data and stitching it together using Excel or Tableau to create a quarterly or annual report that they are required to deliver to business leaders.

Not surprisingly, these individuals are desperate for self-service tools that can make this process easier. Yet, if they didn’t have to create the reports at all, they would be ecstatic. They have better things to do than spend their precious free time creating business views of data. Their job is to lead and make decisions, not slice, dice, and prepare data like a data analyst.

What business managers really want is a well-designed interactive report or dashboard that generates the views they want or can be easily modified, if needed.

What business managers really want is a well-designed interactive report or dashboard that generates the views they want and can be easily modified. They don’t want to create reports from scratch; they simply want the BI team to understand their needs and create pre-ordained business views that largely align with their needs. Ideally, these dashboards are action-oriented and predictive in nature, showing business people's performance to plan and forecast and suggestions for the next steps.

Our rule of thumb is that each business function needs one or more well-designed, interactive dashboards that address 80% of the questions that managers and workers typically ask. Since these dashboards can be complex, often it’s the IT department that creates them, but skilled departmental analysts and report developers can design and build these dashboards as well. In fact, if the IT department has limited resources, a self-service initiative might be the only way these standard dashboards get created.

A well-designed departmental dashboard will obviate much of the demand for self-service.

The point is that organizations should spend more time implementing robust, departmental dashboards than pursuing a self-service strategy for its own sake. A well-designed departmental dashboard will obviate much of the demand for self-service.

Summary 

Self-service analytics seems like a win-win for business and IT: the business gets what it wants, how it wants it, and when it wants it, and the IT department gets to focus on more value-added tasks. The reality is that self-service analytics can go awry quickly, creating data chaos with hundreds of conflicting reports and dashboards.

To avoid data chaos, organizations need to implement “governed self-service” where business and IT collaborate to establish standards that define data elements and the proper use of reports and dashboards, establish processes for ensuring data quality, and develop controls for accessing and safeguarding data, among other things.

The next article in this series will discuss how to govern reports and dashboards to foster trust in the data and ensure organizational alignment. Stay tuned!

Read - How to Succeed with Self-Service Analytics, Part II: Know Thy Customer