beawolf - Fotolia

Tip

How to make a self-service BI tools deployment less painful

Self-service BI can be a big change for everyone in an organization. Expert Rick Sherman offers three principles to keep in mind that could make things easier.

Why does self-service BI have to hurt so much? In theory, it should be the answer to all your problems. But so many enterprises end up disappointed. They're overwhelmed with data, yet suffering from a shortage of useful information. They can't gain insights. And it's causing data chaos and anarchy.

When enterprises embark on their first endeavor with self-service BI tools, they commonly find themselves in one of the following scenarios:

  1. They evaluate BI tools and purchase one for an enterprise-wide deployment, but most business users revert to using spreadsheets quickly.
  2. Small groups of business users acquire self-service BI tools on their own and enjoy success; however, everyone else continues to use their existing reports or spreadsheets.
  3. Many business users throughout the enterprise use a self-service BI tool; however, they operate in silos. As a result, each business unit generates its own numbers, which differ from those in the other business units.

In scenarios one and two, people continue using spreadsheets, thus creating data shadow systems, also called spreadmarts. In scenario three, the self-service BI tools generate shadow systems. In all the scenarios, it's complete chaos and anarchy.

Select the right self-service tool

When evaluating self-service BI tools, the goal is to select the best one for the enterprise, but there's often a bias toward the one-size-fits-all tool. The flaw in this approach is that business users aren't all the same. They have different analytical needs and analytical skills. Yet, tools are selected with the assumption that every businessperson is a power user who wants to create her own dashboards, data visualizations and reports. It's quite the opposite: Most business users need to consume data to do their jobs, not be consumed by it.

There are four types of analytics: descriptive, diagnostic, predictive and prescriptive. And there are different business analytics roles: casual information consumer, business analyst, data analyst, data engineer and data scientist.

The overwhelming analytical need is descriptive analysis, which enables a business user to examine what has happened or what is trending in the business. This is the type of analysis that casual information consumers need. These users don't want to create a dashboard but would like to be able to filter and drill down into data from existing dashboards, visualizations or reports. Putting a data discovery tool with a blank screen in front of them is the surest way to overwhelm them and send them back to the spreadsheets.

Before choosing a self-service BI tool, analyze your customer segmentation in terms of analytical needs and skills.

There is a time and a place for data discovery tools. Their sweet spot for self-service BI is diagnostic analysis performed by business or data analysts. Their job is to analyze data and, if enabled, create dashboards, visualizations and reports for themselves and others. Predictive and prescriptive analytics is the purview of data engineers and data scientists who will need not only the self-service BI tools, but also statistical applications and data preparation tools.

Before choosing a self-service BI tool, analyze your customer segmentation in terms of analytical needs and skills. With that segmentation, you should be able to set up the right tools and analytical environments to enable and encourage widespread use of self-service BI -- and avoid buying a tool no one ends up using.

Shift business and IT roles

Traditionally, businesspeople use reports and dashboards embedded in their enterprise applications or custom BI applications developed by IT. Since the embedded reports are never enough and don't access the enterprise's multiple applications, IT groups find themselves in an ever-increasing backlog of BI dashboards. This backlog is frustrating and makes the business susceptible to the marketing hype of self-service BI tools vendors, whose sales pitches say "no IT needed." This shifts from IT doing all the BI work to the business doing all the BI work; neither benefits the business.

The new formula that enables a successful self-service deployment is business and IT establishing a partnership. IT's roles are to create and manage the data backbone of enterprise applications and a data warehouse and to create the descriptive analysis that the casual information consumers need. Business and data analysts' roles are to create and publish the diagnostic analysis, utilizing IT's data backbone. Data engineers and data scientists will also use the data backbone when they create and publish predictive and prescriptive models.

Implement data governance

Finally, no matter how self-service BI tools are built, dashboards and reports that are shared across business groups absolutely require data governance. Business metrics or key performance indicators (KPIs) are created based on business rules that select data, filter data and apply business algorithms. If an enterprise wants to rise above the data silos, it needs to reach agreement on KPI definitions and then apply them in dashboards and reports. This agreement is achieved when data governance is established and becomes part of an enterprise's accepted business practices.

Successful, painless self-service BI requires that an enterprise follow the three principles discussed here. It's not the tool that leads to self-service BI success, but rather proper focus on policy, process and politics.

Dig Deeper on Business intelligence technology

Data Management
SearchAWS
Content Management
SearchOracle
SearchSAP
Close