Resolving the Tug-of-War Between Self-Service and Governed Analytics-Part 1

By guest author, Dan Vlamis-President,Vlamis Solutions

To get the biggest bang for your analytics bucks, you want maximum flexibility when reporting and analyzing your data, right? That sounds good if you subscribe to an ad hoc, self-service approach to data. But is that the best way? Shouldn't IT manage the data to ensure the greatest security and reliability? Isn't the governed approach the way to go? Self-service vs. governed analytics. There's an undeniable tension between these two approaches, between IT and BI users, between the need for reliability and the desire for agility. Too often, this tug of war is depicted as a winner take all contest - either all analytics must be governed, or all analytics must be self-service. This doesn't have to be a zero-sum game. In our years of experience working with Oracle Analytics, Vlamis Software has found ways to give you the best of both worlds. In Part 1 of this two-part blog, I will look at these two ways of managing your analytics data, and in Part 2 offer ideas for how you can maximize your return with a blend of both approaches.

The Governed Analytics Approach

An analytics system often needs metadata that defines logical and physical schemas, physical-to-logical mappings, as well as how all the data interrelates. In an Oracle Analytics system, this information is stored in a metadata repository (referred to as the “RPD”). Using this repository, IT managers register all data sources, define for the business intelligence tool how the data interrelates, and then control which users have access to that data.

Having analytics governed and controlled by IT provides several benefits:

• Security and integrity. IT can control which users have access to what data, ensuring greater security across the system and thus greater confidence in the data.
• Consistency of navigation. IT can define relationships centrally that enable users to consistently navigate between related sets of data more easily.
• Business alignment. Standardization of business rules aligns decision-makers, reduces time spent arguing over data, and aids coherent understanding of the business.
• Quality and accountability. Data governance makes it easier to hold data stewards accountable for the accuracy of their data.
• Query redirection. The ability to seamlessly retrieve summary data from a summary table and data from a detailed fact table using one logical model offers better performance for end-users. IT has access to tools (such as an RPD) that makes this possible.

The Downside of Centralized Analytics

In my experience, the past 10 years have seen BI users demanding more and more control over their analytics – they want better, more customized answers that they can produce more quickly. The limitations of centralized analytics from their perspective are clear:

• Centralized analytics is incomplete. Users want data that cannot be found in the IT-controlled repository. Sometimes the data does not even belong to the organization, but users need it for the benefits it could bring to the analysis. (For example, they may want to normalize sales in their district based on population count from U.S. Census data.)
• Data is not structured the way people want. The structure of data in IT-controlled repositories can limit how it is viewed – the drill path through the data hierarchy. (For example, IT may set up the hierarchy as CATEGORY drills to SUB-CATEGORY drills to PRODUCT LINE drills to BRAND drills to DETAIL, while users want to drill from CATEGORY to BRAND to PRODUCT LINE to DETAIL.)
• IT can become a bottleneck. The way that centralized analytics is stored and defined can be too rigid and take too long to change.
• Calculated measures aren’t always set up the way analysts want. IT may have set up calculated measures that do not meet the needs of analysts.

The Self-Service Data Approach

In a self-service approach to data, users circumvent the limitations of an IT-controlled repository by exporting the data they need and importing it into other easy-to-use tools such as Tableau®, Excel, or Microsoft PowerBI®. They gather data from auxiliary databases, then join that with the exported data to create their custom analyses.

In contrast to the limitations of governed analytics, the self-service approach offers the following advantages:

• Control in the hands of the business users. Putting the data and tools into users' hands ensures it is closer to the people who need to use that data in their decision-making.
• Agility and fast answers.  With easy-to-use tools, business users can quickly and easily identify trends. They can bring in external data and combine it with corporate data for additional insights. By manipulating the data themselves, they can explore new areas with a quick turnaround and fast return on value.
• Ease of collaboration. People can share analyses among a few individuals and collaborate. They can create a sandbox for exploring different ideas that are not worthy of a larger IT investment.

The Downside of Self-Service Data

Although the self-service approach seems to provide maximum flexibility when reporting and analyzing data, it also introduces some clear limitations:

• Conflicting data – no reliable source of truth. Different analysts could export different data and come up with different results. The CEO hears revenue is $28,000,000 from one person and $23,000,000 from another person. How does the CEO know whose numbers to believe?
• Wrong conclusions from inappropriate use of data. A lack of understanding of underlying data structures makes it oh-so-easy to create reports that are simply wrong. It is very easy to double-count transactions and develop inaccurate reports.
• Extra work and maintenance. People create unnecessary burdens for themselves. Data exported one month has to be refreshed the next month. In the meantime, data manipulations must be redone each month.
• Support issues. As one-off databases are shared and more people rely on them and encounter issues with them, they turn to IT for help, but IT doesn’t know about these one-off creations. These user-created solutions are problematic because they lack the design and planning that make support practical.
• Performance nightmares. As user-created data procedures grow in complexity, they can bog down scarce IT resources and create performance bottlenecks that become critical to the business.
• Inefficiency and indispensable individuals. Every user is working in their data silo. Data procedures are inefficient and haphazard and often rely on a single individual. A lack of standards, documentation, and best practices makes it worse.
• Exposure to security breaches. It’s hard enough to protect data assets when they are centrally located. Once users create their data repositories, it’s next to impossible to keep that data secure.
• Hidden costs. These self-service solutions often become indispensable without anybody even realizing it. In the long run, maintaining self-service solutions often costs organizations more than their short-term benefits justify.

Now that you see the inherent conflict you may be wondering, how can you reconcile these two approaches? Read part 2 to find out how to resolve the tug-of-war and create a win-win situation for your organization.

Learn more about the Oracle analytics platform. To learn more about Vlamis and how they can help with Oracle Analytics projects check here. Follow us on Twitter@OracleAnalytics and LinkedIn.