Topic: Self-Service BI
There's a lot of talk these days about the importance of self-service business intelligence (BI), but little discussion about the downsides. Unfortunately, I've seen self-service BI initiatives go completely awry.
Typically, a small percentage of power users employ the new-fangled tools to create tens of thousands of reports-many of which contain conflicting or inaccurate data-and this makes it harder for casual users to find the right report or trust its contents. Ironically, many turn to IT to create a custom report, which expands the report backlog that self-service BI was supposed to eliminate. In addition, many power users use the self-service tools to query large data sets that they export to Excel for analysis. These runaway queries erode query performance significantly, making the BI environment even less inviting for casual users.
So despite its allure, self-service BI can cause overall BI usage to plummet, report backlogs to grow, query performance to diminish, and information inconsistency to increase. To avoid such unintended consequences, here are five ways to implement self-service BI properly:
1. Deploy the right type of self-service BI
Power users need ad hoc report creation while casual users need ad hoc report navigation. There is a big difference. Powers users need to create reports from scratch, often accessing data from multiple locations. Casual users, on the other hand, simply want a report or dashboard that contains their top ten performance metrics and makes it easy to drill down into details, if desired.
2. Don't abdicate responsibility for creating standard reports
To ensure that business units exploit the full value of the data warehousing resource, BI teams should take an active role in creating standard reports or dashboards for each group. This requires BI teams to get more engaged with the business, not less, which is usually what happens when an organization deploys self-service BI tools. It is not enough to create the data warehouse; if you want widespread adoption, you need to deliver the end-to-end solution.
3. Create a network of super users
Super users are technically-inclined power users in each department who function as extensions to the corporate BI team. Their job is to create ad hoc reports around the edges of the standard reports, suggest enhancements to the standard reports, and work with the BI Competency Center to define standards, select tools, and create a BI roadmap. Super users are the eyes and ears of the corporate BI team within each business group and critical to its success.
4. Liberate and manage power users
Power users can make or break a BI program. To get them on your side, give power users access to the best tools, unlimited access to data, and a sandbox within the data warehouse in which to conduct their ad hoc explorations. To ensure that they don't proliferate spreadmarts, limit their ability to publish reports back to the server and establish a review board of power users (i.e. their peers) to examine new requests for published reports. This two-pronged approach will both empower power users to generate critical insights while minimizing their ability to foment report chaos
5. Educate executives to use standard, certified reports
For self-service BI to really work, executives need to make a companywide declaration and commitment to use standard, certified reports or dashboards when making decisions. Of course, there will always be exceptions, like when standard reports don't contain the data needed to address issues required for a decision. But a corporate edict backed up by action makes all the difference.
Following these five steps will help your BI program achieve the holy grail of self-service BI.
Wayne Eckerson is director of TDWI Research at The Data Warehousing Institute. He can be reached at weckerson@tdwi.org.
