DIY BI: Implementing Self-Service BI with Claudia Imhoff, PhD
Another post from tireless software engineer David Dunham (@radiofreelunch) at the Tableau Customer Conference, on self service BI presented by Claudia Imhoff.
Self-service BI ≠ self-sufficiency! A TDWI survey on self-service BI said #the 1 driver of self-service was constantly changing business needs.
Benefits to IT of providing self-service include:
- In many cases IT can step aside as intermediary
- IT can focus on more value-added activities (not just grinding out reports)
- IT becomes more of a partner rather than a roadblock (and yes, Claudia has actually seen this happen)
Claudia cited a case study: a great self-service project rolled out to doctors and nurses FAILED. The audience didn't want to do the analytics, they just wanted to consume information not produce it.
There are 4 groups of information workers:
- Information Producers: "power business users" and business analysts
- Information Consumers: Task-oriented business users (general public, customers, partners, suppliers, employees)
- Information Collaborators: Improve the knowledge content and expertise of an organization and other information workers (especially information consumers). These are often new & motivated information workers, subject matter experts, or researchers. The problem with this group is they can't make a decision by themselves.
- BI/Data Warehouse Builders: Traditionally responsible for building DW and/or BI solutions.
"We need to build the buffet table that has all these components in it, so the BI user can stand there with their tray and pick and choose," said Claudia.
Guiding principles of self-service BI:
- Make BI tools easy to use
- Make BI results easy to consume & enhance
- Make DW solutions fast to deploy & easy to manage
- Make it easy to access data
On technology solutions, Claudia said that no single technology has all the necessary attributes.
The #1 inhibitor to self service, according to mostly IT responders to TDWI survey, was lack of skills in business users. Also cited was lack of data quality, control, and governance. "No demand" was NOT an inhibitor!
Governance does have a role in self-service BI. This is a big issue. It's different than in the past though.IT is under the misconception that it has to control the information assets that the business people use. They got blamed for bad data, security breaches, compliance problems — even though these weren't their fault. Not all data has to be governed, but some does. IT needs monitoring and oversight capability, not control.
The Line of Business may need to take on some of the previously traditional central IT roles such as security. Governance can be aided by customizable BI components, such as parameterized reports, report templates, or stored analyses.
Claudia cited these best practices:
- Don't assume that installing easy-to-use BI tools make your environment self-service BI enabled.
- IT needs to monitor BI self-service.
- Support collaborative business intelligence.
- Don't give information workers too much responsibility: generating information and reports isn't part of their job.
- Understand information requirements of information workers and provide appropriate tools/reports/dashboards.
- Create a starter set of standard BI components.
- Establish a governance committee.
- Allow the data warehouse system to be used with other types of source data.
- Buffer less experienced information workers from the complexities of the BI envirnoment. This is probably where Tableau shines the most.
- Watch your costs.