Happy New Year from the DGPO!
On January 8, 2020, Cynthia Parsons and Scott Peachey presented Nationwide Insurance’s Award Winning Data Governance Journey in a DGPO webinar. Nationwide is the Winner of the 2019 DGPO Data Governance Best Practice Award.
Submissions are invited for the 2020 Award. The deadline is February 28, 2020. The award is given to practitioners within a customer organization, corporation or government agency in recognition of an outstanding data governance program. Software vendors and consulting companies who provide consulting services, tools, or other software products are not eligible.
Submissions will be judged on their response to the following:
• Business need(s) being addressed
• Approach used to design, develop and implement the program
• Scope and depth of the Data Governance Program as well as program longevity
• Data governance defined roles and responsibilities
• Communications and marketing approaches
• Policies, procedures and processes established
• Program metrics and business value provided
For details please visit the DGPO website.
The DGPO is pleased to have Malcolm Chisholm from Data Millennium as our guest columnist this month.
——————————————-
Dawn of Self-Service BI Governance
Malcolm Chisholm
It seems that Self-Service BI is finally taking off. After decades of hope, hype, over-promising, and disappointment, technology that businesspeople can use is finally becoming available, albeit with some level of data literacy. But before we cheer for this success a little caution is in order. Why? In short, because Self-Service BI needs more than just technology to be successful. It needs governance.
There is no doubt that the emergence of BI technologies that can be widely used by businesspeople is very exciting. It will truly democratize data when anyone in the enterprise can access data and utilize it for reporting purposes. This does not mean that everyone is going to become a data scientist and create statistical models of their little part of the enterprise. More likely in it means that:
- A business unit can create its own operational reports in order to run processes efficiently and effectively. In particular, the business unit can quickly respond to minor changes in the business, which often occur more quickly than is often thought
- A business unit can create its own analytical reports, mostly through filtering, classifying, transforming, sorting, and aggregating data. This is more logic than specialized mathematics, but it is possible to derive insights in this way.
All this can be achieved without the need to go to IT. Involving IT in report development usually involves:
- The effort of trying to communicate business requirements to IT report developers
- The chance that business requirements will not be communicated correctly to, or understood correctly by, IT. The result can then be a wasted effort.
- The prioritization of the report development request into an often obscurely managed IT backlog, where there is little clarity on when the request will be executed.
Conditions for Adoption of Self-Service BI
So, if Self-Service BI is so beneficial, why has it not happened already? Indeed, ever since the introduction of PCs in the mid-1980’s there have been prophecies that it is just around the corner. A critical factor has been the lack of technologies that can be mastered by business users with a reasonable amount of effort. As noted above, this is changing today as more and more user-friendly tools become available. These technologies do not merely have more intuitive user interfaces – they actually perform a lot of data manipulation tasks that had to be done manually in the past, and which required greater technical knowhow.
It is therefore correct to say that lack of adequate technology has been a barrier to Self-Service BI in the past, but that is no longer the case. Does that mean that we just have to introduce the technology and the problem is solved?
There does seem to be a drive to introducing technology. It is difficult to blame the technology vendors, and indeed their tools are really good. And to be fair, in discussing Self-Service BI in the business, it is often pointed out that businesspeople are a lot more savvy about data than in the past. So it does seem that conditions are favorable for adoption. IT too has a stake in this game because they often want to get out of dealing with uncontrollable requests for report creation, especially ad hoc reports.
What Could Go Wrong?
At this point it would seem there is nothing to worry about. But very often new technologies that process data pay little attention to data content, and actually enable unanticipated data-related problems once they are introduced. Let’s look at a couple of examples.
ETL (extract-transform-and-load) technologies enables data to be taken from different source systems and integrated in environments like data warehouses. Data quality problems that were not apparent in the source systems (because they were controlled by program logic, or irrelevant to the use cases there) suddenly became apparent. More recently, eCommerce technologies have changed our entire economy, but now everyone realizes that these have enabled serious problems with data privacy.
When it comes to Self-Service BI it seems that many people think that the new technologies in combination with today’s more data literate workforce is all that is needed for success. This is not likely to be the case. Once businesspeople get into the weeds of Self-Service BI, they may find themselves asking questions such as:
- How do I request getting the new Self-Service BI technology, and how is it set up?
- Am I expected to do the technical administration of my own Self-Service BI technology environment?
- I want to develop a new report. But is there already one that has been developed that is close to what I need?
- This new report I have developed is really good. My boss wants it productionalized. How is that done?
- I understand how to get some of the data I need, but not all of it. How do I get access to all the data I need in a form I can use?
- I want to segment customer data by various demographics. But does this violate any data privacy policies or regulations?
- Am I supposed to develop reports on Excel and Access end-user-computing data sources that I maintain on my desktop?
- I have decided to use the label “Customer Lifetime Value” on my new report. I wonder if this means the same to me as it does to other people.
- If I develop a report, how does it get tested? After all, I will not see my own erroneous assumptions.
The Need for Self-Service BI Governance
One disconcerting factor that occurs when new technologies are introduced is that of “quick wins”. There is always a set of requirements that are fairly urgent and not too difficult to satisfy. The new technology is brought to bear on these, and they are dealt with successfully. At this point, say 6 months in, it seems that everything has been proven to work. But as time goes by, the more difficult use cases arise, and there is the issue of managing what has already been created. And questions like those listed above become more important. This is where the need for governance for Self-Service BI becomes apparent.
The best practice is always going to be to anticipate the need for Self-Service BI Governance, figure out what it means for the enterprise, and come up with a plan that is going to address what are likely to be the most important needs. Every enterprise is likely to be different, but there are also likely to be patterns of governance approaches that work. The Data Governance community should think through what these patterns are and encourage adoption of the successful ones.
——————————————-
About the Author: Malcolm Chisholm is an independent consultant over 25 years experience in data management and has worked in a variety of sectors. Malcolm is a well-known presenter at conferences in the US and Europe, writes columns in trade journals, and has authored three books. In 2011, Malcolm was presented with the prestigious DAMA International Professional Achievement Award for contributions to Master Data Management. He can be contacted at mchisholm@datamillennium.com or via the website www.datamillennium.com.
