When I talk to people about adding self-service BI to their company's environment I generally get a list of reasons why it won't work. Some things I commonly hear:
So what is a forward thinking BI implementer to do? Well, Intel just went out and did it, blowing through the supposed obstacles. Eduardo Gamez of Intel's Technology Manufacturing Engineering (TME) group interviewed business folks to find those who were motivated for change, found a great pilot project with committed employees, and drove the process forward. They put a "sandbox" environment up for the business to use and came up with a plan for monitoring the sandbox activity to find models and reports worth adding to their priority queue for enterprise BI projects. The business creates their own data models and their own reports for both high and low priority items. IT provides the infrastructure and training including products like Analysis Services, PowerPivot, Power View, SharePoint, Excel, SQL Server, and various data sources. The self-service models and reports are useful to the business - they reduce manual efforts, give them the reports they want much faster, and ultimately drive better, more agile business decisions. If a model isn't quite right after the first try, they can quickly modify it. The same models and reports are useful to IT - they are very refined and complete requirements docs that shorten the time to higher quality enterprise models and reports, they free up IT resources to build a more robust infrastructure and allow IT to concentrate on projects that require specialized IT knowledge. Everyone wins with a shorter time to decision, higher quality decisions, and a significant impact on the bottom line.
Learn more about how Intel TME is implementing self-service BI:
Eduardo (email@example.com) and I (firstname.lastname@example.org or @SQLCindy) are happy to talk to you about Self-Service BI - let us know what you need to know!
Self-Service BI might be a relatively new concept though the requirements, solutions and challenges associated with it move 10-15 years back. I suppose that in many organizations are a few non-IT people with enough understanding of the data and data models, who have the technical skills to build a report or a personal solution with the help of MS Access, MS Access or a user report builder capabilities existing in various BI tools, as long the data are in one or more repositories available for consumption. In fact, because of the unavailability of standard BI solutions or BI developers, some of my customers were forced to build their own solutions with generally available and departmental data, resulting ad-hoc personal or departmental solutions of whose existence IT was quite often not aware. Over time the use of such solutions degenerated, becoming a nightmare for IT and business as well, as the infrastructure was more difficult to support, in organization being reported different numbers for the same business problem.
The adoption of standardized BI solutions helped reducing the number of non-standard solutions, allowing the alignment of reports with organizational and departmental goals, and to provide a unique and in theory more reliable view over the business. Despite this, there are still people who want to analyze the data by themselves, look at the same data from different perspectives. As BI consultant I have to say that an organization needs also this kind of people and a self-service oriented infrastructure, and maybe that’s what it stressed by self-service BI nowadays.
The main difference between 10-15 years ago and today is that nowadays tools and architectures are more capable of addressing self-service BI requirements directly. On the other side many of the challenges remain the same:
- Picking the right tools and building a flexible infrastructure for self-service requirements
- Making users understand the data, what they have to consider or ignore, being consistent in criteria they use over time, addressing business and data model changes.
- Minimizing the existence of personal solutions of which IT has no idea.
- Minimizing the duplication of effort (e.g. two different people need same or similar solutions and they build the solutions independently).
- Reintegrating the knowledge accumulated by users, while they play with the data, back in the knowledge base/infrastructure of an organization.
Above the childish arguments against self-service BI there is also an entitled argument – how much has an organization to invest in its infrastructure in order to address self-service requirements? For some organizations with an old BI infrastructure, this effort might pay-off, but for organizations that redesigned recently their BI infrastructure without considering self-service requirements; a redesign can prove to be not feasible.