I read the blog "Move your ass" by my colleague David Vonka and said to myself, "That describes a principle I totally agree with." However, I felt it missed something, clear instructions on how to actually implement the successful operation of a BI knowledge-management system. In this blog I want to describe the "how".
First, it's necessary to find a company that can deliver knowledge management for the BI department. That's not just someone who can install a Confluence wiki or a Sharepoint portal - you can do this easily enough on your own. Crucially, the contractor deploys a system to manage your BI outputs and convinces your team that this system is worth using. How to choose such a supplier is a separate issue and I will address that in my next blog. For now, let's stick with the "how" of the implementation.
What exactly should be the scope of such a delivery? My recommendation is to start small: choose a business area that your BI supports and cover it completely in the knowledge system. The point is to create a working standard (proof of concept) - a demonstration for you and your superiors how things will look when the knowledge system is rolled out enterprise-wide.
Where to start
Select a subset of your reports, perhaps 10-20 of them, preferably from one business area. Reports have business owners and BI stewards for their operation. These people need to be involved in the proof of concept. It's therefore necessary to choose reports by people willing to cooperate and who have time available. Their superiors decide about their time and so it's essential to sell the idea to management.
If you have a well-selected area of BI with people happy to be involved then it's easier. My experience with a PoC delivering a Semanta-made BI Knowledge System (Company Encyclopaedia) is that about 15 man-days of external consultants and 15 man-days from internal staff is sufficient to create a good benchmark area for a BI knowledge system.
The main benefits of a PoC should be for a business user who has access to the service through a Web browser. For the reports included in the PoC, the user should be able to find the following information:
where to find the report in the reporting platform or on the shared drive (there simply must be a link to the report)
who is responsible for the report and whom therefore to call / write to with questions or issues
brief basic info, what the report is good for
We recommend selecting 2-5 reports that are used by a significant number of staff or personnel and document them further in a richer way. The key is a well-annotated screenshot of the report. It will help users understand the documentation better. Another helpful tool is to create a page that describes these data sources and links them to the reports. The result is a single point were a business user and data analyst can find useful documentation. Plus, one can easily verify whether the business and technical description of the report are in compliance.
Finally, other useful information for users could be added (especially for business users, after all, supporting business is what gets you the budget). This might be:
useful how-tos (how to access the reports or DWH etc.)
a dictionary of terms and abbreviations used
the contacts of people from BI and their profiles
an online monitor of BI systems (is DWH or reporting available?)
news about current projects and other items
The training of the end users must be completed within the above-mentioned 30 man-days. During these training workshops, the end users themselves will start typing information into the knowledge system. There must be a presentation of your results to management. Here you can show the usefulness of the finished proof of concept to their business model and analyze what time allocations will be required for the running of the the new processes and operations, as well as how much it would cost to do this.
Most enterprises have invested tens of millions into DWH and BI platforms. Here I've described, using some of my own hands-on experience, how an investment of just $20-30k can be used to make a fully functional proof-of-concept for a BI knowledge system. One that can become the basis for a BI department's showcase of itself to the company. It offers BI outputs to business users in an understandable, barrier-free and accessible BI e-shop. This is exactly the type of result that leads top management to understand that investment in BI is useful for them.