Probing Business Intelligence Questions to Identify the Right Tools to Use

Recently, a colleague asked me if I had a list of solid probing questions about the state of business
intelligence within an organization. Moreover, because I was at the time part
of our IBM Business Analytics Practice, he wanted to know if I had questions
that specifically related to the IBM stack. What he wanted to do was to use the
question to understand the issues companies were facing in order to help
construct appropriate solutions and bring the right tools to the table.

I must say that I made a rookie mistake,
after 25-years in the industry, and the product focus derailed me. I
immediately started to ponder what questions could I ask about IBM Cognos
versus SAP Business Objects versus Oracle Business Intelligence and got
nowhere. I forgot that it does not matter if it is SPSS or SAS, or TM1 or
Hyperion, they can all for the most part do the same things. What matters is
what are you trying to achieve, and how do you want to achieve it.

Here is my list of top-ten questions to ask. At this stage, I am not going to try to
interpret the answers. If you can get honest answers, you should be in good
position to evaluate which tools are right for you and not end up with
shelf-ware and dissatisfied end-users. I am going to make one big assumption on
which to move forward. You have or can get access to the data you need.

1.
What do you want to do with the
data? Is it simple query and data retrieval? Do you need to visualize the
data? Do you want to send out standard reports across the organization? Are you
looking to explore the data and discover nuggets of information? Are you
looking to do advanced statistical analysis and data mining?

2.
What can you not do today that
you need or want to do? Are you facing specific issues that you need to
address? Are you feeling you should be doing something because everyone else is
doing it? Are you spending too much because you cannot access and interpret
information? Are you missing revenue because you cannot see far enough into the
future?

3.
What tools do you have and
use today? Why are they not working for you? Are you stuck in Excel hell? Do
you have a BI tool in place? Do you have many BI tools in place? Are they from
the same vendor? Are they all integrated? Are they relatively new or have they
been around for a while?

4.
Who will be responsible for
support? Who will be building the reports and doing the analysis? Will
end-users support themselves? Is there an IT department who will support
things? Is there a team of developers and analysts? Will there be a need for
outside help and support?

5.
Where and how do end-users expect
to receive the results? Do they want it in Excel? Do they want a
dashboard? Do they want a printed or emailed report? Do they want it on an
iPad, iPhone, Android device, Blackberry, or Windows mobile device? Should
it be sent to the shop floor? Should it be accessible from the Internet? In what
language should they be presented?

6.
What kind of data needs to be
analyzed and where does it come from?  Is it coming from a single
application or many applications? Is it stored in a spreadsheet or Access
database? Is it stored in a data warehouse or data mart? Is it stored in
specialized structure like Essbase, Analysis Services, or Hana?

7.
How much data do you have? Is it
a spreadsheet or two? Is it a few megabytes or gigabytes? Is it a terabyte or
more stored on a data warehouse appliance like Netezza, Teradata, or Exadata?
Is it “Big Data” that needs Hadoop and MapReduce?

8.
When do you need it? What is the
timeline for implementation? Do you need it next week? Is it to support a
specific initiative such as next year’s budget? Is it worthless if the deadline
is missed?

9.
What is the budget? Does it have
to be carved out of existing spend? Are there dedicated funds allocated? What
is the scale of the budget? Can it sustain a number of dedicated resources? Can
the costs be capitalized or will it all be operating expense?

10.
What is the technology landscape?
Is one vendor preferred over others? Does it matter if it is Windows, or
Unix, or Mainframe? Does it matter if it is custom code, open-source, or
commercially available software?

If you can answer these questions, you should
be in a good position to:

Good luck with your implementation!