David Toole
Business
Analytics Specialist
Each business intelligence initiative comes
with a unique set of challenges and obstacles that must be overcome before reaching
the proverbial golden analytics palace on the hill; where the reporting is accurate,
the discoveries are actionable, and the ROI flows like wine. While no two implementation journeys are the
same, there are indeed several identifiable similarities in the roadmaps, which
if followed, will help lighten your load as you progress towards the front gates.
Let’s take a look at 4 important ones:
·
Say What?
–Many (many, many…) years ago I took my first trip to Massachusetts to visit
with family. One afternoon, my uncle
decided to take my cousins and me to the movies. After buying the tickets we
quickly made our way to the concession stand, where I attempted to ask the
clerk for an “orange pop.” “Pop, do you
mean like, a popsicle?” the clerk replied. “No, orange pop, in a cup,” I retorted. This continued on for a bit until my uncle,
fluent in both Western PA and New England-speak, translated… “He means soda.” I tried to interject as to me, soda was the
nasty, clear, and tasteless version of pop that my mom used to get stains off
of clothing. The clerk seemed to
understand as I was quickly handed a gigantic cup of pseudo orange flavored
goodness. This story illustrates a
common obstacle among businesses of all sizes where different vernacular is used
to describe the same thing. For example,
Sales Analyst, Tom, may call the company who purchased the product as the “Customer” but VP of Sales, Sally,
views the “Customer” as the company who the products were shipped to. Even more
troubling is defining business rules. For
example, a sales department and a finance department may explain “gross margin”
differently or have a disconnect on which costs should be categorized as “overhead”
vs “cost of goods sold.” The power of BI
is bringing data from several different silos together so the business can be
viewed and analyzed as a whole. Without uniformity in data definitions this
cannot be achieved. Before beginning a
BI implementation there should be open discussion between departments to find
commonality among the language used to define their business.
·
Divide and Conquer – Drinking from a bottle is a lot easier
than drinking from a fire hose. The same concept is true when implementing
BI. Instead of tackling the whole
project at once, focus on a single key area within the business, build a few
accurate, easy-to-digest reports based on this focal point, and release them to
a limited number of end users and business analysts for validation. While small
and seemingly insignificant, getting a few quick wins into the hands of the
users will build trust and hopefully excitement about what else is yet to
come! At the same time, the slow and
steady release schedule gives the end user time to grow comfortable with the
look and feel of the tool and oftentimes leads to specific formatting, query,
or layout requests in future reports. As
a BI administrator, getting feedback from users is essentially winning the
analytics development lottery! Positive
or negative, receiving feedback means the users care about the tool, see
potential, and (whether they know it or not) have actively expressed a desire
to help shape the future of the business.
That brings me to the next suggestion for success.
·
Power to the People- You could have the cleanest data,
the most efficiently modeled warehouse, and
the trendiest BI tool on the market, but if no one at the company is
taking advantage of it, your BI is about as valuable as the chewing gum on the
bottom of my shoe. To maximize adoption
rates and to prevent your BI initiative from becoming an unseen and trampled
upon minty mess, the end-user perspective must be considered from the very
beginning. By involving end-users early
on in the process, you’ll grow into the software together, making the process
less intimidating and creating an environment where it’s easy to ask questions or
to suggest improvements. The user will essentially shape the solution to their
needs, allowing business requirements to be met more effectively and a broader
overall use of the tools.
·
The Data Cleanse- It’s not the newest health trend to sweep the
nation, but it should be a top priority for companies looking to implement a BI
initiative. Whether the underlying data
is incomplete, outdated, or contaminated with misspellings and typos generated from
the use of manual data entry, it needs to be rehabbed as much as possible
before the modeling begins. Skipping
this step will end up causing headaches down the road where your BI team will
be spending their time correcting mistakes vs utilizing the in-depth analysis
offered by the tools. To avoid this
nightmare, sit down with your team and brainstorm reporting requirements. Any fields that come up in conversation (ex.
salesperson, customer location, etc.) should be identified in the database
tables, inspected for completeness (are all desired attributes available), consistency
(capital vs lowercase, 5 digit zip or 9 digit zip, etc.), and quality
(spelling, partially completed data, etc.).
Any issues that arise should be dealt with in this phase. Your analytics,
reporting, and forecasting will only be as good as the stuff you have in your
database so put the time in early on and avoid the need for Advil down the road!
