Perfect Cubes and Golden Chalices
It's compelling to think of data as going through a one-time
process of cleansing and mapping, resulting in a "perfect spending cube" that represents Truth.
That's nonsense, of course. It's been our experience that,
regardless of the time and money expended on a cleansing/mapping effort,
an experienced commodity manager can drill into the tail of the
distribution and almost immediately find spend that's poorly or
incorrectly mapped. We've seen this sucker punch delivered many
times to spend analysis vendors who claim to have "special knowledge"
or "proprietary techniques" or "custom databases" that supposedly
make them extra-special at cleansing and mapping.
Instead, if we admit that perfection is unattainable, this
clears the air and lets us reach a reasonable solution.
Since it is inevitable that we will find poorly-mapped
|
Why doesn't every vendor approach the mapping process in an incremental fashion? Because their rules systems
are offline systems.
|
|
spend whenever we do a deep dive into a category, the BIQ
philosophy is "fix it when you find it." Our mapping system operates
in real time, right off the data viewer. Rules can be added and
executed immediately, without any complex offline processes.
Our collaboration feature allows multiple users to make changes
in the rules and hierarchies, and coordinate those changes later.
So, BIQ users first do a straightforward and inexpensive overall
mapping exercise (see
Mapping Spend: Three Easy Steps). Then they
examine the results, and focus more closely on categories of strong
interest. For example, if Commercial Print is a relatively uncontrolled
spending area, as it often is for financial services businesses, it's
appropriate to dive in and create more intelligent mapping for the
various types of Print spend (envelopes, lettershop, and so on).
When that's done, it's time to move onto the next category. Errors
are fixed as they're found, right then and there.
Why doesn't every vendor approach the mapping process in this
common-sense, incremental fashion? Because their rules systems
are offline systems. Their data cubes are read-only datasets.
Offline procedures and mechanisms have to be brought into play
for any changes. Then the dataset has to be "re-cooked" and "re-published."
So these vendors must think about cleansing and mapping as a monolithic process,
not as a series of incremental changes.
That's a losing proposition, and it gets worse with time. As offline
changes are being made (typically when the data cube is refreshed, at
the end of the month or quarter), more changes pile up at the door.
It's the proverbial dog chasing the fire truck — he'll never catch up,
and he falls farther and farther behind. As one BIQ customer put it,
after converting from another spend analysis system, "With BIQ, we
can get through many cycles of data clean-up before the next batch
of data comes. We're finally ahead of the curve."
