Courtesy of Lorraine Lawson of IT Business Edge
Jignesh
Shah (@jshah0209
), Software
AG's vice president of business infrastructure
products and solutions, discusses with IT Business Edge's Loraine
Lawson the benefits of taking a process-driven approach to master
data management, as opposed to a data-driven approach. (In the
interest of disclosure, this month Loraine Lawson began writing
for B2B.com
, a business-to-business site owned by Software AG and
overseen by Shah's division. However, this interview was scheduled
in response to an ITBE blog post Lawson wrote on
marrying BPM with
MDM.)
"People often equate BPM with automation of processes.
There are processes that do lend themselves to automation, but
there are many things that you can do to a process that have
nothing to do with automation ..."
Jignesh
Shah
VP of Business Infrastructure
Software AG
Lawson: I know that in 2010, towards the
end of the year, Software AG - which is known for its BPM tool -
bought Data Foundations, an MDM company. How do you see BPM and MDM
coming together?
Shah: What's interesting is that the
process-driven MDM story at Software AG precedes Data Foundations,
and not a lot of people know this because it actually stems from
the IDS Scheer side of the house. Definitely we put process-driven
MDM messages on the forefront for the Data Foundation acquisition,
but IDS Scheer, which became a part of Software AG in 2009 and has
been around for about 20 years, has been doing process-driven MDM
for about 15 years.
IDS Scheer started out as a process improvement company and a
lot of their practice around process improvement was around SAP
implementations. So they took a very process-driven approach to
implementing SAP. They would document customers' processes,
how they could be changed, modified, automated, and then use all
that information to then guide the implementation of SAP systems
and modules.
Even with SAP, the databases and the data models are not
homogenous, so frequently they would implement master data
management within SAP modules or between SAP modules and other
systems that clients may have. And guess what? Being a process
company, they adopted a process model to do MDM as well. Initially,
it started out as an extension of just process thinking, but very
soon they actually had developed a whole methodology and discipline
around how MDM can be implemented and tied to process improvement
initiatives. It's a six-point methodology starting from strategy,
organization, process, data, applications and governance.
The article that you wrote and some of the things that Andrew
White of Gartner writes, they often kind of equate process with
BPM, which maybe it's true these days, but you know processes have
been around as long as we've had the human race, right?
So, BPM and processes are not one and the same, is the point I'm
trying to make. And IDS Scheer was doing process improvement before
BPM became fashionable and they've been doing process-driven MDM
before people started talking about marrying BPM and MDM, shotgun
or otherwise. So that's the backgrounds. So we've been doing this
part of the initiative for about 15 years, and then when we
acquired in 2010 OneData, it just made natural sense to say, "Hey,
we're going to go to market with this tool and with the methodology
that we have developed over the last 15 years as a combined,
unified offering," and that's how we came about the process-driven
MDM message, the process-driven MDM dummies book and that's how we
go to market these days.
Lawson: So if you're not doing BPM, how
are you managing processes? Is that just a methodology
approach?
Shah: People often equate BPM with
automation of processes. There are processes that do lend
themselves to automation, but there are many things that you can do
to a process that have nothing to do with automation - redesigning
a process, making specific pieces of the process more efficient by
having the right people, the right training, the right governance
and so on. There are many aspects of process improvement that fall
outside of scope of a classic BPMS.
When we look at MDM, we don't assume you are using a BPMS or you
have a formal BPM program in place. Regardless of whether a company
has a formal BPM program or not, every company is interested in
process improvement.
Lawson: Can you give me an example of what
you would use a BPA tool for versus a BPM tool, which I know can do
some of the integration and automating, acting as a sort of
middleware. I assume you mean that BPA does not do that.
Shah: To be clear, BPA is considered part
of BPM, a very key part of BPM. A BPA tool like Software AG's ARIS
is a process-modeling and design tool. The other part of BPM is the
automation of these processes often using a BPMS like Software AG's
webMethods BPMS. Both aspects of BPM have important touch points
with MDM.
Lawson: How does MDM fit in with your BPA
tool?
Shah: To start with, our BPA tool, ARIS,
has the ability to identify the data elements that these processes
need to operate. So in addition to defining the process logic and
the process flows and the decisions and so on, we also are able to
define what we data entities and business entities use by process
steps. Typically they tend to be things like customers, products,
orders and so on. So the analysis helps you identify these entities
and develop attributes of these entities that are needed to operate
the processes correctly.
That data model then flows into our MDM tool, which becomes the
starting point for deciding what data will you master, what
attributes will you master, and what policies do you need to put in
place to govern the changes to the master data. So all the work
that you do up front in terms of identifying these processes that
create, consume and change master data, that then flows into MDM in
the forms of data models and policies that have to be
implemented.
The other thing that happens is, as these data elements are
identified, the BPA tool also helps you link them back to the
source systems that supply information. That information, again,
becomes very useful on the MDM side because you'll decide: How am I
going to consolidate information from the sources, how am I going
to cleanse it? How am I going to put it back into these
systems?
Lawson: Why the focus on the BPA
portion?
Shah: I guess the greater point that I was
trying to make is that BPM has two components, right? It has a
process analysis and redesign component and then it has a process
automation and integration component.
People often talk about the automation and integration portion
more than they talk about the analysis and process-design
component. What I'm here to tell you is that from an MDM point of
view, it is the analysis and design component that is much more
relevant and valuable in driving MDM than the automation and
integration component.
The classical MDM approach is to do this boil-the-ocean approach
of studying all the systems that have customer information and the
hundreds of attributes associated with the customer entity and how
they differ and so on. If you do this in a vacuum, as an academic
exercise, it's just never ending. And often, MDM has been accused
of not delivering results fast enough, because people go through
that academic exercise.
When you look at it from a process point of view, then you
quickly can separate the wheat from the chaff and say, look, if the
business goal is to improve customer service and new subscriber
on-boarding processes, then here is the section of the customer
data model that is the most relevant that we need to worry about,
in terms of making it pristine and error-free and complete.
Lawson: So, it sounds like the BPA tool is
the more strategic part of this?
Shah: Exactly. A lot of the discussion
often tends to relate the automation and integration component of
BPM to MDM, and of course there is linkage there and we support it.
But when you look at it that way, it almost says, "Hey, it's only
relevant to people who are doing process automation and process
integration." What about all other companies or departments or
projects that don't necessarily have a separate process-automation
tool? Are they not interested in improving their processes with
better master data?"