O P E R A T I N G M O D E L S
◂
139
analytics as a real competitive rather than just a useful source of
insight.
Productivity and effi ciency comes from:
◼
Defi ning an engagement model that identifi es the handover
points
between individuals
◼
Establishing standardized development and deployment processes
Getting this right helps drive process effi ciencies, ensure quality con-
trol, and simplify the application of competencies across new prob-
lems. It also ensures that governance is tailored appropriately. Too
much, and innovation suffers; too little, and operational risks increase.
It ’s a key part of an effective operating model. The major focus in
this area is on managing workfl ow and
facilitating collaboration and,
just like integration and asset management, this occurs at two levels:
1.
Coordinating the development of analytical assets
2.
Coordinating the deployment of analytical assets
As with asset management, there is no reason why both of these activ-
ities can ’t take advantage of a common technology platform and man-
agement approach. It ’s important
to remember, however, that while
the vast majority of their requirements overlap, the level of emphasis
placed on specifi c requirements varies between the two.
Achieving best practice within this process requires at a minimum:
◼
Establishing a clear operating model that outlines roles, respon-
sibilities, and handover points
◼
Documenting and following
standardized processes
◼
Having well-defi ned points of ownership with the power to
make decisions
◼
Ensuring a high degree of quality through explicit quality con-
trol activities
Unclear processes almost always create highly variable outcomes
and process ineffi ciencies. It ’s hard for an organization to drive con-
tinuous improvement when everyone follows a different process.
Some people will naturally do things more effi ciently than others.
Unfortunately, when everyone
does things differently, it
’s almost
140
▸
B I G D A T A , B I G I N N O V A T I O N
impossible to replicate those effi ciencies. Having standard processes
not only increases agility through making sure everyone has clarity
on how to execute but also ensures that everyone benefi ts when team
members fi nd new effi ciencies.
This does not necessarily mean that activities need to be defi ned to
the lowest level of detail possible—a certain degree of pragmatism and
realism needs to be applied when working out an appropriate level
of granularity. It ’s also true that too much rigidity stifl es innovation;
when people aren ’t given the freedom to experiment,
improvements
tend to be the fi rst thing that suffers.
Underpinning these processes are roles and responsibilities. To be
effective, everyone must be crystal clear on what they are respon-
sible for delivering as well as when they need to get involved. This
helps provide certainty as well as reduce transaction costs. By link-
ing roles to activities, the workfl ow system itself can automatically
notify stakeholders when their interaction is required. If a champion
model has
been submitted by an analyst, the next logical step would
be for the information management team to deploy that model
into production and validate the predictions against the original
model. Ideally, the system itself handles all the necessary notifi ca-
tions based on a combination of asset registration or workfl ow and
process management.
Finally, effective governance requires a high degree of quality con-
trol. When it comes to dealing with operational systems, repeatability
and transparency are critical. Every process
must be exhaustively tested
prior to moving it into production lest it fail and cost the organiza-
tion real money. Minimizing risk involves ensuring that standard tests
are applied, making sure that appropriate signoffs are followed, and
ensuring that outputs and predictions are the same in production as in
development. While the checks will vary between information assets
and analytical assets, the need for a high degree
of quality control is
a constant.
Do'stlaringiz bilan baham: 
