| Strategy Risk Assessment When we consider replacement as a technical strategy, we look to see how
closely coupled the business process is to the application process and how flexible the
business process is. If the business process is very tightly coupled and also relatively
inflexible, then this is a high-risk project regardless of the technical strategy chosen.
If the coupling is relatively loose or the business process relatively flexible, then the
project would be seen as moderately risky. Finally, if the coupling were loose and the
business process relatively flexible, then the project would be seen as relatively low
risk. Regardless, there is no such thing as a no risk project, even among replacement
projects implementing off the shelf packages.
This issue of tight versus loose coupling can be seen in
terms of the degree of automation of the business process. For example, a typical
warehouse picking system is highly automated and very tightly coupled; little if anything
happens without the computer working correctly. Banking applications are also highly
automated, but there is an accompanying oversight component that makes the process
somewhat looser. Incorrectly rejected transactions, for example, may be recoded as a
slightly different transaction that can be accepted, providing a workaround pending
resolution of the problem. At the other end of the spectrum are businesses such as a auto
dealership, for example, that may use the computer when it fits the task but can function
perfectly well with a couple of telephone calls if it's down. If the software doesn't
provide all the services required, it is supplemented with manual processes, just as
unused transactions can be re-used in novel ways. In general, if the business can continue
to operate without the computer for an extended period of time without significant cost,
or if the business operates in many respects in very different ways from the way the
software designers expected, then we consider the coupling to be loose.
The flexibility of the business process is less an issue of
dependence on the computer as it is the degree of structure of the process itself. For
example, an automobile manufacturing process has a huge investment in equipment and
processes, and cannot change the way cars are built without spending a lot of money and
time. Services can be similarly inflexible, such as insurance claims processing where
manual processes are highly evolved and detailed, and have to dovetail exactly with
contractual commitments as well as governmental regulations. By contrast, a consulting
company can be very flexible in its business process, providing a unique service to each
client. Markers for inflexible processes include highly tuned, very efficient processes,
those with very large capital investments, and with problematic labor relations
agreements.
| |
The higher the risk assignment (4 or 5), the greater is our degree of bias toward
to a conservative technical strategy such as renovation, all other issues (cost, delivery
time, return on investment, etc. being equal). The lower the risk assignment (1 or 2), the
greater our degree of bias toward a more aggressive technical strategy, such as
replacement with new development. For a moderate risk assignment (3), our bias is toward
an in-between strategy such as re-engineering.However,
completing the risk assignment from coupling and flexibility assessments is only a first
step. Then we look to an assessment of the cost of an error if and when it does occur in
the resulting updated, re-engineered or new application system. Applications for which the
cost of an error can be $10 million trade execution, or for which the cost of downtime can
be $1 million or more per hour, have a risk profile of a wholly different order of
magnitude than those for which the cost of an error is lost productivity but little or no
direct cost in terms of lost revenue, legal liability, or out of pocket costs. A related
issue can be the cost to the business of significant delays in the delivery of a reliable
and useable system, which can be disproportionate to the length of the delays in some
cases. High potential cost per error or for delays translates into an argument for a more
conservative technical strategy and for significant investments in effective testing of
the results, regardless of the
strategy actually chosen. |
