The
objective of the SOX test is to prove that the control activities included
in the Risk and Control Matrices are functioning as stated.
The first part of the article focused on SOX testing approaches
used, test data to be used and sampling approaches.
The
most critical component of the SOX test is to define the conditions in
which a test would pass or fail. In
some cases, there are multiple pass/fail conditions per test sample.
For example, there are two pass/fail conditions for the test
objective to ensure that properly authorized requests exist.
The test data required to perform this would be a system generated
list of security changes made. The first test condition would be to
determine whether a security request exits for the security change.
The second test condition would be a validation against the
authorized approver list to ensure that the security change was approved
by the appropriate person. If
there was not a formal request then the second test condition to validate
the approver cannot be performed. Regardless,
both test conditions would need to pass for the test sample to be
categorized as a test that passed. A
case could also be made to separate the two test conditions into separate
tests which would increase the total number of tests performed.
Controls
are defined as control objectives which are transposed into control
activities. Each control
objective could have multiple control activities. The frequency of the
control activity which determines the sample size could be different for
each control activity which is associated with a particular control
objective. For certain
control activities, the frequency is intermittent.
For example, the control activity to ensure that security patches
are applies to the production database servers for a SQL Server
environment is intermittent. There
has not been a security patch for SQL server since 2004 therefore there
would be no tests which could be performed for this control activity.
Another example of an intermittent control frequency would be to
ensure that security requests are formally approved.
In a small environment there may only be a few security changes per
year in which case the entire population would be included in the annual
sample selection. Controls
with intermittent frequencies should be re-evaluated on an annual basis.
Each
control activity would have their own test.
In some cases there may be multiple tests per control activity with
each test having different test frequencies and sample sizes.
When a control is an automated control, the test frequency consists
of one annual test. For
instance, an email notification which is sent for each backup failure
which supports the control to ensure that failed backups are identified
and acted upon would only require an annual validation to prove that the
email notification process is functioning.
However, the second test consists of ensuring that proper follow-up
occurred for failed backups identified.
The determination of whether this is a daily control can be debated
(i.e., the frequency of the backup as compared to an intermittent control
which is based on the frequency of the failure).
Another
example of multiple tests for a control activity relates to the detective
review processes that are in place. For
instance, a control activity which is comprised of identifying and
notifying the owners of IDs when there are more than four successive
invalid logon attempts to their ID within a single day involves two tests.
This control requires an area responsible for performing this
review and documenting the review that occurred and notifying the owners
of the Ids to determine whether they were the cause of the invalid logon
attempts. The first test is a
reperformance test to ensure that review process is being performed
correctly. The
reperformance test consists of reperforming the system operation function
to gather the data which is analyzed based on pre-defined rules.
The compilation of the incidents derived from the reperformance
review is then compared to the summary reports used to support the weekly
review process to ensure the review process also identified these
incidents. The
second test consists of ensuring that the review is actually occurring and
the associated emails notification to the owners of the Ids is being sent
out. The second test’s
sample size is based on the frequency of the review process.
The reperformance test is typically an annual review.
SOX
testing require the tester to provide a rational for the type of test
performed and the sample selection. The
one lesson which everyone has learned with SOX is that every SOX tester
may have a different view on how the test should be performed.