← Back to Insights

InsightInternal Controls

When a Control Exists Only on Paper

A control can be documented, performed, and signed off and still fail to address the risk it was meant to cover. The harder question is diagnosing what is actually wrong before deciding how to fix it.

Key takeaways

  1. Start with the risk the control is meant to address, not the control description.
  2. Separate documentation, design, execution, ownership, and data problems before deciding what to fix.
  3. The right fix is not always another control. Sometimes the answer is to redesign one or retire it.

The control is documented. Someone can show you evidence that it was performed. The required sign-off is there.

The problem is that the same errors keep appearing.

Exceptions are missed. Reviews happen without catching problems that later surface somewhere else. People work around the process because the documented version no longer fits how the business operates.

When that happens, figuring out what is wrong with the control becomes more important than confirming that it exists.

The issue may be the documentation, the design of the control, how it is being performed, or the conditions surrounding it. More than one of those can be true at the same time.

Start With What the Control Is Supposed to Do

When a control is not working, start with the risk rather than the control description.

What is supposed to be prevented or detected?

Controls accumulate over time. A review gets added after an audit finding. A sign-off appears after an error. Someone creates a spreadsheet because the system does not provide the information the team needs. Years later, several activities may surround a process without anyone stopping to ask what each one is meant to accomplish.

Knowing the risk makes the next question easier: if the control were performed exactly as intended, would it address that risk?

Sometimes the answer is no. That tells you something very different from finding that a sound control was simply missed.

It can also reveal that the control is not the problem at all.

A detective control may repeatedly identify the same transaction error and be described as ineffective because exceptions keep appearing. If the control identifies those errors with enough precision and early enough for someone to address them, it may be doing what it was designed to do. The recurring errors may instead point to a weakness somewhere upstream that needs its own attention.

When the Documentation Has Fallen Behind

A difference between the documented process and the actual process does not automatically mean the control is ineffective.

The business may have changed roles, replaced a report, adjusted a workflow, or found a better way to perform the control without updating the documentation. The control itself may still work.

The situation becomes more concerning when nobody can clearly explain which version is correct.

In practice

The control documentation says a manager reviews a report monthly. The procedure assigns the review to a supervisor. The manager says the report is no longer used, and the supervisor describes a different review performed as needed.

Before updating anything, someone needs to establish what people are really doing, whether that activity addresses the intended risk, and what the expected control should be going forward. Otherwise, it is easy to update the documentation to match current practice without first deciding whether current practice makes sense.

Would the Control Work If Everyone Did Exactly What It Says?

This is where design and execution can become difficult to separate.

In practice

Take a monthly reconciliation that requires supervisory review. The preparer completes the reconciliation, the supervisor signs it, and the file shows that both steps occurred.

Significant reconciling items continue to age.

If it is design

One explanation is that the review is poorly designed. Perhaps the supervisor is only expected to confirm completion, with no clear expectation to investigate unusual balances or challenge old items.

If it is execution

But there is another possibility. The control may already require the supervisor to investigate those items, and the supervisor simply is not doing it.

The evidence can look similar while the underlying problem is different.

Before redesigning the control, find out what the reviewer was expected to do. If the expectation itself was insufficient, the design needs attention. If the expectation was reasonable and was not followed, the issue is execution.

The harder situation is when the control simply says that a supervisor “reviews the reconciliation.” Management may believe it is obvious that the reviewer should investigate old reconciling items. The reviewer may believe the required review was completed because nobody ever established what level of investigation was expected. What looks like an execution problem can turn into a design discussion very quickly when the precision of the review was never defined.

Exception reports create similar questions. Generating a report and assigning someone to review it is not enough if nobody has established which exceptions require investigation, how they should be resolved, or when unresolved items should be escalated.

An approval can have the same weakness. The approver may act before the transaction is processed every time, but the approval provides limited protection if the information needed to make the decision is missing.

When Execution Is Really the Problem

Not every control issue has a deeper structural explanation.

Sometimes someone misses a step.

A new employee may misunderstand one part of a review that everyone else performs consistently. The control is reasonable, the information is available, responsibilities are clear, and there is no history of similar exceptions. Clarifying the expectation or providing targeted training may be enough.

In practice

A different problem can be harder to see. A review is required by the fifth business day, and the file shows that the owner completed it on time. During the walkthrough, you learn that the information needed for the review is rarely complete by then.

The owner explains, “I review what I have so I can meet the deadline. The rest usually comes in later.”

Now the problem is not a late review. On paper, the control is timely. The question is whether a review performed over an incomplete population can accomplish what the control is supposed to accomplish.

The same kind of problem can show up in other ways. A reviewer may have too many transactions to review at the level expected. A manual control that worked when volume was low may become difficult to perform after the business changes. A report may still arrive on time but no longer contain all of the data the reviewer needs.

A recurring problem deserves a different diagnosis from an isolated missed step. That distinction helps explain the cause, although it does not necessarily change the testing conclusion. One exception in a control that operates quarterly can have a very different significance from one exception in a control performed hundreds of times.

Who Owns What Happens Next?

The person listed as the control owner is not always the person who can explain how the control works.

That becomes apparent quickly in walkthroughs. You ask the named owner how the review is performed, and the response is, “Let me bring in the person who actually does this.”

There is nothing inherently wrong with someone else performing the underlying activity. But if the named owner is also supposed to be reviewing or approving that work and cannot explain what the review involves, the question changes. What is the owner's sign-off evidencing?

Ownership also includes knowing when the control no longer fits the process and who has the authority to change it. Those responsibilities are not always as clear as the owner field in the control documentation suggests.

What the Evidence Can Tell You

For a review control, one of the first things to understand is what the reviewer looked at and what would have caused the reviewer to take action.

Questions to ask

  • If there were exceptions, what happened to them?
  • If there were no exceptions, does that make sense given the underlying activity?

This becomes especially important when judgment is involved. Two reviewers can leave nearly identical evidence behind while performing very different reviews. One may challenge unusual items, trace information back to support, and follow an exception through resolution. Another may scan the same information and conclude that nothing looks unusual.

A sign-off shows that someone recorded a review. How much it tells you about the quality of that review depends on the control and what else the file shows.

A well-designed review also depends on the information provided to the reviewer. If a report is incomplete, a field changed after a system update, or the population excludes certain transactions, the reviewer may follow the expected steps and still miss the problem. The reviewer may have performed the assigned activity, but the control design still needs to address the completeness and accuracy of the information the review depends on.

A sign-off shows that someone recorded a review. It does not, on its own, show how good that review was.

The Problems Rarely Stay in Their Own Lanes

Documentation, design, execution, ownership, data, and the underlying process are useful distinctions. Real control problems do not always stay within one of them.

How it compounds

An experienced employee leaves. The replacement follows the written procedure, but several judgmental steps the former employee performed were never documented.

The replacement does not perform those steps because there is nothing telling them to. Exceptions begin getting through.

Important elements of the review were never captured. Knowledge was concentrated with one person. And if the control only worked because that person knew what to look for, the design may have been more dependent on individual experience than anyone realized.

A neat classification is less important than understanding the chain of events well enough to know what needs attention.

Sometimes the Answer Is Less Control

There is a natural tendency to respond to control problems by adding something: another approval, another review, another spreadsheet, another sign-off.

That is one reason controls accumulate. Diagnosis can lead in the other direction.

Two controls may be addressing essentially the same risk. A manual review may have remained in place after a system change introduced a more reliable control upstream. A detective control created years ago may no longer serve the same purpose because the process around it has changed.

Removing a control should require the same kind of thought as adding one. Before it goes away, someone should understand whether another control depends on it, whether an auditor or other party relies on it, whether it was established in response to an earlier issue, and whether the underlying risk remains covered.

And sometimes the control that exists on paper no longer needs to be there at all.

Have a Specific Problem to Work Through?

An aging finding, control issue, documentation gap, or workstream short on capacity is enough of a starting point. Tell us what is happening and what you need to be true at the end.

Tell us what is stuck, what changed, or what deadline is driving the need. We can determine whether BPS Advisory Group is the right fit and what a sensible scope would look like.