Suppose an operator finds an unexpected result during a production run. Can they tell from the procedure whom to notify? Can the quality unit see the original data, along with any related equipment or maintenance records? And if an investigation identifies a problem, does someone go back later to check whether the action taken actually worked?
These questions are a practical way to examine a good manufacturing practice (GMP) quality system. The answers depend on how well the people, records and procedures at your site work together.
Can your quality unit get the information it needs?
The quality unit has defined responsibilities under U.S. current good manufacturing practice (CGMP) regulations for finished pharmaceuticals. These include reviewing records and making decisions about product approval or rejection. That responsibility carries weight only if the unit can access the information needed for its review.
An equipment alarm offers a useful example. Production personnel may know what was happening on the line when it sounded. Engineering may hold the maintenance history. Laboratory staff may have test results that help assess the batch. The quality unit needs a way to bring those pieces together before deciding what further work is necessary.
It also needs enough time and qualified staff to do that work. If investigations regularly wait for records to arrive or for someone to become available, look at the handoffs between departments. A delay may be a scheduling problem. It may also point to a process that does not make responsibilities clear.
Your records should tell the whole story
When someone reviews a batch record, they should be able to understand what was done and identify anything that needs further attention. The same goes for laboratory records: a reviewer needs access to the relevant data, including unexpected results that may require investigation.
In a June 2026 warning letter to Genzyme Ireland Limited, the U.S. Food and Drug Administration described test failures and non-viable particulate excursions that appeared in instrument histories but were not documented in laboratory records for review and investigation. The FDA also raised concerns about the traceability of some environmental monitoring records.
That is a useful case to consider when reviewing your own systems. Where does an instrument store its original data? Who checks it? Can the reviewer connect the results to the correct sample and batch? The answers may involve several systems and people, but the information still has to come together.
What happens after a deviation is reported?
A deviation tells you that something needs attention. It does not tell you why it happened or whether the product was affected.
Imagine that a temperature reading falls outside a specified range during production. The team documents the event and takes the immediate steps required by the site’s procedures. An investigation can then examine how long the excursion lasted, what material was present and what the equipment records show. The team may also need to consider whether other batches or equipment were affected.
Be careful about settling on the first explanation that appears plausible. A sensor fault, for example, may seem an obvious place to look, but the investigation needs evidence before it can support that conclusion. If the cause remains uncertain, the report should say so and explain how the team assessed the available information.
The scope of the review deserves attention, too. In the Genzyme letter, the FDA noted that the company had retrospectively reviewed 74 cancelled deviations identified during the inspection and concluded that 36 required investigation. The agency requested a broader assessment because it considered the selected review timeframe insufficiently justified.
A site dealing with repeated deviations or a backlog of investigations may bring in GMP consulting services to support the review of records, investigations and proposed corrective actions. The manufacturer still needs to understand the findings and retain responsibility for its quality decisions.
Before you retrain someone, look at the task
Retraining is an attractive response to an error. It is familiar, it can be scheduled, and the completed training can be recorded. But first ask what the employee was being asked to do.
Consider a hypothetical batch form with two fields that look alike. Employees repeatedly enter information in the wrong one. The investigation might find a training gap, but it should also examine the form and the instructions that accompany it. Are the fields clearly labeled? Does the order match the way the task is performed?
If the form contributed to the problem, revising it may be part of the corrective action. Employees would then need to understand the change. The point is to choose the action after examining the cause, rather than assuming every error has the same solution.
How do you know a corrective action worked?
A corrective and preventive action, or CAPA, can be completed on schedule and still fail to solve the original problem. A procedure may have been updated and training delivered, but those steps do not show what happened afterward.
For the form in the example above, the site could review a defined sample of later records for the same error. The review needs a suitable timeframe and enough information to support a conclusion. If the error continues, the team should revisit its assumptions and decide what further action is needed.
This follow-up is easy to overlook when teams are busy closing investigations. Yet it provides the answer management needs: did the change address the problem it was designed to fix?
Consider the effects of a change before it reaches the floor
A software update may look like a technical task until an operator opens a familiar screen and finds that the fields have moved. Depending on the update, the change may affect how data is entered, reviewed or stored. It may also require changes to procedures, testing or training.
The same thinking applies to a new equipment component or a revised cleaning method. Who uses it? Which records refer to it? Does the change affect an existing qualification or validation? The answers will vary, but asking the questions before implementation gives the relevant teams time to assess the impact.
A training record can show that an employee completed an assigned activity. For an unfamiliar task, observing the work or discussing the revised steps may help a supervisor assess whether the instruction is understood.
Give management something more useful than a closure rate
Knowing how many investigations were closed this month is useful. Knowing whether the same problem keeps returning is more useful.
Review recurring deviations alongside overdue actions, investigation findings and the results of effectiveness checks. Look for connections across batches, equipment and shifts. If several events point to the same unclear instruction or recurring equipment fault, the individual investigations may be telling you something about the wider process.
A manufacturer can seek pharmaceutical GMP consulting when it needs an independent view of those patterns or support assessing gaps in its quality system. GMP compliance consulting may help identify where written procedures and daily practice have drifted apart.
The site then needs to decide what happens next: who will investigate, what information they need and how management will know whether the resulting action has worked. That is where a quality system earns its place in day-to-day manufacturing.