What are the most common batch record errors found in review?
By Savi C., Co-Founder · Published Nov 16, 2026
Why small errors become big findings
Most batch record errors are small on their own: a missing initial, an unexplained crossing-out, a weight copied wrongly from a printout. They matter because the record is the evidence that the batch was made as approved. 21 CFR 211.188 lists what the record must document, including each significant step, the identity of the people performing and checking it, and any investigation. 21 CFR 211.192 requires the quality control unit to review the record before release and to investigate unexplained discrepancies.
When a review misses errors, or finds them but does not resolve them, the problem becomes an inspection finding about the review itself. That is why the same errors are worth trending and preventing, not only correcting. A recurring documentation error can also be a signal that a step is hard to perform as written, which is a process question as much as a paperwork one.
The errors reviewers find most often
Signatures and dates
- A step performed but not signed, or not checked by the second person the record requires
- Signatures added after the fact, out of sequence with the times recorded
- Dates in an ambiguous or inconsistent format
- Initials that do not match the signature log
Completeness
- Required entries left blank
- Unused fields not voided, signed and dated
- Missing pages, attachments or printouts
- Ditto marks or arrows used in place of a value
Corrections
- Overwriting, or correction fluid that obscures the original
- A correction without initials, date or, where needed, a reason
- Entries made in pencil or erasable ink
Data and limits
- Transcription errors between instrument printouts and the record
- Calculation errors in yields, potency adjustments or material quantities
- A value outside its limit with no deviation referenced
- A step performed outside its allowed time or hold window
Deviations and reconciliation
- A deviation from written procedures that was not recorded and justified, as 211.100(b) requires
- A referenced deviation or investigation that is still open at review
- Yield or label reconciliation outside limits without an investigation
Where each error usually starts
Error
Missing signatures and blank fields
Common root cause
Forms with crowded or unclear entry spaces, and records completed at the end of a shift rather than at the time of the step
Error
Poor corrections
Common root cause
GDP training that is generic rather than drawn from the site’s own findings
Error
Transcription and calculation errors
Common root cause
Values copied by hand between printouts and the record, and manual arithmetic
Error
Unreferenced deviations
Common root cause
No clear trigger in the record telling the operator when an event needs a deviation
Error
Late or out-of-sequence entries
Common root cause
The record is not available at the point of use when the step is performed
From correcting errors to preventing them
Correcting each error in review keeps a batch moving but does nothing for the next one. Prevention needs data: a log of every review finding with its type, the form and step it came from, and how it was resolved. Within a few months that log often shows findings clustering in a handful of fields and steps.
Those are the places to act. Redesign a crowded form section, add an explicit prompt where a deviation is needed, replace a manual calculation with a pre-calculated table, or move the record closer to the point of use. Then watch whether the count for that error falls over the following batches, and keep the change only if it does. If the same error keeps recurring after training, the cause is usually the form or the process, not the operator.
How Litewave fits
Each error above can be checked by rule on every entry. Litewave reads paper or electronic records, checks every entry against your master record and specifications, and shows QA only the exceptions, each linked to the page it came from. Because the same checks run on every batch, findings can be counted by type and process step, which is what prevention needs. Litewave proposes, qualified people approve.
Questions we get asked
Which batch record error is most serious?
Errors that hide or alter data, such as obscured corrections or entries made long after the event, and out-of-limit results with no investigation. These go to whether the record can be trusted, not just whether it is complete.
Is a missing signature a deviation?
That depends on your procedures. Many sites handle documentation errors through a GDP correction and a review comment, and escalate to a deviation when the error affects product quality data or recurs. Your SOP should define the threshold.
How do we reduce repeat errors?
Trend review findings by error type, form and process step, fix the record design where the same error recurs, and train on real examples from your own records.
Sources
- The FDA Group, 21 CFR 211.192: An Introduction and Compliance Guide
- 21 CFR 211.100, Written procedures; deviations (eCFR)
- 21 CFR 211.188, Batch production and control records (eCFR)
- 21 CFR 211.192, Production record review (eCFR)
- PIC/S PI 041-1, Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments
- EudraLex Volume 4, EU GMP Chapter 4, Documentation
