In 2023, roughly 40% of FDA citations involved a data integrity issue. What is surprising to people is where those violations originate. Most quality directors imagine data integrity failures as something dramatic, such as deleted records, backdated entries, falsified results… and those things happen. But a significant number of citations trace back to something far more ordinary: a rounding error in a formulation record.

That is what Significant Figures enforcement in Dynamic Fields is built to solve.

 

The Violation Nobody Sees Coming

ALCOA+ is the framework the FDA uses to define what compliant data looks like. Attributable, Legible, Contemporaneous, Original, Accurate plus Complete, Consistent, Enduring, and Available. Every one of those principles must hold for every data point in your electronic batch records.

The A in Accurate is where significant figures come in.

When an operator enters a measured value with too many digits, they are implying a precision the instrument never provided. When they round down too aggressively, they are discarding precision the process requires. Either way, the record no longer accurately represents the measurement. This is an ALCOA+ violation. It is also exactly the kind of finding that appears in warning letters as “failures in accuracy and integrity of formulation data” – language that is seen frequently since the FDA’s 2025 Current Good Manufacturing Practice (CGMP) guidance introduced formal data lifecycle requirements.

The uncomfortable reality is that most manufacturers know this is a risk and still rely on operator training to manage it. Train people well, the thinking goes, and they will record values correctly. That logic worked when batch sizes were small and the same two people ran every process. It does not hold when you have multiple shifts, high turnover, and batch records with dozens of numeric fields.

 

Why Training Alone is Not a Control

Training is important, training is required, but training is not a preventive control.

An FDA investigator looking at a data integrity citation will not accept “we trained our operators” as a corrective action. They will ask what system-level control you put in place to prevent the error from occurring again. If the answer is “we retrained them,” you are going to have a difficult conversation.

The FDA’s 2025 CGMP data lifecycle guidance made this expectation more explicit. Data must be generated, captured, and reviewed with integrity across its entire lifecycle. That means the control needs to exist at the point of generation, the operator enters the number, not downstream in a review step that may catch it hours or days later, if at all.

This is the gap that automated significant figures enforcement closes.

 

What Enforcement at the Point of Entry Actually Means

Dynamic Fields in InstantGMP™ PRO and InstantGMP™ MES allow manufacturers to build custom numeric fields directly into batch records, formulation steps, and production documents. With significant figures handling now built in, the system enforces precision requirements upon entry.

If a field requires three significant figures and an operator enters four, the system flags it before the record is saved. If they enter two, they have the same result. The configuration is set by the quality team in advance, tied to the instrument, method, and specification for that measurement. The operator does not make a rounding judgment. The system does not allow a non-conforming entry to become part of the record.

This is system-level control. This is what an FDA investigator is looking for when they ask how you prevent data integrity errors at the source.

The downstream effects are equally important. Your audit trail shows that every numeric entry was reviewed against a defined precision requirement at the time of entry. Your batch reviewers do not spend time hunting for rounding inconsistencies across dozens of fields. Your quality management system does not generate a CAPA because a review step caught something the entry step should have prevented.

 

The Data Lifecycle Closes When the First Entry is Correct

The FDA’s 2025 CGMP guidance introduced a concept that deserves more attention than it has received, which is the data lifecycle. Data does not just exist at the moment it is reviewed. It is generated, captured, processed, and retained with integrity requirements apply at every stage.

Significant Figures enforcement addresses the generation and captures the stages directly. It ensures that numeric data enters the system with the correct precision from the first moment it exists as a record. Every downstream step:  review, approval, release, archival, inherits that accuracy. There is no error introduced at entry that a downstream control must find and correct.

For pharmaceutical manufacturers operating under 21 CFR Part 211, dietary supplement producers under 21 CFR Part 111, biotech firms, contract manufacturers, and compounding pharmacies or any operation where formulation accuracy and data integrity are regulatory requirements; this is the kind of control that closes the gap between a training record and a defensible compliance posture.

 

The Question to Ask Yourself Before the Next Inspection

When an FDA investigator asks you how your system ensures the accuracy of numeric data at the point of entry, what do you say?

If the answer involves describing your training program, your batch review checklist, or your second-person verification procedure, those are all legitimate parts of a quality system. None of them are a complete answer to the question.

If the answer is “our Dynamic Fields are configured to enforce significant figures at entry and will not accept a non-conforming value,” that is a system-level control. That is the answer that closes the finding before it is written.

Data integrity failures that trace to rounding errors are preventable. The technology to prevent them exists. The only question is whether you have it configured before the inspector walks in.

Schedule a demo to see how InstantGMP’s Dynamic Fields and significant figures enforcement work inside a fully integrated electronic batch record platform.