Which metrics show whether batch record review is working?
By Savi C., Co-Founder · Published Nov 16, 2026
Why measure review at all
Batch record review is a control, so it needs evidence that it works. ICH Q10 section 2.3(e) calls for performance indicators that measure progress against quality objectives, and section 4.1 lists deviation, CAPA and change management processes among the things management review should assess. Section 3.2.1 expects the monitoring system to take in feedback from nonconformances, deviations, audits and inspections. Review findings are one of the richest sources of that feedback.
Speed metrics on their own are dangerous: a review that finds nothing can look very efficient. Pair every pace metric with a quality metric, and look at trends rather than single months. No published industry benchmark for these numbers was found, so compare your own sites, products and periods with each other.
Normalize before comparing. A 40 page packaging record and a 150 page manufacturing record are not the same unit of work, so report findings per batch and cycle time by record type, and note when a master record revision changes the number of entries. A metric that moves because the denominator changed tells you nothing about review quality.
The metrics and how to define them
Metric
Right first time
Definition and counting rule
Batch records with zero corrections requested at first QA review, divided by batch records reviewed in the period. State whether minor GDP corrections count
Metric
Findings per batch
Definition and counting rule
Review findings divided by batch records reviewed, split by category: signatures and dates, missing entries, calculations, out-of-limit values, GDP corrections, unrecorded deviations
Metric
Review cycle time
Definition and counting rule
Elapsed time from executed record handed to QA to disposition, reported as a median and a high percentile, with waiting time separated from hands-on review time
Metric
Open deviations at review
Definition and counting rule
Deviations linked to a batch that are still open when its record reaches QA, per batch. A rising number predicts release delays
Metric
Escaped findings
Definition and counting rule
Errors found after disposition, for example in APQR, audits or inspections, that first review should have caught. The most direct measure of review effectiveness
Making the metrics trustworthy
Definitions
- Each metric defined in a controlled document, with numerator, denominator and exclusions
- One finding taxonomy used by every reviewer and site, so categories can be compared
- Start and stop events for cycle time fixed to recorded timestamps, not estimates
Data
- Findings captured at the time of review, not reconstructed at month end
- Each finding linked to the batch, page and reviewer, so trends can be traced back to evidence
- Escaped findings recorded with the date of original review and the date found
Use
- Findings trended by line, product, shift and form, and recurring categories routed to CAPA
- Metrics presented in management review alongside deviation and CAPA indicators, as ICH Q10 section 4.1 describes
- Trends summarized in the annual product review required by 21 CFR 211.180(e)
- Targets set from your own baseline, and revised when forms or processes change
How this relates to FDA quality metrics
FDA’s revised draft guidance on submission of quality metrics data (November 2016) proposed three site-level metrics: lot acceptance rate, product quality complaint rate and invalidated out-of-specification rate. That was a draft, not a requirement. The review metrics above sit one level lower: they explain why lots are rejected or delayed, which is what a site needs to improve. For reducing cycle time, see How long should batch record review take, and how do you shorten it?
How Litewave fits
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, so findings are recorded in a consistent form as review happens. Litewave proposes, qualified people approve. At a leading global generic API manufacturer, hands-on review time fell from days to hours and batch disposition became 85% faster, on 100 to 150 page paper batch records with no eBPR migration.
Questions we get asked
What is a good right first time rate for batch records?
There is no published industry benchmark we could cite. Set a baseline from your own records over a defined period, using a written definition, and track the trend by product and line.
Should review cycle time include waiting time?
Report both. Total elapsed time shows the delay to release; hands-on time shows reviewer effort. Most of the gap between them is usually queueing, open deviations and missing attachments, which need different fixes.
Is a low number of findings per batch good news?
Only if escaped findings are also low. A falling findings rate can mean records are improving or that review is missing more, so always read it alongside errors found after disposition.
