Applications · Quality
Continued Process Verification
Stage 3 of process validation, run by a Digital Scientist. Quickflow CPV takes CPP and CQA values from eBMR, LIMS and the historian as each batch closes, charts every parameter, applies the Nelson rules and tracks Cp, Cpk, Pp and Ppk. When a process starts to drift, it explains why and what to do. QA reviews and signs.
Quickflow CPV · Product A 10 mg tablets · CQA-03 Dissolution (30 min)
Stage 3 · liveBatches trended
30
Rolling Cpk (12)
1.04
State
Signal
Digital Scientist · reasoning
- 1The shift starts at B-2419, the first batch made with MCC lot 7731 from the new supplier (ERP receipt GR-5520).
- 2Mean dissolution moved from 90.2% to 88.5%. Rolling Cpk fell from 1.70 to 1.04; every batch is still within specification.
- 3Recommend: open a deviation, trend MCC particle size for the next 10 batches, no impact on batches already released.
QA reviews the case and signs
Approved · A. Mehta, QAQuickflow CPV at a glance
Continued process verification: batch data charted and trended as each batch closes, with drift explained.
- Who uses it
- QA scientists
- Process owners
- Manufacturing sites
- CDMOs
- What is included
- Control charts
- All eight Nelson rules
- Cp, Cpk, Pp and Ppk
- Multivariate analysis
- Golden-batch comparison
- APQR-ready reports
- Works with
- eBMR / MES
- LIMS
- Historian
- ERP
- eQMS
- APQR
- Built for
- FDA Process Validation (2011)
- ICH Q8, Q9, Q10
- EU GMP Annex 15
- 21 CFR Part 11
- ALCOA+
The FDA’s three-stage lifecycle
Validation does not end at PPQ. Stage 3 never ends.
The FDA’s 2011 guidance splits process validation into three stages. The third, continued process verification, is the ongoing proof that a commercial process is still under control. It is also the stage most sites still run on spreadsheets.
Stage 1
Process design
Build the scientific understanding: which parameters matter, and how far they can move.
- Pick the critical parameters and attributes
- Set the design space and the safe range for each parameter
- Assess the risks and decide how each one is controlled
- Plan what Stage 3 will monitor
Stage 2
Process qualification
Prove the design reproduces at commercial scale.
- IQ, OQ and PQ of equipment and facility
- PPQ batches
- Initial capability indices
- Control strategy confirmed
Stage 3
Quickflow CPVContinued process verification
Keep proving it, batch after batch, for the life of the product.
- Every critical parameter and attribute watched and trended
- Statistical process control charts
- All eight Nelson rules checked
- Cp, Cpk, Pp and Ppk tracked over time
- Alerts when the process drifts
- Evidence straight into the APQR
Stage 3 feeds the APQR. The same charts, capability figures and conclusions go into the APQR, so nobody rebuilds a year of data in December.
Where CPV breaks today
The quarterly scramble: export, paste, chart, repeat
On most sites, Stage 3 means a QA scientist pulling data from five systems into a spreadsheet, drawing control charts by hand and writing the same report every quarter. Drift hides in the gaps.
When is a drift found?
One product, 40 batches, a shift from batch 19. Illustrative.
Spreadsheet CPV
Quickflow CPV
- In control
- Drifting, unseen
- Drifting, being charted
- Under investigation
Data scattered across systems
CPP and CQA values sit in LIMS, eBMR, ERP and spreadsheets. Someone exports, pastes and reconciles them by hand.
Drift found after the fact
Without live monitoring, a shift shows up after a failed batch or at the annual review, not when it starts.
Different statistics everywhere
Each site and analyst uses a different tool, limit rule and chart. Results are hard to compare or defend.
Subtle trends missed
Manual reviews miss the quiet patterns: runs, trends and slipping capability that the Nelson rules exist to catch.
CPV and APQR in silos
Continuous monitoring and the annual review use different data pulls, so conclusions do not always match.
Weak link across stages
Stage 1 design limits and Stage 2 PPQ capability are not traceable into Stage 3 monitoring.
Reports eat QA time
Charts, summaries and narratives are assembled by hand every quarter, pulling QA away from investigations.
Hard to show control in an audit
Static snapshots and scattered files make it hard to show an inspector the process is still under control.
How it works
5 steps. Every one recorded.
- Step 1Collect CPP and CQA dataSystem
- Step 2Chart and apply Nelson rulesSystem
- Step 3Explain the signalSystem
- Step 4QA reviews and signsHuman decision
- Step 5Feed the APQRSystem
Real-time monitoring
Every CPP and CQA, on every batch, as it closes
Values arrive from eBMR, LIMS, the historian and ERP the moment a batch step or test is complete. Nobody exports a file. Each parameter is charted, checked and compared with the batches before it.
What the patient receives: the results that prove each batch is right.
- CQA-01Assay99.1%
- CQA-02Content uniformity, AV3.9
- CQA-03Dissolution, 30 min91.0%
- CQA-04Total impurities0.34%
- CQA-05Hardness120.3N
- CQA-06Water content2.1%
Batch-wise trending
A trend line for each parameter and each yield, updated batch by batch without anyone drawing it.
Alerts that escalate
OOS, OOT and drift alerts reach QA and the production team, and escalate if nobody picks them up.
Golden batch benchmarking
Mark your best batches and compare every new batch against them, parameter by parameter.
Many products, many sites
One view of every product and site, from the portfolio down to one parameter’s chart.
Statistical process control
The statistics, done the same way every time, on every site
Individuals, moving-range, X̄ and R charts are drawn for each critical parameter and attribute, with its own control and specification limits marked. No CSV exports, no desktop statistics package, no chart that only one analyst knows how to rebuild.
All eight Nelson rules, on every point
Each violation is flagged the moment it appears and opens an investigation, with the chart and the rule attached.
Rule 1: One point beyond 3σ
A single result far from the rest: a special cause.
A Digital Scientist for Stage 3
It reasons, you approve
Agentic AI would draw the charts. A Digital Scientist reads them: it works out why a process moved, builds the case with the evidence and recommends what to do. A named QA scientist reviews it and signs. It never signs, and it never releases a batch.
1. Monitor and collect
Digital ScientistProcess and quality values flow in from the systems that produce them.
2. Analyse
Digital ScientistPatterns, anomalies and predicted drift over many batches and products.
3. Surface signals
Digital ScientistEach signal comes with the evidence and a recommended action.
4. Review and decide
QA scientistA named QA scientist reviews the case and signs.
5. Improve
Digital ScientistConfirmed findings improve monitoring, investigations and the APQR.
Predicts drift
Learns from batch history to warn that capability is slipping before quality is affected.
Explains the trend
Interprets the statistics, traces the likely root cause and recommends the next action.
Finds anomalies
Spots unusual multivariate patterns that single-parameter control charts can miss.
Reads paper records
Pulls values out of scanned batch sheets, handwritten logs and old files.
Answers in plain language
Ask about any product, parameter or period and get the answer with the records behind it.
Writes the narrative
Drafts each section of the CPV report with conclusions. QA edits, approves and signs.
Every recommendation, edit and signature is in the audit trail. See the Digital Scientist Trust Charter.
Connected
One source of process truth, fed by every system
CPV is only as good as its data. Quickflow CPV reads from the systems where the numbers are made and writes its findings back to where the work gets done.
- APQRCharts, capability figures and conclusions go straight into the yearly product review.
- eQMSDeviations, CAPAs, OOS and OOT linked to the CPV finding that raised them.
- LIMSTest results, stability pulls and laboratory incidents.
- eBMR / MESValues recorded at each process step, plus yields.
- eDMSThe approved SOP or protocol behind every limit and method.
- eLogbookWho used which equipment and room, when, and for what.
- CalibrationWhether each instrument was calibrated and each machine maintained.
- Cleaning validationSwab and rinse results from cleaning studies.
- ERPWhich material lot and supplier went into each batch.
- HistorianTime-series process data from SCADA and PLCs.
Already on Quickflow? Quickflow eQMS, Quickflow eDMS and Quickflow eLogbook connect out of the box.
What changes
From reactive checking to proactive control
Quality, compliance and efficiency move together when the evidence builds itself.
Proactive quality control
Catch drift before it becomes a failed batch, not after.
Far less manual effort
Data collection, charting and report writing run on their own. QA gets back to investigations.
Real-time visibility
Live process health for every product and site, down to one parameter’s chart.
No more manual stats
Replace CSV exports and desktop statistics packages with analysis that is automatic, audited and current.
APQR without the rebuild
The same data and methods flow from Stage 3 into the annual review.
Faster batch release
Monitoring and reviews already done shorten the path from review to release.
Always inspection-ready
Tamper-proof audit trails, e-signatures and report templates, ready any day an inspector arrives.
One method, every site
Harmonised CPV methods and reports across sites, with room for local data and SOPs.
Regulatory compliance
Built for the guidance inspectors quote
The controls and records each regulator expects are part of the application, not a validation project of their own.
FDA
Process Validation guidance (2011)
Everything the guidance expects of Stage 3, from monitoring through to the report.
ICH
Q8, Q9 and Q10
Development knowledge, risk-based decisions and a working quality system, joined up.
EU GMP
Annex 15
The EU rules for qualifying and validating, including verification that never stops.
21 CFR
Part 11
Trustworthy e-records, compliant signatures, complete audit trails and controlled access.
MHRA
GxP data integrity
ALCOA+ applied from data capture to the final report.
WHO
Technical Report Series
The WHO’s expectations for validating a process and keeping it verified.
Industries
Wherever a validated process has to stay validated
Across life sciences manufacturing, from API to finished dose, and from biologics to devices.
Pharmaceuticals
Trending and batch comparison for tablets, capsules, liquids and injectables.
Biotechnology
Biologics and biosimilars, where many assays and a complex process must be trended together.
API manufacturers
Capability and impurity profiles followed through each synthesis step.
CDMOs
Monitoring and report templates per client and product, kept separate.
QC laboratories
Lab results analysed across many variables, for development and commercial supply.
CROs
Data for each sponsor kept apart, with its own analytical trends.
Medical devices
Manufacturing performance trending linked to design controls.
Vaccines and steriles
Environmental monitoring and aseptic process trending alongside the batch.
Full overview: Continued process verification, Stage 3 of the FDA lifecycle
Process validation has three stages. Stage 1 works out which parameters and attributes matter, Stage 2 proves the process at commercial scale with PPQ batches, and Stage 3, known as continued process verification, keeps proving it for as long as the product is made. The FDA expects that proof to be continuous, statistical and documented.
Quickflow CPV takes process and quality values from eBMR, MES, LIMS, the historian and ERP as each batch closes. It draws individuals, moving-range, X̄ and R charts for each parameter, checks the eight Nelson rules on every new point, and trends capability (Cp, Cpk, Pp, Ppk) over time. Six-pack views, golden-batch comparison and multivariate analysis are built in.
A Digital Scientist reads the statistics. When a process moves, it traces the likely cause across connected records, sets out the evidence and recommends what to do next. It drafts the narrative for the CPV report and carries the same data into the Annual Product Quality Review. A named QA scientist reviews, approves and e-signs every conclusion, and every step is in the 21 CFR Part 11 audit trail.
The application is built around the FDA 2011 guidance on process validation, EU GMP Annex 15, ICH Q8 to Q10, MHRA expectations for data integrity, WHO guidance and 21 CFR Part 11, and it lets several sites run CPV the same way with one set of methods and report templates.
Quality
Related applications
See Quickflow CPV on your own process.
Tell us how this runs in your organisation today and we will walk through the application against it, with your QA and IT teams in the room.