Person-in-Plant Oversight: What a PIP Is Actually For

A person-in-plant, usually shortened to PIP, is a sponsor representative present, physically or virtually, at a contract manufacturer during a defined manufacturing campaign or a critical activity. The role is easy to describe badly and easy to do badly, because it sits right next to two roles it is not: the contract manufacturer’s own quality unit, and an auditor. Done well, a PIP gives a sponsor real-time visibility into how its product is actually being made, catches issues while they are still fixable, and builds the working relationship a long CDMO partnership depends on. Done poorly, it either misses everything that matters because nobody knew what to watch for, or it turns into an awkward standoff where the sponsor’s person is trying to run someone else’s floor.

This is a starting point for the role: what it is for, what to actually watch, and where the line sits.

What a PIP Is For

The core value of a PIP is timing. A batch record review after disposition tells you what happened, days or weeks after it happened, filtered through whatever the record captured. A person on the floor sees the event as it happens, while it can still be corrected, investigated properly, or at minimum documented accurately instead of reconstructed from memory later. That difference matters most during a first campaign at a new site, during a complex or historically failure-prone unit operation, and for products where there is no do-over, an autologous cell therapy batch, a short-shelf-life intermediate, or any batch large enough in value or small enough in volume that losing it is a real problem. This is exactly the moment a technology transfer’s engineering run and early validation batches are most at risk of an unrecorded surprise.

A PIP does not sign batch records, does not release batches, does not direct the CDMO’s operators, and does not substitute for the CDMO’s own quality control and quality assurance functions. Those functions carry the actual GMP accountability for the batch under 21 CFR 211.22 and under the quality agreement between the two companies. The PIP’s job is to observe, verify against what was planned, and escalate through an agreed channel. Confusing that role with the CDMO’s own QA is the single most common way a PIP program goes wrong, in both directions: either the PIP starts acting like a shadow QA unit and steps on the CDMO’s own chain of command, or the PIP treats the visit as a courtesy tour and misses things a trained eye would have caught.

Where PIP Authority Actually Comes From

A PIP’s standing to observe, ask questions, and escalate is not inherent to the role. It is created by contract, and it evaporates the moment a floor supervisor asks “who authorized this” and nobody can point to where that authorization is written down. FDA’s November 2016 final guidance, “Contract Manufacturing Arrangements for Drugs: Quality Agreements,” describes how the parties to a contract manufacturing relationship use a written quality agreement to define which party performs which GMP-relevant activity, and a sponsor’s quality or technical agreement with a CDMO is the natural place to define a PIP program’s scope, the escalation channel, expected response times, and the CDMO’s obligation to grant access and cooperate. A PIP program that runs on an informal understanding between two individuals, rather than language in the governing agreement, has no documented standing once either of those individuals leaves.

This is also where the sponsor’s underlying accountability connects back to the PIP program specifically. The sponsor’s quality unit retains responsibility for a product’s quality regardless of where it is manufactured, consistent with ICH Q10 Section 2.7’s principle that outsourcing the work does not outsource the responsibility. A documented PIP program, with its authority traceable to a real quality agreement rather than assumed, is one of the concrete ways a sponsor demonstrates it is exercising that responsibility instead of simply trusting the contractor. It is the same governance instinct behind assigning roles by function rather than by name in a Validation Master Plan: the program has to survive the departure of any one individual on either side of the relationship.

Currency: What FDA’s August 2026 CGT FAQ Guidance Means for the Floor

FDA finalized “Frequently Asked Questions, Developing Potential Cellular and Gene Therapy Products” in August 2026 (docket FDA-2024-D-4311, first issued as a draft in November 2024), and two of its answers land directly on what a PIP watches. The guidance draws a sharp line between characterization testing, which generates information about the product and does not need predefined acceptance criteria, and release testing, which is part of the specification, must have predefined acceptance criteria, and must comply with current good manufacturing practice. On the floor, that distinction shows up as two different sampling events that can look identical to an untrained eye: a technician pulling a vial at a step in the process. A PIP who cannot tell which sampling event just happened, a characterization pull headed for a research assay with no bearing on batch disposition, or a release pull headed for the specification, cannot verify that the batch record correctly labels which is which, and a mislabeled sample is exactly the kind of gap that surfaces during a data integrity review months later rather than on the floor when it could still be corrected.

The same guidance answers a second question a PIP program depends on without always naming it: how manufacturers should evaluate comparability of pre- and post-change products. When a CDMO implements a manufacturing change mid-relationship, a raw material substitution, an equipment swap, a process parameter adjustment, the PIP’s job during that first post-change batch is not just the usual floor watch. It is verifying that the change was implemented exactly as the approved change control describes, that the comparability sampling plan the change control specifies actually gets executed at the right steps, and that any deviation from the as-approved change gets flagged the same shift rather than absorbed into “close enough.” See CDMO deviations and change control for how the change control record itself should be structured before it ever reaches the floor. That real-time comparability evidence is also what eventually supports the narrative a CMC author writes into a filing; see writing the CMC validation summary a reviewer can follow for how floor-level consistency evidence turns into the language a reviewer actually reads.

Deciding When You Need One

Not every batch at a mature, well-performing CDMO needs a person on site. Presence should be risk-based and it should scale down as trust and performance history build, ideally through a graduated model written into the quality or oversight agreement rather than decided fresh each time.

As a campaign proves itself over successive batches with a clean deviation trend, oversight intensity can reasonably step down one row at a time, provided the decision is documented, tied to the actual batch history, and not just assumed because nothing has gone wrong yet.

Portfolio-Level PIP Resourcing: When You Have More CDMOs Than PIPs

The risk table above answers whether a single campaign needs a PIP. It does not answer the question most oversight programs actually run into once a sponsor is manufacturing at more than two or three CDMOs at once: which campaigns get covered when there are more risk-flagged campaigns running concurrently than there are qualified PIPs to staff them.

Ad hoc allocation, whoever asks first, or whoever the program manager happens to think of, quietly turns a documented risk-based program into an undocumented availability-based one, and that gap is exactly what an inspector or auditor finds when they ask why a known-risk campaign at one site went unstaffed while a routine campaign at another got full coverage.

A portfolio-level prioritization framework scores every concurrent campaign against the same kind of risk factors that already drive the single-campaign decision, then ranks the list and allocates capacity top-down, with the cut line and its rationale documented rather than assumed.

Score each active campaign, rank the list, and draw the coverage line where capacity actually runs out. Every campaign below that line gets a documented rationale, stepped down to virtual oversight or, in a genuine capacity crunch, an explicit accepted-risk decision signed by sponsor quality leadership, not a silent gap nobody chose. The PIP Portfolio Coverage Prioritization Matrix carries this decision, campaign by campaign, so a reviewer sees a ranked, dated allocation rather than a program manager’s memory of what seemed most urgent that week.

The line itself should move. A capacity shortfall that recurs every quarter is a staffing gap, not a prioritization exercise to keep re-running with the same headcount; a portfolio matrix that consistently cuts real risk-tier-one campaigns is the evidence that justifies the next PIP hire, not a permanent operating mode to accept indefinitely.

Virtual vs On-Site Models

On-site presence buys things a camera cannot: real-time observation of gowning and aseptic technique, a feel for material flow through the room, the environmental cues that don’t show up in any record, an unusual noise, an unplanned pause, a tone change in the room when something goes wrong, and the ability to have a spontaneous conversation with an operator instead of a scheduled one.

Virtual PIP relies on a live video feed of critical steps, remote review of the electronic batch record as it is completed, and scheduled check-ins with the site team. It is a legitimate model for lower-risk campaigns, for situations where travel is not practical, or as steady-state oversight once a site relationship is established and performing well. It has a structural limit worth stating plainly: a virtual observer only sees what the camera is pointed at, cannot hear or smell the room, and depends entirely on the CDMO’s choices about what to show. Treat virtual oversight as a real tool with a real limit, not a lesser version of the same thing.

A hybrid model, on-site for the highest-risk unit operations, viral inactivation, fill, potency-critical process steps, and virtual for the rest of the campaign, is often the most efficient answer. Whichever model is used, the choice should be a documented decision tied to the risk assessment for that campaign, not something improvised visit to visit.

Data Integrity in Virtual and Remote Oversight

Virtual oversight introduces a category of risk an on-site model never has to manage: the observation itself now travels through a network connection, a screen-sharing session, or a remote electronic batch record login, and every one of those hops is a place ALCOA+, attributable, legible, contemporaneous, original, accurate, plus complete, consistent, enduring, and available, can quietly fail without anyone on either end noticing until much later. A sponsor that stands up a virtual PIP program by pointing a webcam at a fill line and calling it done has not actually built oversight; it has built a stream of unverified footage with no defined retention, no access control, and no answer for what happens when the connection drops mid-step. Scoring the virtual oversight channel itself against each ALCOA+ element, rather than assuming the underlying electronic batch record system’s own validation covers the added network and screen-sharing hop, is what turns that list of nine letters into a checkable basis; a ready-to-use worksheet for that scoring exercise is at ALCOA+ data integrity risk assessment worksheet.

Why this needs its own qualification, separate from the PIP’s own training. The video feed and the remote electronic batch record connection are technology, not judgment, and technology used for a GMP purpose needs the same before-you-rely-on-it verification any other computerized system gets, proportionate to its role. The feed itself is rarely a retained GMP record in its own right; what matters more is that the PIP’s own contemporaneous notes, built from watching it, are trustworthy, and that depends entirely on the PIP actually seeing what happened, unambiguously, in real time, through a connection that was tested before the campaign, not discovered to be inadequate during it.

What the setup has to get right before the first virtual batch. Four things, decided and tested before go-live, not assumed because the vendor’s marketing page says the platform supports GMP use. Camera placement and resolution have to actually resolve the specific action the PIP is there to verify, a label check or a line-clearance confirmation needs a close, steady shot, not a wide angle that technically includes the area. Electronic batch record access granted to the PIP has to be read-only; a remote reviewer who can edit the very record they are supposed to be independently observing has erased the independence the whole model depends on, and that access itself needs to generate its own audit trail entry, showing who viewed what and when, the same expectation 21 CFR Part 11 places on any other electronic GMP record access. Network latency and feed stability need a real test under production conditions, not a quiet office demo, because a feed that lags by even a few seconds during a fast manual step can make the PIP believe they observed something they did not actually see in time to catch a deviation. And a defined response to a feed interruption has to exist before the first batch runs: does the campaign pause until the feed is restored, does the step get re-verified afterward against the electronic batch record, or does the interruption itself get logged as a documented gap in oversight coverage. Silence on that fourth point is the most common gap, because it is the scenario everyone assumes will not happen until it does.

Worked example. During a virtual-oversight fill step at a CDMO, the video feed drops for six minutes right as the line begins a planned changeover between two fill volumes. The PIP, following a documented interruption protocol agreed with the CDMO in advance, immediately calls the floor quality representative to pause the changeover until the feed is restored rather than allowing it to proceed unobserved, then logs the interruption itself, start time, restoration time, and what step was affected, as a documented oversight gap in the visit report, independent of whether the changeover itself later turns out to have gone correctly. Without that protocol, the same six minutes would have passed as if nothing happened, and the batch record would show a changeover that occurred with no independent verification at all during the exact window a camera was supposed to be watching it, indistinguishable after the fact from a changeover that was properly observed.

Roles. Sponsor quality defines the access and retention requirements a virtual oversight platform has to meet before it is used for any GMP-relevant campaign. The CDMO’s information technology function, or its third-party platform vendor, implements and maintains the feed and the access controls, and is accountable for the audit trail showing who accessed what. The PIP is accountable for treating a feed interruption as a real event requiring a real response, not a technical annoyance to note in passing, and for confirming access is read-only before the first login of every new campaign, not assumed unchanged from the last one.

Shift Handoff and Continuity of Observation

A single PIP cannot watch a campaign that runs around the clock. Cell therapy processing, a multi-day fill campaign running two or three shifts, or a bioreactor harvest that does not pause for a convenient end-of-day all create the same structural problem: the moment of highest risk during a multi-shift campaign is often the handoff itself, the window where one PIP’s attention is winding down and the next PIP’s attention has not yet picked up the thread, on a floor that does not stop moving just because the person watching it changed.

Why this needs its own written step, not an assumed continuation. A PIP who arrives for a shift with no structured handoff is starting cold: unaware of which open items are already in flight, what verbal commitment a CDMO supervisor made three hours ago that has not yet been logged, or what step the batch was actually at when the prior PIP left. The gap is invisible in the record because nothing shows a handoff was skipped; it only shows up later as an item that never got escalated, a commitment nobody tracked, or two PIPs each assuming the other covered the same window.

What a handoff has to capture. Four things, transferred the same way every time rather than reconstructed from memory in a hallway conversation: the batch’s current step against the master batch record, every open item already escalated and its current status, any verbal commitment made by CDMO staff that has not yet been converted into a tracked CAPA or tracker entry, and anything observed that did not rise to a formal finding but is worth the next PIP watching for.

How to run it without creating a coverage gap of its own. A brief overlap window, even fifteen minutes of both PIPs present, physically or by call, closes the handoff gap far more reliably than a written note passed through a third party, because it lets the outgoing PIP answer a question the written note did not anticipate. Where an overlap window genuinely is not practical, a structured handoff log completed before the outgoing PIP leaves the floor, not after, stands in, provided the incoming PIP reviews it and confirms receipt before assuming coverage.

Acceptance criteria. A handoff is complete when the incoming PIP has reviewed the batch’s current status against the master batch record, every open item’s status and owner, and every unlogged verbal commitment, and has signed the handoff log confirming receipt before the outgoing PIP leaves the site or ends the call.

Worked Example: A Commitment That Almost Fell Through the Gap

During an overnight cell therapy processing campaign, the day-shift PIP observes a brief pause in the closed-system transfer step and is told verbally by the CDMO’s shift supervisor that the pause was a planned equipment check, not a deviation, and that documentation of the check will be added to the batch record by end of shift. The day-shift PIP mentions this during a 15-minute phone handoff, but it is not yet in the escalation log because the day-shift PIP’s own written report is not due until the next morning. The night-shift PIP, now aware of the open commitment, checks the electronic batch record at the start of the shift and confirms the equipment check documentation has not yet been added; a prompt to the CDMO’s night quality representative gets it entered before the batch disposition review, rather than the gap surfacing for the first time at that review with no PIP left on the account who remembers the context. Without the handoff, the night-shift PIP has no reason to check for an item they never knew existed, and the commitment would have depended entirely on the CDMO remembering it on its own.

Roles. The outgoing PIP owns completing the handoff log before leaving the floor or ending the call, with no exceptions for a quiet shift. The incoming PIP owns reviewing it and confirming receipt before assuming coverage, and escalating immediately, not waiting for the next scheduled report, if the handoff surfaces an item that needs attention before the next observation window. Sponsor technical operations decides, before the campaign starts, whether multi-shift coverage uses an overlap window or a structured log alone, documented in the campaign’s oversight plan, not improvised the first time a campaign runs past a single shift.

The PIP Shift Handoff and Continuity Log is built to carry exactly these four fields between PIPs, so a reviewer sees a documented, signed transfer of coverage rather than an assumption that nothing happened in between.

What to Watch on the Floor

A short list of things to watch consistently does more good than an unstructured wander through the suite.

Deviations in real time matter most. When something unplanned happens, does it get written up at the time, in the paper record or the electronic batch record, or does it get handled verbally with documentation added later, or not added at all? Contemporaneous documentation, recording the event when it happens rather than reconstructing it afterward, is the single best predictor of whether the rest of the record can be trusted.

Batch record discipline is the second thing to watch. Entries should be made at the time of the action, not batched up and filled in at the end of a shift. Corrections should follow a proper convention, single line through the original entry, initialed, dated, with a reason where required, never obscured or written over.

Segregation and material flow deserve a direct look: are different products or lots physically separated the way the facility’s own procedures say they should be, and is line clearance actually performed and documented before a changeover, rather than assumed because the last product is “obviously” gone.

Gowning and aseptic technique, where relevant, should match what was actually qualified and trained, not a shortcut version that shows up on a night shift when fewer people are watching.

Equipment status is worth a direct check that sounds too obvious to need saying and is a real recurring finding anyway: does the equipment actually running the batch match what the batch record says should be running, and is status labeling current.

Environmental monitoring excursions are worth watching as they happen, not just reading about weeks later in a summary report, because the real-time response to an excursion says more about the site’s discipline than the final report does.

Worked Example: What Real-Time Presence Actually Catches

Acme Therapeutics has a PIP on site at Meridian Fill Services for the third batch of a monoclonal antibody fill campaign. During a planned changeover between two different fill volumes of the same product, the PIP notices the line clearance checklist gets signed complete before the changeover crew has actually finished clearing the filling line’s tubing set, a gap of about five minutes between the signature and the physical clearance.

Flagged in real time to the site’s floor quality representative, this becomes a documented, contemporaneous deviation, investigated the same shift, with video review confirming the prior product’s tubing set was still in place when the record was signed. The corrective action, a hard stop requiring a second person to verify clearance before the signature is entered, not after, went into the batch record for that same campaign.

Found instead during a batch record review three weeks after disposition: the same gap would have been indistinguishable from a documentation timing error, if noticed at all, because nothing in the paper record shows the line was ever actually out of clearance. The batch would likely have already shipped. The value of the PIP here was not catching a batch failure, the batch itself was fine, it was catching a systemic gap in the site’s line-clearance discipline before it caused one.

Roles and Responsibilities

A PIP program crosses two companies’ organizations, and the fastest way to blur accountability is to leave any of these roles undefined.

PIP Training and Qualification

Assigning a PIP is a staffing decision that carries its own quality risk if it gets treated as an afterthought, whoever happens to be free that week. A PIP has to walk into an unfamiliar facility, recognize a deviation the moment it happens, and know when a variation is routine operator technique versus a real gap against the batch record, all without directing the floor. That takes a specific mix of training, not general GMP awareness alone.

Qualification should be documented the same way any GMP-relevant role’s training record is documented, not assumed because the person has done the job before at a different site. A PIP qualified for a monoclonal antibody fill campaign is not automatically qualified for a first-in-class cell therapy campaign at a new CDMO; the product and process knowledge element resets with the program.

PIP Independence and Conflict of Interest

A PIP’s value depends entirely on a floor supervisor, and eventually a reviewer or an inspector, trusting that what got written down is what was actually seen, undistorted by any reason to see it more favorably than the facts supported. That trust is worth checking before training, because training cannot fix a program built on a PIP who has a reason, disclosed or not, to go easy on the site being watched.

Why this needs its own screening step, separate from qualification. Qualification asks whether someone can recognize a deviation. Independence asks whether, having recognized one, they would actually write it down. A PIP who recently worked for the CDMO being observed, who has a close personal or family relationship with staff on that floor, who holds a financial interest in the CDMO, or who reports, directly or through an incentive structure, to someone under schedule or cost pressure to keep the campaign looking clean, carries a bias that a technically perfect training record does nothing to address.

What to screen for before assignment. Recent prior employment at the specific CDMO within a defined lookback period is the clearest disqualifying or at minimum disclosure-triggering fact; a personal or family relationship with site staff, a financial interest in the CDMO beyond an ordinary index holding, and a reporting line where the PIP’s own performance review is influenced by the campaign’s schedule or cost outcome all belong on the same screening form, completed and signed before the first assignment to a given site, not assumed unchanged from the last one.

How disqualifying is decided versus managed. Not every disclosed fact disqualifies; some are manageable with a documented mitigation. A table of common scenarios, decided in advance rather than argued case by case under schedule pressure, keeps the line consistent.

Acceptance criteria. An assignment meets the bar when the screening form is completed and signed before the first day of the assignment, every disclosed fact is dispositioned against the table above or an equivalent documented rationale, and any disqualifying fact results in reassignment rather than a waiver granted under schedule pressure.

Roles. The PIP discloses fully and proactively, including a fact that only becomes relevant partway through a long-running relationship with a CDMO, such as a new family hire at the site. Sponsor quality leadership, not the technical operations function scheduling the campaign, makes the disqualify-or-manage decision, specifically because the function with the most schedule pressure to keep a PIP in place is the one least suited to deciding whether a conflict is acceptable.

Common inspection framing. An otherwise accurate PIP report, once it comes out during an inspection or an audit that its author had an undisclosed prior employment relationship with the CDMO, does not get re-litigated line by line; the program’s credibility takes the hit regardless of whether any specific finding in that report turns out to have been wrong. Disclosure before assignment is the only defense that holds up after the fact.

The PIP Independence and Conflict of Interest Disclosure carries this screening, signed before assignment, the same way the campaign readiness checklist carries the rest of pre-campaign qualification.

Being Useful Without Becoming a Second QA Unit

The line to hold is simple to state and harder to hold under pressure: observe, document, ask questions, and escalate through the channel agreed before the campaign started. Do not directly instruct an operator to change what they are doing, do not touch product or equipment, and do not override or countermand the CDMO’s own quality assurance function.

This matters for two reasons. The CDMO’s quality unit carries the actual accountability for GMP compliance and batch disposition, and a PIP who starts directing the floor blurs that accountability in a way that can weaken, not strengthen, the record if it is ever examined during an inspection. It also matters practically: undermining a site’s own supervisory chain in front of its own operators erodes exactly the trust a multi-year manufacturing relationship depends on. If something needs to stop immediately, that decision runs through the CDMO’s own line supervisor or floor quality representative, prompted by the PIP’s escalation, not executed by the PIP directly.

After-Action Reviews

Capture observations the same day, while the details are still fresh: what was observed, what matched the batch record and what didn’t, what deviations opened, and any verbal commitments made by CDMO staff on the floor. That last category deserves particular attention, because a verbal commitment is not a corrective action; it needs to convert into something tracked, with an owner and a due date, or it will quietly disappear. Use a consistent structure batch to batch so that patterns across a campaign become visible, not just the details of a single visit. This is also the point in the process where the technical fit questions asked before signing the contract, see technical assessment of a CDMO, either hold up in practice or don’t.

CAPA Tracking to Closure

The trap to name directly: a floor conversation where the CDMO says “we’ll fix that” is not a corrective action. Part of the PIP’s job is making sure every real finding lands in the CDMO’s actual CAPA or quality system, or in the sponsor’s own oversight tracker where the CDMO’s system does not give the sponsor visibility, with a named owner and a due date. Then the finding has to be followed to verified closure, not closure as claimed.

A concrete version of the trap: a CDMO commits, after a PIP flags a slow environmental monitoring excursion response, to retraining the shift on the excursion SOP within a week. The CAPA form closes on schedule with training records attached. That is closure as claimed. It becomes verified closure only when the next excursion, whenever it happens, gets the correct response time on the first try without the PIP having to prompt it again. If the same delay shows up on the next campaign, the training happened and the problem didn’t move, which is exactly the pattern a closure-verification step exists to catch before it repeats a third time.

This discipline is consistent with ICH Q10 Section 2.7, Management of Outsourced Activities and Purchased Materials: the sponsor’s pharmaceutical quality system retains responsibility for the quality of outsourced operations for the life of the relationship, and cannot delegate that responsibility away by delegating the work itself.

Measuring Whether the PIP Program Itself Is Working

A single trip report tells you about one visit. It does not tell you whether the program as a whole is catching what it exists to catch, and a program that never asks that question at the portfolio level can run for years on the comfortable assumption that no news is good oversight, when the actual explanation is that nobody is tracking whether findings ever really close.

A small set of metrics, reviewed on a fixed cadence rather than only when something goes wrong, answers that question with evidence instead of impression.

Review this set quarterly at a level above any single campaign, sponsor technical operations and quality leadership together, and treat two consecutive quarters of decline on any one metric as a trigger for a program-level root-cause look, not just a note for the next individual visit. The PIP Program Metrics and Trend Scorecard is built to carry this rollup on a repeatable cadence, so the program’s own trend is visible the same way a single site’s deviation trend is.

PIP Support During a Regulator’s Inspection at the CDMO

Every PIP program eventually gets tested by the one scenario its documentation is least likely to cover: a health authority inspector, an FDA investigator on a pre-approval or for-cause inspection, an EU qualified person’s auditor, or another agency’s inspector, arrives at the CDMO while the PIP is on site. This is not a bigger version of routine oversight. It is a different situation with different rules, and a PIP program that has never written down what the PIP does in that moment is asking someone to improvise the answer under real pressure, in front of a regulator, with the sponsor’s name attached to whatever happens next.

Why this needs its own written protocol. The CDMO, not the sponsor, is the party under inspection, and the CDMO’s own management and quality unit run the response: hosting the inspector, producing records, answering questions, and closing out the visit. A PIP is a sponsor employee present on someone else’s regulated premises during a live regulatory event, and the underlying quality or technical agreement should state, in writing and before any inspection happens, what that person is and is not authorized to do if the inspector asks a direct question. A PIP who answers a question about the sponsor’s own specifications without the CDMO’s inspection host present, or who volunteers information outside the inspection’s stated scope, can create exactly the kind of inconsistency between what the sponsor’s filing says and what a visitor said on the floor that turns a routine observation into a written inspectional finding for either company.

What the protocol has to decide in advance. Three questions, settled in the quality agreement long before an inspector’s car is in the parking lot, never negotiated in the lobby that morning: whether the PIP may be present in the inspection room at all, whether and how the PIP is introduced to the inspector if present, and who speaks for the sponsor if the inspector directs a question toward “the company that owns this product” rather than the CDMO answering only for its own operations. A defensible default position states that the PIP does not initiate contact with the inspector, defers every question to the CDMO’s designated inspection host, and, if directly asked something, states factually what role the PIP occupies and refers the substantive question back to the CDMO or, for a sponsor-specific matter, to a named sponsor regulatory affairs contact identified in advance for exactly this purpose.

During and after the inspection. The PIP’s role shifts from routine floor observation to real-time internal reporting: documenting which records the inspector requested, which questions were asked, and any commitment CDMO staff made touching the sponsor’s product, then relaying that to the sponsor’s quality and regulatory affairs leads the same day, not at the next scheduled call. A commitment made to a regulator, “we will revise the batch record template by next quarter,” carries more weight at the next inspection than the same commitment made only to a customer, and the sponsor needs to track it to verified closure with the same discipline as any other CAPA rather than assuming the CDMO alone will remember it. The PIP’s after-action report for an inspection week adds three fields to the routine visit report format: any observation the inspector raised that touches the sponsor’s product specifically, any commitment the CDMO made that the sponsor now needs to track, and whether anything in the CDMO’s inspection response appeared to contradict the sponsor’s current approved specification or filing language. A contradiction caught here, while the inspection is still open, is a solvable problem; the same contradiction surfacing for the first time in the inspector’s written observations is not.

Roles. The CDMO’s designated inspection host owns the inspection response in full and is the only party who should be answering substantive questions on the CDMO’s own behalf. The PIP observes, documents in real time, and defers; the PIP does not answer for the CDMO and does not negotiate commitments on the sponsor’s behalf. The sponsor’s named regulatory affairs contact, identified in the quality agreement before any inspection happens, is the only person authorized to speak to sponsor-specific questions if the CDMO’s host asks that contact to join, in person or by phone. Sponsor quality leadership tracks every commitment and observation the PIP relays to verified closure, the same discipline applied to any other CAPA.

Common Inspection Findings in Outsourced Oversight Programs

A pattern of findings shows up repeatedly across sponsors when an inspector or auditor examines how a company oversees its contract manufacturers, independent of any single PIP visit.

• No documented quality agreement covering the manufacturing relationship, or one that is silent on oversight scope, escalation channel, and expected response time.
• A sponsor quality unit unable to produce evidence that it reviewed a contractor’s deviation and CAPA data before releasing product manufactured under a known, open quality concern at that site; see CDMO deviations and change control for how that review discipline is normally structured.
• Batch disposition decisions resting on a contractor’s certificate of analysis alone, with no evidence the sponsor exercised its own technical review of the underlying data.
• A PIP or oversight visit report that exists in a file but shows no trace of its findings ever reaching a real CAPA or quality system, sponsor or CDMO’s, for tracking to closure.
• Escalations resolved verbally on the floor, with no written record, discovered only when the same issue recurs on a subsequent campaign and there is nothing to show it was ever raised the first time.
• Oversight intensity that never changed across many campaigns at the same site, with no documented risk basis for either the original level or the fact that it stayed fixed.
• Oversight capacity allocated across multiple concurrent CDMO campaigns with no documented, risk-based prioritization when the number of risk-flagged campaigns exceeds the number of available PIPs.
• No documented protocol for a PIP’s role during a regulator’s inspection at the CDMO, discovered only when an inspector directly questions the PIP’s presence in the inspection room.
• A virtual oversight platform used for a GMP campaign with no evidence it was qualified before use: access controls untested, no defined response to a feed interruption, or electronic batch record access broader than the read-only scope the role requires.
• A multi-shift campaign with no structured handoff between PIPs, discovered when an open item or a verbal commitment from one shift never reaches the next and surfaces only at batch disposition.
• A PIP assigned with no documented independence or conflict-of-interest screening, discovered when a prior employment relationship with the CDMO or a family connection on the floor comes out after the fact.
• A PIP report containing speculative or opinion-laden language about a specific operator or the site’s general competence, rather than a factual, contemporaneous account of what was planned and observed.

Common Mistakes New PIPs Make

What a Good PIP Report Looks Like

A useful report is short and structured the same way every time. The difference between a report that carries decision value and one that only proves the visit happened is specificity, in the entries themselves, not in the section headings.

The report goes to the sponsor’s technical operations and quality leadership, and, in a form the CDMO can act on, back to the site itself. A report that only exists to prove the visit happened is not doing the job.

PIP Report Records: Retention and the Discipline of Writing Fact, Not Opinion

A PIP’s visit report is a GMP-adjacent business record the moment it is written, and like any business record describing a manufacturing event, it can be requested later by a regulator, by the sponsor’s own quality system during an investigation, or, in the worst case, in litigation discovery if a batch’s quality is ever disputed in a legal proceeding. That possibility should shape how a report is written from the first entry, not get discovered as a problem the day a legal hold notice arrives asking for every PIP report touching a specific batch.

Why fact and opinion need a hard line between them. A factual, contemporaneous observation, what was planned, what was observed, what deviation number was opened, what a named individual said and when, holds up under any later scrutiny because it describes what happened. A speculative characterization, a guess at why something happened, an opinion about a specific operator’s competence, or a conclusion about whether the batch is compliant, does not belong in a PIP report at all, both because the PIP is not the party with disposition authority to reach that conclusion and because that kind of language is exactly what a reviewer in a later dispute seizes on, whether or not it was ever meant as a considered judgment.

What this looks like in practice. The difference is not subtle once it is named, and a single report entry shows it clearly.

The left column survives scrutiny because it states what was observed and leaves the conclusion to the party with authority to reach one, the CDMO’s quality unit for disposition, the sponsor’s quality unit for oversight-program decisions. The right column substitutes the PIP’s own impression for a conclusion nobody asked the PIP to reach, and it reads very differently to a lawyer five years later than it did to the PIP writing it that evening.

Retention and access. A PIP report’s retention period should match or exceed the retention period of the batch record it accompanies, not a shorter, convenient default set independently by whichever system happens to store PIP reports. Access should be limited to sponsor quality and technical operations leadership with a genuine need, not open to the full organization by default, and any legal hold notice received by the sponsor’s legal function suspends normal retention and disposal for every report the hold identifies until the hold is formally lifted, a step that has to be built into the records procedure in advance rather than improvised the first time a hold notice actually arrives.

Roles. The PIP writes fact, not opinion, in every entry, trained to that standard before the first assignment, not corrected after the fact on a report already distributed. Sponsor quality owns the records retention schedule and reviews reports for the fact-versus-opinion line as part of normal report review, not only when a dispute is already underway. Sponsor legal counsel owns issuing and lifting a legal hold and defining its scope when one applies.

Interview Q&A: Person-in-Plant Oversight

How do you decide whether a campaign needs an on-site PIP versus virtual oversight? Start from the risk table: a first campaign at a new site, a recent inspection finding relevant to the product, a complex or historically failure-prone unit operation, or a non-recoverable batch all point toward on-site presence, at least for the highest-risk steps. A mature relationship with a clean deviation trend across multiple campaigns can step down to virtual with periodic on-site spot checks. The decision should be documented and tied to actual batch history, not defaulted to whatever the last campaign did.

A CDMO’s floor supervisor questions why your PIP is asking to observe a changeover. How do you handle it? Point to the quality agreement or technical agreement governing the relationship, where the PIP program’s scope and access should already be defined. If it is not written down anywhere, that is the actual finding, not the floor supervisor’s pushback, and it gets fixed by adding that language to the governing agreement before the next campaign, not by arguing authority on the floor in the moment.

Walk me through closing the loop on a verbal commitment a CDMO makes during a visit. A verbal “we’ll fix that” is not a corrective action. It has to convert into a tracked item with a named owner and a due date, in the CDMO’s actual CAPA system or the sponsor’s own oversight tracker if the CDMO’s system does not give the sponsor visibility. Closure is verified, not claimed, when the next relevant event, the next excursion, the next changeover, shows the fix actually worked, not just that a training record was filed.

What is the actual difference between a PIP and an auditor? An auditor samples a system periodically against a defined scope and reports a compliance conclusion at a point in time. A PIP is present continuously or repeatedly during live manufacturing, watching real-time execution rather than sampling records after the fact. A PIP does not issue an audit report or a compliance finding in the formal sense; a PIP escalates observations through the agreed channel and depends on the CDMO’s own quality system, and the sponsor’s CAPA tracking, to convert those observations into corrective action.

A sample gets pulled during the batch you are watching. How do you know whether it matters to disposition? Ask what the sample is for before assuming. Characterization sampling generates information about the product and carries no predefined acceptance criteria; release sampling is part of the specification, has predefined acceptance criteria, and has to comply with current good manufacturing practice. A strong answer explains that this is not a distinction a PIP can guess at from watching the pull alone; it has to be confirmed against the batch record and the sampling plan, because the two can happen at visually identical steps and only the paperwork tells them apart. A weak answer treats every sample pull as equivalent, which misses that a mislabeled release sample is a disposition risk in a way a mislabeled characterization sample is not.

How would you staff and qualify a PIP for a new program? The strong answer treats PIP assignment as a training and qualification decision, not a scheduling one: current GMP training, product and process knowledge specific to the program, aseptic qualification where relevant, deviation-writing competence, and a documented walkthrough of the escalation channel before day one. It states plainly that experience on a different product or a different CDMO does not automatically qualify someone for this program, because the product and process knowledge element has to reset. A weak answer assumes anyone with general GMP background and free time on their calendar is interchangeable for the role.

You have three qualified PIPs and six campaigns that all score as risk-flagged this quarter. How do you decide who gets covered? Score every campaign against a common set of portfolio factors, risk tier, batch value or patient impact if lost, proximity to a regulatory interaction, site maturity, and PIP qualification match, then rank the list and allocate capacity from the top down. Document the cut line and get a named Quality Assurance sign-off for any risk-tier-one campaign that ends up below it. The failure mode to avoid is deciding by whoever asked first or whoever is easiest to reach, which turns a documented risk-based program into an undocumented availability-based one with no defensible rationale if an inspector asks why a similar-risk campaign at a different site went unstaffed.

How do you know if a PIP program is actually working, not just running? Measure it, the same way a single site’s deviation trend gets measured: visit-to-report turnaround, the percentage of escalated findings that actually convert into a tracked CAPA or tracker entry, the gap between claimed closure and verified closure, and coverage rate against the risk model. A program with clean individual trip reports can still be failing at the portfolio level if closed items keep recurring or risk-flagged campaigns keep going uncovered with no one tracking the pattern across visits.

An FDA investigator arrives unannounced at your CDMO while your PIP is on site. What does the PIP do? The strong answer states that the CDMO, not the sponsor, runs the inspection response, and the PIP’s role in that moment is defined in advance by the quality agreement, not improvised: defer to the CDMO’s inspection host, do not initiate contact with the inspector, and if directly asked a question, answer factually about role only and refer substantive questions back to the CDMO or to a named sponsor regulatory affairs contact. It adds that the PIP’s real job during the inspection is internal, real-time reporting of what was requested, what was asked, and any commitment made touching the sponsor’s product, so that gap can be tracked to closure afterward. A weak answer either has the PIP staying silent and uninvolved with no reporting value, or has the PIP answering questions directly as if representing the CDMO.

Your virtual PIP’s video feed drops for six minutes during a critical fill-line changeover. What do you do? The strong answer treats this as a foreseeable event with a pre-agreed response, not an improvised judgment call: the changeover pauses until the feed is restored, or, if it cannot pause, the interruption itself gets logged as a documented oversight gap regardless of how the changeover later turns out. It names the underlying requirement, that a virtual oversight setup has to be qualified, camera placement, read-only electronic batch record access, feed stability, and a written interruption response, before it is relied on for a live campaign, not discovered to be inadequate the first time something goes wrong. A weak answer either ignores the gap because the changeover appeared to go fine, or has no answer for what “qualified” means for a virtual setup in the first place.

How do you keep oversight continuous across a multi-shift campaign without gaps between PIPs? A structured handoff, not an assumed continuation, transfers four things every time: the batch’s current step against the master batch record, every open item’s status, any verbal commitment not yet logged, and anything worth the next PIP watching for. A brief overlap window where both PIPs are present, even briefly, catches a question a written note alone would miss; where that is not practical, a completed handoff log the incoming PIP reviews and signs before assuming coverage stands in. The failure mode to name is a gap that is invisible in the record precisely because nothing shows a handoff was skipped.

How do you screen a PIP for independence before assigning them to a campaign? Treat it as a separate question from qualification: qualification asks whether someone can recognize a deviation, independence asks whether they would actually write it down. Screen for recent prior employment at the specific CDMO, a personal or family relationship with floor staff, a financial interest beyond an ordinary diversified holding, and any reporting-line pressure to keep a campaign looking clean. A disqualifying fact gets reassignment, not a waiver granted under schedule pressure, and the decision sits with sponsor quality leadership rather than the function scheduling the campaign.

What is wrong with writing “the operator didn’t seem to take it seriously” in a PIP report? It substitutes the PIP’s own impression for a conclusion nobody asked the PIP to reach, and a PIP report is a business record that can surface later in an inspection or in litigation discovery. The defensible version states what was observed, factually and with timestamps and reference numbers, and leaves any conclusion about competence or compliance to the party with authority to reach one. The test is simple: would this sentence still look fair and accurate read back five years later by someone who was not in the room.

Monday morning, if you have an active PIP program right now: pull the last three reports and check whether every open item has a named owner and a due date in a real CAPA system, not just a bullet point in a trip report. That is usually where the discipline breaks down first, and it is the cheapest place to fix it.