Table of contents
At a glance
How does limited experience affect the assurance needed for a new function?
Start with the intended use and possible failure consequences. Identify which behaviour has been demonstrated and where experience or reliable data are lacking. Use those gaps to define targeted tests, limits on use and criteria for later expansion. Novelty alone does not determine a GAMP category or a standard test scope. For GMP applications, Annex 11 and risk-based quality decisions inform the regulatory assessment.
- Document separate uncertainty from severity of deviation and existing controls.
- Supplier evidence refers to the actual new function and environment.
- Build experience with verifiable boundaries and clear criteria for expansion decisions.
1. Identify what is actually new
A new version number does not automatically imply a new technology. Conversely, a minor enhancement may open a previously unexamined decision pathway. Describe the change at the functional level: input data, processing, output, user, and downstream decision. Also note which existing controls are waived or whose effectiveness is altered.
For an SME, a specific question helps: Would the previous evidence recognize the same error in the new process? If manual sighting is replaced by automatic sorting, for example, the sorting function is not only relevant. Also, the visibility of deferred cases and the responsibility for their processing change. These effects belong to the consideration, even if the platform is already used for a long time.
2. Separate GAMP methodology from applicable requirements
The public version of the second edition of GAMP 5 emphasizes critical thinking, appropriate methods, tools, and the handling of emerging technologies. GAMP 5 is not a regulatory requirement and does not prescribe any specific document template as binding. Public access does not replace the complete guide when your organization intends to implement its detailed methodology formally.
For the human medicines GMP use considered here, Annex 11 describes a risk-based life cycle and a justified scope of validation. ICH Q9(R1) addresses uncertainty and the appropriate formality of risk management. A new product name alone determines neither a risk score nor a software category.
3. Prepare a knowledge gap matrix
The following internal guidance document clarifies frequently conflated issues. A well-documented failure with serious consequences requires effective controls. An unclear failure rate demands investigative insight. A supplier claim not yet substantiated requires a credible source or independent verification. A single traffic light color is insufficient to clearly distinguish these differences.
Draft each gap so that it can be closed. For example, "insufficient experience" becomes "behavior in the case of two simultaneously processed cases is unknown". Assign a responsible party and specify an expected outcome. The table is not a prescribed GAMP template.
| Aspect | Already known | Still open | Next test result |
|---|---|---|---|
| Data entry | Current format inspection confirms | How does import behave with empty mandatory fields in the new source? | Targeted import cases with missing values |
| Processing | Individual processing traceable | Simultaneous changes unknown | Competing processing with state comparison |
| Decision | Manual inspection planned | Are pending cases visible within the actual workflow process? | Check visibility and workflow process |
| Operation | Support pathway available | Response to new error class unclear | Accountability and escalation procedures |
4. Select evidence appropriate to the open question
Supplier documentation may include relevant experience and testing. Verify version, configuration, data requirements, and claimed functionalities. A robust assessment of the actually used module can provide greater value than an extensive list of certificates. Document which questions have been addressed and which deviations remain relative to your intended use.
Additional internal testing should address this gap. Repeating the same simple case does not necessarily provide more insight into an unknown edge case. Combine targeted edge case testing, observation of typical work processes, and independent result evaluation. For hard-to-detect errors, identifiable disruptions and a practical rollback pathway are particularly important. Scope and depth remain justified project decisions.
5. Example: Recommendations for Document Assignment
A team wishes to implement a new function that assigns incoming documents to existing procedures. The function may initially generate suggestions only; formal acceptance by an authorized personnel remains a prerequisite for binding assignment. This boundary is explicitly stated in the example and must be verified both technically and organizationally.
The internal audit check includes 30 pre-prepared documents: 18 clearly assignable, eight deliberately ambiguous, and four lacking process identification. The expected response follows the previously defined requirement, not a subsequently selected success rate. The figures do not constitute a statistical evidence of effectiveness.
| Group | Count | Expected behaviour | Additional inspection point |
|---|---|---|---|
| Unambiguous | 18 | Suggest matching option; no automatic booking | Acceptance and process traceability |
| Ambiguous | 8 | Ambiguity visible; no hidden pre-selection | Selection and rejection possible |
| Identification missing | 4 | Clarification required; no established assignment | Dataset remains retrievable |
| Full reconciliation | 30 | 30 inputs in the traceable processing inventory | 18 + 8 + 4; no losses or duplicates |
6. Define usage boundaries prior to pilot deployment
A pilot is not a general exemption from required controls. The key factor is whether the function is already being used for regulated decisions. Therefore, identify the allowable data, users, decisions, and transitions. A purely observational parallel operation has different implications than a pilot whose results govern actual processing.
In the example, a later automatic assignment must not be activated unnoticed through a setting change. The extension alters the intended use and impacts the remaining human oversight. Prior to such changes, the affected requirements, potential consequences, and validation evidence must be re-evaluated. A robust observation phase provides relevant insights but does not substitute the decision regarding the expanded functional scope.
7. Leverage operational experience with traceable criteria
Define the observations that trigger a re-evaluation: undetected misclassifications, unprocessable inputs, new data formats, or a changed user group. Capture not only the number of incidents but also the pathway by which they were discovered. A small number of reported defects may indicate either excellent performance or poor detectability.
The review should lead to a decision: retain the limit, improve the control, perform additional testing, or extend the function. For a small team, a clear list containing findings, impact, action taken, and closure evidence is often sufficient. The scope of review is determined by the system and its risk profile; a one-size-fits-all monthly or annual interval is not justified.
8. Checklist for Justified Decision on Intended Use
Close the evaluation with a specific statement regarding the approved use. A document containing numerous test cases provides little value if it remains unclear under what conditions and which new function is permitted for use. The checklist below serves as an independent working tool for consolidating the decision.
Open issues should not be eliminated merely by stating "observed during operation." Describe the remaining uncertainty, its significance, and the effective mitigation measures. If the decision-critical function is not adequately supported under the intended operating conditions, the use must be appropriately restricted or deferred until clarification is achieved.
- Scope limited to function, data, and subsequent decision-making.
- Evidence associated with version and usage conditions.
- Gaps in knowledge and error consequences described separately.
- Retestings and expected results defined.
- Pilot limits, rollback path, and responsibility assignable.
- Triggers for further evaluation identified.
- Authorized use and open items documented.
Project experience and collaboration
Identify remaining uncertainty before production use
For a new technology, your process owners define the intended task and the decisions a person will continue to make. A limited pilot with controlled data can reveal unfamiliar failure modes. Expanding the function to several products or sites requires a fresh assessment of what the earlier tests demonstrate. With IT and Quality, we use those findings to decide the next step. Success in a demonstration case does not establish suitability for the whole intended use.
Frequently asked questions about GAMP 5 risk-based assurance for new technologies
Is a new technology automatically considered particularly high risk?
No. Novelty may increase uncertainty, but it says little on its own about the potential consequences of a failure. Assess the function's usage, data impact, and existing controls. A new auxiliary function with no decision impact may be evaluated differently from a known function that performs a critical decision without additional controls.
Does GAMP 5 require Category 5 for new technologies?
Such a general inference is not justified. Product name, novelty, and software category are distinct considerations. Evaluate the components used and the modifications implemented. The scope of validation arises additionally from intended use, risk, and existing evidence; the public ISPE guidance does not replace a complete categorization methodology.
How do we specifically describe the absence of experiential knowledge?
Define a testable unknown: for example, behaviour under parallel processing or in the absence of identification. Elaborate on its significance and the corresponding insight step. "The product is new" serves as a useful starting point, but remains too vague for selecting a specific test.
Does the supplier's validation package suffice?
This depends on the actual scope of coverage. Compare functionality, version, configuration, and intended operating conditions against your specific requirements. Justify the acceptance of appropriate evidence and proactively address any remaining gaps. Neither a blanket rejection of all supplier test results nor an unverified full transfer adequately addresses the specific usage scenario.
Sources: [1]
May we first conduct only a limited pilot?
A limited use may be justified if its impacts and controls have been adequately assessed. A pilot whose results already influence regulatory decisions requires appropriate safeguards. Identify authorized data and decisions, as well as conditions under which the pilot is stopped or expanded.
Do 30 successful test cases demonstrate reliability?
Not general. The 30 cases in the example deliberately assess selected requirements and boundary conditions. They neither represent a universal minimum number nor constitute a statistical reliability justification. The statement is meaningful only in that each case addresses a relevant question and identifies the remaining uncertainty after the test.
Must each knowledge gap be addressed by a separate test?
Not necessarily. Suitable supplier testing, reliable technical documentation or a demonstrably effective restriction can also contribute. Choose evidence that answers the open question. A general product brochure is usually insufficient to demonstrate the required behaviour of a particular critical function.
When is additional operational experience considered sufficient?
Experience is helpful when conditions, observations, and statement limits are known. A low number of error reports alone is insufficient. Additionally, assess whether errors could be detected and whether the observed data cover the intended future use. An expansion decision requires a justified overall evaluation.
Who should prepare the decision?
Knowledge of the specialist process, technical functionality, and current quality requirements is required. The individuals authorized to make decisions are determined by your organization in accordance with the intended use. A fixed, organization-wide signature combination cannot be derived from the approach described herein.
Sources: [1]
Does this approach also apply outside of GMP?
The work aid can be transferred, but the cited requirements are not automatically applicable. For clinical studies, medical devices, or non-regulated activities, you must establish a separate framework. GAMP is an industry-wide methodology; the regulatory examples specifically pertain to computerized systems used in human medicinal product GMP.
Sources and review date
Source date: 7 October 2026. Regulatory requirements, guidelines, and industry practices must be categorized separately. Tables, figures, and case examples are to be considered standalone reference materials; they do not constitute completed product testing.
- EU GMP Annex 11 – Computerised SystemsRevision January 2011; listed in the human medicinal product section. Principle, §§1–17.
- EMA: ICH Q9(R1) – Quality Risk Management, Corr.2EMA version dated 23 January 2025, Corr.2. This is Q9(R1), applicable in the EU since 26 July 2023. See sections 1–6, particularly 4.3 and 5.1–5.3.
- ISPE: GAMP 5 Guide, 2nd EditionPublic publisher description and edition information; no access to paid full text.
Sources checked: .
Jointly develop the specific evidence process
We support your team in professionally defining a new function, categorizing existing evidence, and making remaining evaluation decisions transparent and justifiable.
- GAMP 5 consulting →
Understand how GAMP 5, EU GMP Annex 11 and 21 CFR Part 11 apply to your systems. We translate the relevant requirements into proportionate validation work and evidence your team can defend in an audit.

