Table of contents
At a glance
What evidence do you need before a GxP go-live?
Before go-live, assess the actual process and system version, required functional evidence and operating controls. Evaluate open findings by their consequences. Giving an issue a due date does not replace a necessary control. Authorised decision-makers document the permitted scope, conditions and outcome of the release decision. Include reachable owners, appropriate training and workable incident handling.
- Determine the specific intended use and system status.
- Evaluate each open finding individually based on root cause, evidence of occurrence, and effective control.
- Assess functions alongside recovery procedures, user roles, and operational workflows.
- An go-live date or a green light does not replace a justified decision.
1. Understand go-live as a concrete usage decision
Technically provided, tested, and available for a specific process are distinct statements. A deployment confirms the availability of a system instance. Test results address predefined evaluation questions. The decision to use the system integrates these results with their scope, open findings, and operational readiness.
The framework under consideration is Human Medicinal Product GMP. Annex 11 is the relevant GMP guideline for computerised systems; Annex 15 complements the qualification and validation context and refers to Annex 11 for computer systems. Industry-specific or additional national and FDA-related requirements necessitate a separate assessment. The system usage decision is not the certification of a medicinal product batch.
Before making a decision on who may perform which activity, with which configuration, from when onward, describe the intended use, including all relevant interfaces, data sets, and locations. Any unanticipated functions must be explicitly excluded. The term "system approved" without this contextual reference may imply a greater level of assurance than the available validation evidence supports.
This approach addresses the planned transition of a defined stage into operational status. The general remediation of long-standing missing validation documentation or a renewed supplier selection are separate tasks.
2. Compare the evidence of compliance against the intended use
Present requirements, relevant risks, results, and findings in a way that enables a decision-maker to understand the relationships and dependencies. A simple list of existing documents does not demonstrate that the intended use has been adequately addressed. Highlight specific functions whose failures could lead to incorrect quality decisions or the loss of significant data.
Verify the compliance of documentation: Does the software version, configuration, roles, interfaces, and test data align with the intended use? Passing a test against an older mapping table does not automatically demonstrate correct processing following its modification. Therefore, capture changes between the last review and pre-launch preparation.
Existing supplier documentation may cover appropriate requirements. Assess the actual scope and supplement selectively where gaps are identified for internal use. A document prepared in-house is not inherently stronger; what matters most are credible assertions and appropriate contextual conditions.
A concise conclusion identifies met criteria and remaining limitations. It provides evidence instead of reiterating all raw data. If a result is not assessed, this status remains visible. "Test performed" and "acceptance criterion met" must not be assigned to the same designation.
3. Evaluate open findings based on their actual impact
Separate missing foundation, unverified critical functions, and known defects. In the case of a missing foundation, appropriate existing evidence may be traced. An unverified critical function remains a knowledge gap. A known defect requires an assessment of its impact and, if applicable, the presence of an effective control.
The term "minor" is insufficient. Describe the affected process, the potential consequence, the existing evidence, and the remaining uncertainty. Assess whether multiple individually small findings collectively compromise a critical control point. An average derived from green and red positions would be inappropriate for this purpose.
Practical decision rule: If a necessary critical prerequisite is missing for the intended use, the scope will not be approved. A corrective action with responsible party and defined timeline does not render a missing control effective. For a reasonable residual task, it must be demonstrably justified that the current usage is already adequately supported.
Annex 15 Section 2.10 addresses the authorized progression to the next qualification or validation stage under the conditions specified therein. It does not constitute a blanket authorization for productive GxP start with open critical points.
4. Verify operational readiness against specific actions
A system may pass technical testing yet remain unmanageable in day-to-day operation. Therefore, assess who takes ownership of an incident, evaluates its technical significance, and makes a decision regarding continued use. A general support contact does not necessarily clarify who is responsible for addressing a potentially misinterpreted laboratory result.
Assign a feasible accountable individual and a clear procedural workflow to each required operational process. This may include user access provisioning, configuration changes, monitoring of interface interactions, handling of error alerts, and review of relevant event logs. The scope and frequency shall be risk-based.
Test the intended reporting pathway with a realistic scenario: A system interface reports success, but a decision-critical assignment is missing in the target system. Which user identifies the omission? Who restricts further usage? Where is the evidence retained? The answer should not depend on the random availability of a project team member.
The transfer is practically only considered viable when the operation is able to assume the required documentation, access rights, and decision-making responsibilities. A calendar entry alone does not establish this capability.
5. Backup, Recovery, and Separation of Failure Procedures
A successful backup job is an important operational finding, but it does not constitute evidence that the required data can be accurately restored. According to Annex 11, Section 7.2, the evaluation of integrity, accuracy, and restoreability is included in the validation and subsequent monitoring. Define the relevant data and system scope for this assessment.
This does not automatically imply a complete restoration of all enterprise data into the production environment. Select an appropriate, controlled validation and justify its representativeness. Consider the information whose loss or incorrect assignment would impair the intended process.
Meanwhile, the failure procedure for critical processes is defined. Data restoration may function technically, yet still take longer than the process can tolerate. Therefore, specify what occurs during a failure, how alternative records are maintained, and how their subsequent takeover is controlled.
Before making a decision, the proposed solution must meet the required restart conditions. A vendor's unverified assurance or the statement "backup runs daily" does not address this question. Appropriate supplier evidence can, however, be utilized if the statement actually reflects the required scope.
6. Training, access rights, and data transfer to be coordinated with each other
Identify the user groups that are to be involved at startup. Verify that their training covers the actual duties, critical error scenarios, and new operating procedures. General product walkthrough participation alone is not automatically sufficient. An untrained substitute may compromise an otherwise sound plan.
Compare the assigned roles with the actual access rights. Particular attention should be given to rights related to modifying master data, deleting records, and bypassing technical control checks. Technical accounts required for system startup must also be assigned, as must personal user accounts.
When data is migrated, you must reconcile the values and meanings of the data in the new system. Simply confirming complete dataset counts does not demonstrate correct assignment of unit, status, and sample. Define which data may be used immediately following the transition and specify how incomplete activities shall be handled.
These points are interrelated: The correct dataset provides little value if the responsible personnel cannot locate it or processes it with an inappropriate role. Therefore, verify a complete technical workflow under the intended start conditions.
7. Use a brief decision rationale supported by reliable references
The following table is an internal working aid, not an officially prescribed template. It consolidates the current status for decision-making purposes. Each statement must be linked to the corresponding evidence; detailed protocols and assessments remain at their designated archive location.
The outcome may relate to the release of the requested scope, a decision on a demonstrably bounded scope, or a deferment. A restriction is only viable if it genuinely operates technically and organizationally. The excluded process must not continue to influence the remaining operations undetected.
The responsible personnel and their authorities are defined in accordance with the quality system and the specific organisational structure. Do not create a general list of signatures. However, document the relevant technical and operational inputs considered, as well as who has the authority to make decisions.
| Field | Specific entry | Assessment question |
|---|---|---|
| Scope of Use | Process, System Status, Location, Interfaces | What exactly is being decided? |
| Current evidence | Criteria, Result and Reference Documentation | Does the evidence support the intended use? |
| Open finding | Impact, uncertainty, existing control | Is the current usage already acceptable? |
| Operational readiness | Workflow, responsible party, accessible evidence | Can the operation actually be carried out? |
| Decision | Scope, Prerequisites, Authorized Person, Date | Are boundaries clear and effective? |
| Traceability | Action, Due Date, Completion Criteria | Who verifies the substantive completion? |
8. Example: Laboratory ERP with unverified restoration capability
A pharmaceutical SME has implemented a laboratory ERP system. The sample receipt, result assignment, and status transfers for the requested scope have been demonstrated. The system is intended to maintain the critical electronic laboratory records. To date, evidence of successful data restoration—either through internal capabilities or through supplier documentation—has not been provided.
The project overview indicates that most tests have been completed. Despite this, a required foundation for the intended operation is missing. In this example, the team delays the productive use. A verbally proposed alternate note is not accepted as a substitute, since neither completeness nor subsequent traceability of this workaround has been verified.
The IT and process responsible parties subsequently define the representative recovery scope. In an appropriate testing environment, the required records, relationships, and access rights are verified post-recovery. The responsible parties assess the results; in the course of the example, the previously defined criteria are met.
A remaining display error in an internal reference list will be addressed separately. The required laboratory procedure does not utilize this list, and the rationale will be supported by a clear and traceable comparison. Only now does a new usage decision arise for the designated site. It is no longer the number of open items that has changed, but the significance of those items.
| Finding | Evaluation prior to startup | Action and Decision Outcome |
|---|---|---|
| Technical procedure completed | Meet the tested functions | Link the evidence to the version submitted. |
| Reproducibility unverified | Fundamental basis for critical records is missing | Shift productive utilization; establish appropriate validation evidence. |
| Parallel note only proposed | Efficacy and completeness unclear | Not to be counted as an existing control. |
| Restoration of later recorded data | Fixed criteria in the example are met | Evaluate the results and make a new decision. |
| Help list with representation error | No impact on the requested procedure is demonstrated | Monitor the residual action with closure criterion. |
9. Prepare for system switching and startup verification
Between the decision record and the first productive use, further changes may occur. Define how the actually delivered state is aligned with the decided state. This includes relevant configuration, activated interfaces, released data, and user roles.
Define specific termination and escalation conditions for the cutover. Which unexpected deviation halts the transition? Who determines whether to revert or establish a controlled intermediate state? A technical rollback may become complex once new production records have been generated. Therefore, the handling of these data must be carefully considered in advance.
Perform an appropriate start control for the critical processes. It complements the existing evidence under the actual start conditions, but does not replace a lack of basic validation. If results remain unclear, the use is limited and evaluated according to the agreed procedure.
Maintain records of deviations from the planned transition. A subsequent change to the system must not be implicitly accepted under the previous decision if such a change affects the validity of that decision.
10. Make the remaining work and transfer of operations lockable
Assign a measurable closure criterion to each remaining action. "Improve documentation" is too vague to serve as a closure criterion. "Add missing reference to the previously assessed restore proof and verify the linkage" is specifically actionable and verifiable. The substantive evaluation should remain distinguishable from simple attendance or scheduling records.
The transition to routine operation also includes open incidents, known limitations, and pending changes. Agree on when the special start-up supervision ends and the conditions that must be met for this to occur. A fixed number of days does not align with every system and process.
For SMEs, effort arises primarily when decisions are consolidated only on the eve of implementation. Gather evidence and clarify authorizations already during the project. This ensures the go-live discussion focuses on a known state, rather than an improvised search for documentation.
- Scope and actually provided system availability are clearly defined.
- Critical requirements and results are traceable and logically linked.
- Open findings are individually assessed; required controls are already in place.
- Relevant data, roles, and training align with the launch.
- Failover and recovery procedures have been validated and demonstrated to be appropriate.
- Operation, support, and escalation have identifiable responsible parties.
- Decision authority, results, and limitations are documented.
- Corrective actions have responsible parties, timelines, and closure documentation.
Project experience and collaboration
Test go-live readiness against an actual operating scenario
Shortly before the planned start, we walk through an operating scenario with the people responsible. Can they retrieve the approved version, relevant test results and procedures for incidents, access and changes? We group open points by their effect on this intended use. A missed handover may be recoverable in time. An unresolved critical function needs a wider decision about go-live. We prepare the evidence with QA, the process owners and IT. The authorised decision-maker determines which use is allowed and under what conditions.
Frequently asked questions about GxP go-live
Does a signed validation report suffice for Go-live?
It may contain or support the usage decision, provided its scope and authorized responsibilities are appropriate. The key question is whether it reflects the actual current status, evaluates pending findings, and takes into account required operating conditions. A signature on a simple test summary does not automatically address these questions. Additional justification is only required if the decision otherwise remains unclear.
Must all actions be completed prior to startup, without exception?
Not every residual activity carries the same significance. However, prerequisites necessary for the intended use must be met or supported by appropriate and effective controls. A non-critical, clearly defined and traceable residual activity may be separately tracked. Document its impact and the closure criterion. The scheduling of a measure alone does not justify its use if a current critical basis is absent.
Does Annex 15 Section 2.10 permit a conditional production start?
The function addresses the transition to the next qualification or validation stage and includes a documented assessment of the impact on the subsequent activity. This does not constitute a general product start authorization. For specific GxP use, the required controls and supporting evidence remain decisive. Therefore, an open critical prerequisite cannot be waived solely by referencing this section.
Who must sign off on the go-live decision?
The authority for approval is defined by your quality system. In the context of qualification, Annex 15 specifies the appropriate personnel responsible (No. 2.2) and quality oversight throughout the validation lifecycle (No. 1.3). A possible assignment structure is: the process owner evaluates the technical usage, the system owner assesses technical operational readiness, and QA addresses quality-related concerns. The individual who makes and signs the decision must have actual authority consistent with their role. You must also establish effective delegation in the event of absence.
Can the contractor assume responsibility for the decision on usage?
A third-party provider can deliver agreed-upon testing, assessments, and documentation. Whether a specific decision can be delegated depends on the task, agreement, and actual authority. The provider's technical readiness assessment does not automatically replace the operational decision of the regulated organization. In particular, verify who evaluates and formally accounts for their own process boundaries and remaining risks.
Must a complete production restore be performed for go-live?
A controlled test in a representative environment or suitable reliable supplier evidence can contribute to this. Reason for coverage and limitations. However, the mere success of a backup job does not yet prove that the required data is available again in good time.
Sources: [2]
Can we only take a portion of the system into production?
A defined scope of use can be individually assessed if it is genuinely separable from a technical and scientific standpoint. Review shared data, roles, interfaces, and dependencies. The excluded process must not impact the authorized workflow. Document and communicate the effective boundary. A mere statement in the release certificate is insufficient if users continue to uncontrolled access to the excluded process.
What significance does the training level have for the decision?
Verify whether the personnel assigned at startup have the capability to perform their specific duties and are familiar with relevant error scenarios. Consider substitutes and newly implemented operating procedures. A general participation certificate is only as credible as the scope of training it reflects. If the required user group has not yet been adequately trained, this must be factored into the planning and decision-making regarding the actual scope of usage.
Sources: [4]
What occurs when a change is made shortly after release?
Assess whether the change affects the conclusions of previous validation or approval decisions. Ensure that configuration, impacted requirements, and any additional verification activities are clearly documented and traceable. The proximity of the change to go-live does not automatically render it non-critical. Depending on the impact, additional validation evidence and a renewed decision may be required; however, not every change necessitates a complete rebuild of all documentation.
Sources: [2]
How do we avoid an costly release meeting with no outcome?
Distribute an initial draft document containing scope, supported basis, decision-relevant findings, and clear questions. Invite individuals with the necessary authorizations and expertise. Assign a responsible person and a specific accountability for any missing information. The output must clearly state what may be used or explain why a decision cannot yet be made. A further general project status update provides little added value.
Sources and review date
Source date: October 6, 2026. Consideration of human medicinal product GMP. Tables, checklists, and laboratory ERP example are standalone reference tools; they do not replace an evaluation of the specific system.
- European Commission: EudraLex Volume 4 – Current Human Medicinal Product GMP Framework
- European Commission: EU GMP Annex 11 – Computerised Systems
- European Commission: EU GMP Annex 15 – Qualification and Validation
- European Commission: EU GMP Chapter 2 – Personnel
- European Commission: EU GMP Chapter 4 – Documentation
- EMA/ICH: Q9(R1) – Quality Risk Management
Sources checked: .
Prepare the go-live decision specifically in advance
We help you translate evidence of compliance, open findings, and system readiness into a traceable decision-making basis. The actual scope of use determines the extent of testing.
- Computer system validation →
Bring your GxP systems into validated operation, from laboratory software, MES and eQMS to ERP. We define the scope, plan risk-based tests using GAMP 5 and hand the evidence to QA for release.

