Regulations and compliance · Practical guide

GxP configuration baselines: preserving the tested state

The application bears the same name as the version tested, but a mapping or limit has changed. A configuration baseline identifies the settings that matter and enables comparison of the intended, tested and deployed configurations.

Table of contents

At a glance

What belongs in a GxP configuration baseline?

A GxP configuration baseline records the approved state in enough detail to identify significant changes and deviations. Include versions, settings, roles and dependencies relevant to intended use. A folder of unfiltered screenshots alone is insufficient. The baseline should allow comparison of the tested state with the system actually in operation.

Scope and Applicability

Annex 11 refers to change and configuration management. The specific form of a baseline is a methodological definition. Not every technical setting requires the same level of documentation and evaluation effort.

Sources: [1]

Describe the released status as a related system

A configuration baseline denotes the reference state relevant to the intended use. This may include software version, critical parameters, roles, interfaces, and subject-matter master data. Selection is based on actual impact. A screenshot of the version dialog rarely fully describes a complex application.

Distinguish current configuration, approved change, and reviewed state. These may differ. An approved change is not yet confirmation of implementation. A current export may already contain an unreviewed change. The baseline should therefore clearly illustrate the relationship between the intended state, evidence of implementation, and actual operational state.

Annex 11 addresses controlled changes to computerised systems. The term Baseline serves as a methodological aid to ensure traceability. A universal form with a prescribed number of fields is not accepted. Where applicable, existing configuration and change management procedures should be utilized if they adequately support the required documentation.

Sources: [1]

Use a lean, risk-based baseline template

Capture components whose modification could affect functionality or data integrity. Decorative settings do not need to automatically inherit the same hierarchy level as a calculation parameter. However, describe the selection criteria to ensure that critical dependencies are not lost simply due to difficulties in capture.

State the source and date of each status verification. A manually maintained value may be outdated; an export may contain more or less than the required extent. Verify that the verification reliably reflects the actual state. A filename such as "final" is not sufficient as a version reference.

Internal work aid: Use a lean, risk-based baseline template
Baseline scopeOwn contentReference to the evidence
Application Product, Version and relevant enhancements Proven Functional Status
Configuration Critical Parameters and Rule Versions Requirements and Test Cases
Access rights Roles and relevant exceptions Access matrix and action review
Interfaces Mapping, Recipient and Technical Identity End-to-end evidence
Technical Data Relevant methods or specifications Selection and Decision Rule
Infrastructure Platform- and operation-dependent requirements Appropriate qualification and operational records

Sources: [1]

Example: unchanged software, modified interface rules

The application version remains unchanged, but a mapping update alters the conversion of a measurement value. An inventory that records only product versions will not detect such a difference. Nevertheless, the technical processing has changed. The baseline must therefore include the relevant mapping version.

Prepare a known data case. Compare approved rule, actual configuration and result in the receiver. Document the source of the configuration proof. If a difference is detected, check time and data volume concerned. A subsequent adaptation of the Baseline sheet does not evaluate an unchecked change retroactively.

After correction or appropriate assessment, the new state is clearly validated. Recreate relevant reference cases. Retain the prior reference to ensure historical results remain interpretable. The baseline thus represents a controlled transition between states, not a continuously overwritten list devoid of history.

Own work aid – Example: unchanged software, modified interface rules
Test caseActionExpected finding
Comparison of expected and actual results Verify mapping against approved reference Correct version or visible finding
Reference value Transfer known value Expected professional transformation
Unauthorized modification Alter critical parameters in a test environment Appropriate detection works
Recovery Make the specified version available again Critical function corresponds to the reference
Historical question Investigate previous result The rule in force at the time can be identified

Sources: [1]

Link the approved version to its subsequent changes

Link the baseline to release and the associated validation evidence. A list lacking a date, system reference, or responsible party leaves open whether it pertains to a plan, a test run, or production. Use unique identifiers and controlled storage.

When changes occur, it must be evident what has changed compared to the previous state. A complete new export can assist in identifying such differences, but it does not replace the evaluation of relevant discrepancies. Automated comparisons should deliberately address dynamic and professionally insignificant fields, ensuring that significant deviations do not get lost in noise.

Sources: [1]

Assess differences between expected and actual results

A difference may be documented or technical in nature. It is possible that an authorized change was implemented, but the associated evidence was not updated. It is also possible that a change was made without review. Such cases require different handling. Investigate the root cause and potential data impact before formally correcting the list.

State which usage is appropriate until clarification is obtained. A falsely documented cosmetic parameter has different implications than an unexplained result rule. The decision depends on the specific impact. Blanket blocking of all differences and blanket acceptance of all differences are equally imprecise.

Link the assessment to Change, Incident, or Deviation, as appropriate. The status record should reference the primary decision. Avoid contradictory copies of the same release across multiple files. A clear reference is often preferable to redundant text.

Sources: [1]

Use baseline for changes and restoration

Before a change, the baseline helps identify affected components and dependencies. After implementation, it confirms the current actual state. For restoration, it can indicate which versions and settings are required together. A backup without corresponding configuration information can only partially fulfill this function.

Check differences between test and production. Other endpoints, roles, or data volumes may alter the validity of a test. The baseline should identify relevant differences rather than asserting complete identity. Appropriate additional controls should confirm transferability to the intended operation.

For cloud services, not all technical details are always available. Document available provider information and custom configuration. Lack of transparency is assessed risk-related. An unknown component should not appear to be fully documented by a invented version number.

Sources: [1]

Practical organization of care and handover

State the responsibility and trigger for updates. A baseline that is only opened during annual audits is unlikely to support an ongoing change process. Integrate maintenance into actual takeover and operational procedures. The required form depends on system size and existing tools.

Conduct a review: What was the status at the time of a specified result, and which change led to the current state? A duly authorized personnel should be able to locate the answer in the prescribed documentation. If only the original administrator can provide this information, there may be a lack of appropriate handover.

Sources: [1]

Draft: Compare Intended Use, Test Status, and Actual Operational State

Compare with the table the relevant statuses for a critical function. Evaluate differences before simply updating the baseline: a documentation gap and an unverified technical change are distinct findings. Clear references can replace redundant copies of individual values, provided another authorized employee can actually reconstruct and understand the combination and its significance.

Sample filling example – internal work aid
Version informationExample entryWhat the test establishes
Expected Approved conversion rule. Expected technical knowledge is known.
Test Version designated with reference cases. Appropriate evidence available.
Actual Actual mapping and parameters in operation. No unresolved difference.
Dependency The report and interface use the same rule. Joint effect is considered.
Change Change explains the transition. Implementation and decision-making linked.
History Previous status remains retrievable. Old results are explainable.

Sources: [1]

Next steps

  • Select and justify critical configuration elements.
  • Ensure the verified status with system reference and date is maintained.
  • Evaluate changes against this baseline.
  • Regularly and as-needed reconcile active operating status.
Further reading:

Project experience and collaboration

Compare the tested configuration with the system in operation

For a critical function, the system owner, process owners and IT compare the intended, tested and production configurations. These can include the application, mapping and relevant parameters. A documentation gap warrants a different response from an untested change. We investigate the cause and potential effect on data before revising the baseline. Records and references should allow another authorised colleague to find the configuration in operation and the evidence supporting its approval.

Frequently asked questions about Configuration baselines

Does every setting require a screenshot?

No. Suitable export files or traceable configuration documentation can be more comparable. The evidence must be reliable and understandable.

Sources: [1]

Is the baseline equivalent to the system specification?

No. The system description explains the structure and interrelationships; the baseline maintains the relevant approved version. Both should be consistent.

Sources: [1]

Can an unverified supplier export be accepted?

Its scope and intent must be evaluated. It may omit critical settings or relate to other environments.

Sources: [1]

Must every system setting be documented in the baseline?

Not automatically generated with the same depth. Selection is based on impact on intended use, data, and controls. Justify critical components and dependencies. An unstructured full list may assist, but does not replace professional judgment regarding the settings relevant to validation and subsequent changes.

Is a baseline identical to a backup?

No. A backup supports recovery. A baseline describes the relevant reference state and its relationship to documentation. Both can interact. A backup alone does not explain why a configuration is appropriate; a baseline list does not automatically enable data and application restoration.

How do we handle a deviation between baseline and production?

Clarify whether the documentation or the actual configuration is incorrect and assess the potential data impact. Evaluate the time period and scope of affected elements. Simply updating the list does not automatically render an unverified change acceptable. The decision must reflect the actual current state.

What can a baseline provide during a restoration?

It describes the required combination of components and settings together with their validation rationale. This supports resumption of operation. However, it neither replaces backup data nor practical functionality testing. Upon restoration, the critical usage relevant to the intended application must be appropriately verified.

Sources and review date

Annex 11 provides the framework for documented system statuses and controlled changes. Selection fields, comparison matrix, and examples are proposed features for a traceable baseline; not every technical setting carries the same professional justification.

  1. European Commission: EU GMP Annex 11, Computerised Systems (2011)§ 10: Controlled changes; baseline documentation represents the own methodological implementation. Current version according to EudraLex as of 14.10.2026; no universal project or deadline specification.

Sources checked: .

Daniel Herrmann, Daniel Herrmann Consulting

Daniel Herrmann Consulting

Identify the version actually in operation

We support you in assessing the current situation, clarifying responsibilities, and implementing appropriate evidence within operations. The type of support that is most effective depends on your system, your team, and the specific objective.

  • 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.

Send a project enquiry Customer testimonials →