Table of contents
At a glance
How do you validate Odoo for GMP use?
Demonstrate that the installed Odoo applications and data flows are fit for the intended GMP process. Identify the edition, deployment model, extensions and permissions. Test how item, lot, quantity and quality decision are linked. A matching total inventory is insufficient. Requirements, failure consequences and existing evidence determine which configurations, handovers and failures to test.
- Community and Enterprise are editions; Online, Odoo.sh, and On-Premise describe different deployment models.
- Retrospective batch assignment requires reliable origin documentation in addition to quantity reconciliation.
- A completed quality check only verifies the defined testing, not a complete batch release.
- Multiple JSON-2 calls do not automatically form a single transaction.
First, separate Edition, Hosting, and Extensions.
The question "Which Odoo version are you using?" is insufficient to provide adequate evidence. It is necessary to capture the core version, edition, hosting configuration, applications, and additional modules together. Two installations with the same primary version may process bookings or review decisions differently. Furthermore, implementation partners may have introduced changes that are not documented in the general end-user documentation.
Odoo distinguishes between Community and Enterprise editions. The hosting overview assigns Online to the Enterprise edition; On-Premise can be operated with both editions. The Odoo.sh setup guide requires a compatible Enterprise subscription. Custom modules are not supported by Online, according to the documentation. However, this does not imply a blanket prohibition against other configurations. Similarly, the feature set of the Quality application is not definitively established for every Community configuration.
The worksheet distinguishes these levels. Check licensing and the actual installation separately. For every relevant extension, identify its maintainer and the tested version in production.
| Level | Status to record | Critical evidence question |
|---|---|---|
| Editions and Applications | Community/Enterprise; installed apps and licenses | Which of the described functions are actually available? |
| Online | Core/Intermediate Version and Used Configuration | Which changes are provided by the supplier and which by the internal team? |
| Odoo.sh | Subscription, Repository, Module Stand and Database Status | Does the tested combination correspond to the production level? |
| On-Premise | Edition, Infrastructure and Operator Responsibility | Who is responsible for operation, backup, and updates? |
| Expansions and Data Paths | Own/external modules and specific interfaces | Which actions alter the regulated process? |
Which Odoo processes are covered under GMP validation?
This article examines Odoo within the manufacturing of human medicinal products. Annex 11 is the relevant EU GMP guideline for computerised systems in GMP-regulated activities. The current EudraLex index still references the revision from January 2011. Consultation drafts are not considered valid requirements herein.
Describe the specific impact: Does Odoo control the availability of a raw material, transmit batch identity, or retain decision-relevant records? A purely commercial process and a quality-related storage process do not automatically require the same scope of validation. However, the distinction must not overlook indirect effects, such as a shared master data field.
The validation of the application and the appropriate qualification of the underlying IT infrastructure constitute distinct evidence requirements. The product name does not imply a specific software category or a general GxP license. Requirements from other areas—such as medical devices, pure wholesale trade, or FDA-regulated records—must be determined separately for the specific use case.
Specify the approved process requirements
A useful requirement defines a decision. For example, material whose lot identity is unresolved must not enter the intended manufacturing process. Then decide which combination of software controls, warehouse procedures and authorised decisions meets the requirement. A requirement that merely calls for batch management leaves important exceptions open.
Follow a typical workflow from warehouse receipt to usage. At each transition point, verify identity, quantity, unit, status, and responsible party. A misidentified batch with correct quantity may have more significant consequences than the absence of an immediately visible data record. The testing depth should reflect these distinctions.
Existing supplier documents can answer a part of the questions. Arrange each document used to the appropriate level of functionality and version. Often the own configuration and the connected systems remain open. Unknown behavior is documented as uncertainty; missing information does not justify a low risk classification.
Post-booking batch assignment: Clarify origin prior to booking
Odoo defines a special transition process for existing inventory: if tracking by lot or serial numbers is activated only later, inconsistent data sets may arise. Document 20.0 outlines two inventory correction procedures, which replace unassigned quantities with assigned quantities. All relevant warehouse locations must be considered during this process.
This is a technical functional description, not approval for unverified production corrections. Prior to any change, a known starting inventory and independent evidence of the actual batch identity must be available. A delivery note alone is insufficient, for example, if the quantities listed do not align with the existing material. Unclear origin cannot be resolved by subsequently assigning a plausible batch number.
For the planned transition, warehousing and QA should define how open movements, reservations, and unassignable quantities shall be handled. The approach should be tested in an appropriate test environment. Prior inventory, assignment basis, reason for change, and outcome shall be retained together. Only after evaluated deviations and a duly authorized decision may the production process be finalized.
Test example: 100 units are still not a correct batch quantity
The following example does not describe an executed Odoo test. In an isolated test environment, 100 units of an item are initially present without lot assignment. Unaltered, independent stock locations show 60 units of Lot A at location P1 and 40 units of Lot B at location P2. Changes in quantity due to other activities are excluded during the test.
The designated target allows only Batch A to be used; for Batch B, the quality decision remains open. Each table case starts independently from the described initial state. In the origin error scenario, a different staffing level for five units at position P2 does not permit a unique assignment. The expected outcome is then 60 clearly assigned to A, 35 clearly assigned to B, and 5 unresolved units, totaling 100 units as before.
The quantity calculation is independent: 60 + 40 = 100. Even an incorrect assignment of 40 to A and 60 to B still results in 100. Only the individual comparison reveals a deviation of −20 for A and +20 for B. Document actual dataset entries and system responses during testing alongside the pre-defined expected values; a retrospectively adjusted expected value would not constitute an independent comparison.
| Test case | Controlled entry | Independently established expectation | Evidence to be ensured |
|---|---|---|---|
| Correct assignment | Take 60 from A/P1 and 40 from B/P2. | A = 60; B = 40; Total = 100 | Before-and-after balance with original assignment basis |
| Swapped quantities | Assign Target 40 to A/P1 and 60 to B/P2; retain original records. | Total 100, but deviation −20/+20; not accept assignment | Comparison by lot and bin, with assessment of the finding |
| Unknown origin | For 5 of the 40 pieces on the P2 there is no loadable lot certificate | 60 A + 35 B + 5 unclassified = 100; no established assignment | Need for clarification and decision on affected quantity |
| Use of B | Request from B for an actually used downstream processing route | No use prior to authorized decision in accordance with the established control concept | Response and efficacy of the intended control |
| Dual acceptance | Complete the initial 60/40 takeover; offer the same migration dataset again | Final balance remains at 100; repetition does not generate a second regular balance amount | Identity of receipt, processing status and final stock |
Quality check and usage decision are binding.
Odoo Quality describes pass/fail inspections that may arise individually or through Quality Control Points. They can relate to inventory, production, or work orders. A quality check first verifies the criterion defined within it. Whether a reservation, withdrawal, or release is permissible thereafter is a separate process question.
Therefore, assess the relationship in your configuration: Which check corresponds to which quantity? Who is authorized to record or modify the result? What occurs after a failure? The procedure outlines a potential Quality Alert step, but it does not establish a universal lock on all other data pathways. Additionally, in the shop floor workflow, it references a checkbox through which a check can be directly marked as passed. This alternative user pathway should be considered in the evaluation when it is feasible to use.
An operational material decision is to be distinguished from the certification of a finished product batch. Annex 16 describes in the EU context the QP certification and the subsequent transition to salesable stock. The word "contained" in Odoo does not replace this decision and its evidence.
JSON-2: Successful calls may result in an incomplete workflow.
Document 20.0 describes JSON-2 as the interface introduced starting from version 19.0. Available models and methods depend on the specific database. The documented pricing note references external API usage for Custom and excludes Standard and One App Free. Verify this condition for your agreement; it does not constitute a complete statement regarding any self-operated community installation.
Technically critical: Each JSON-2 call executes within its own SQL transaction. Consecutive calls cannot be automatically aggregated into a single, unified transaction. If an integration first reads a stock level and later performs a modification, interim operations may have altered the original state.
Therefore, define the logically relevant operation and the expected error pathway. Interrupt the test between the actually executed calls. Subsequently, verify the stored state, the notification to the sender, and a retry. A successful HTTP response for a partial call does not confirm that the entire inventory process has been correctly completed. The selection of an appropriate atomic method or alternative effective controls must align with the actual integration requirements.
Examine permissions and comprehensible changes
The JSON-2 documentation references user rights, record rules, and field access. For permanent integrations, Odoo recommends the use of dedicated technical users. However, a complete history of every quality-related change cannot be derived from this setup. For your data, determine which information is actually logged and identify who has access to read or modify it.
A suitable review request begins with a specific change: a user corrects the batch assignment, another modifies the inspection result. Can the reviewer trace both the original and updated content, the identity of the individual who made the change, the timestamp, and the required justification? Are the affected quantities clearly identifiable? Additionally, consider an integration user whose technical name does not fully reflect the prior business decision.
Additionally, specify which records are required, for what time period, and how they remain retrievable and linked. A list of holdings without a clear assignment basis may suffice for a current overview, but it is insufficient to adequately explain the historical decision-making process. The suitability of export, historical view, and supplementary documentation is assessed for this purpose.
Updates, module changes, and restoration linked to the same process
An update may impact the described evidence at various points: new core version, modified add-on module, altered dataset rules, or adjusted interface. Compare the affected process with existing evidence. This leads to targeted additional assessments; neither automatic full re-execution nor unverified adoption of prior results is appropriate.
The Odoo.sh guide distinguishes major versions of online intermediate versions and describes an import that overrides existing branch data. Clear compatibility and relapse before execution. A successful import does not yet prove the correct allocation of all data and files.
For restoration, select a dataset with credible history: for example, a corrected batch assignment together with the associated test sample. After restoration, demonstrate that identity, quantity, and required context remain available. The hosting organization's organizational responsibility does not substitute for verification that your process is adequately supported by the system.
What your team has prepared and what determines the effort
The scope is less determined by the number of menu items than by the distinct data pathways and outstanding proof requirements. Warehouse and QA provide the authorized workflow and verifiable assignment records. IT and implementation partners define components, rights, and technical handovers. Authorized responsible parties assess open findings and determine usage.
A standard installation with a well-documented process requires a different approach than a mature environment with multiple extensions. Additional effort is introduced primarily by unclear origin, missing version histories, alternative booking pathways, and a non-representative test environment. Start with this checklist before ordering test cases or finalizing a transition.
- Edition, hosting, versions, and affected modules are clearly documented.
- The GMP-relevant process and authorized exceptions are described.
- Identity, quantity, unit, and storage location have independent target records.
- Unexplained quantities receive an authorised decision-making process.
- Quality status and actual usage are verified through relevant channels.
- API Termination, Retesting, and Concurrent Changes have clear expectations.
- History, roles, and storage address the required review questions.
- Change evaluation, restoration, and open findings are considered.
Project experience and collaboration
Test how the system handles failed quality checks
For Odoo, the apps, customisations and integrations actually used determine the scope. We follow a relevant transaction with users, including what happens downstream after a failed Quality Check. A defined set of apps may need a focused package. Before introducing a new process chain, we establish requirements, permitted operating paths and responsibilities with IT and process owners. We then agree the tests and evidence needed.
Frequently asked questions about Odoo
Can Odoo Community be used in a GMP environment?
The edition alone does not address the suitability question. Key considerations are the intended process, actual functionalities, and effective controls. Enterprise- or quality-related descriptions must not be transferred to Community without proper evaluation. First, assemble module details, operators, and available evidence. Open requirements may necessitate an alternative configuration or solution; they are neither automatically fulfilled nor excluded by the open-source status.
Do we need to validate the entire ERP in Odoo?
Determine the scope based on GMP impact and dependencies. A commercial workflow may fall outside the relevant application scope, while a shared material field must be considered. Document this boundary and its rationale. A simple list of excluded applications is insufficient if rights, master data, or interfaces still affect the regulated process.
How should we handle stock whose lot is unknown?
Separate the existing quantity by source of record. Use appropriate, unchanged storage and delivery labels for assignment. If these are insufficient for a partial inventory, the source remains open and requires a duly authorized evaluation. Enter no plausible batch number; do so only to ensure complete data structure. A technical stock correction does not generate historical evidence.
Is the total inventory sufficient as a validation criterion following a change?
No. Additionally verify the assignment per item, batch, unit, and relevant storage location. In the example of this item, both 60/40 and 40/60 combinations total 100 units. The sum comparison fails to detect swapping errors. Define the required quantity from independent documentation prior to testing and maintain the original documentation even if a correction to the transfer is required.
Does a failed quality check automatically block the shipment?
A failed check or a Quality Alert alone does not constitute an effective shipment hold. Verify whether the established follow-up process actually prevents shipment or whether another decision is required. Consider alternative user pathways and integrations. The Pass-Fail procedure describes the check and a potential Quality Alert; your required control must be demonstrated in actual operational flow.
Why does a JSON-2 integration require a termination test?
Since multiple calls constitute separate transactions, an earlier call may successfully be recorded while a subsequent one fails. At an appropriate interruption point, verify the remaining state and the ability to resume. The expected handling must align with the technical operation. A general success message from the connector neither replaces the reconciliation of the booking nor provides evidence that retrying does not generate duplicate quantities.
Is an integration user simultaneously sufficient for an audit trail?
No. A dedicated user assists in assigning technical access rights but does not automatically populate prior field values, change justifications, or the preceding decision. Review the actually stored history of the relevant objects. The reviewer must be able to reconstruct the required context. Missing information should be evaluated in terms of its significance and addressed through an appropriate overall concept.
Can we immediately transition online to Odoo.sh?
The documentation specifies version boundaries: Odoo.sh supports major versions, not arbitrary online intermediate versions. Additionally, modules, databases, and files must be compatible with the target environment. Therefore, clarify the actual current state and the intended migration path prior to proceeding. Perform a controlled transition, including rollback and data mapping. A technically successful import does not constitute professional release or approval.
Does every module update require a complete revalidation?
No, the scope of inspection follows the impact. Identify which requirements, data pathways, and controls are affected by the change. Existing documentation may remain applicable. For new or previously insufficiently covered risks, additional validation evidence is required. Document the rationale for why adjacent processes remain unchanged or are specifically re-evaluated.
What should QA and IT provide prior to a cost estimation?
The following are required: system and module list, process boundaries, available supplier documentation, roles and interfaces, as well as existing inventory and disposal history. Known deviations and available test setup should be supplemented. This enables differentiation between cases where existing evidence is primarily available for assignment, or where unclear processes and data holdings require prior cleansing. A credible assessment follows from these actual tasks.
Sources and review date
As of 6 October 2026. Supplier reference: Odoo Documentation 20.0 and public hosting overview. Separate verification of app, license, and customer configurations is required. Regulatory classification applies to human medicinal products under GMP; other areas of applicability are not hereby extended. Tables and target values are internal work aids and do not constitute completed product testing.
- Odoo Hosting Types
- Create a project — Odoo 20.0 documentation
- Reassign lot/serial numbers — Odoo 20.0 documentation
- Pass - Fail quality check — Odoo 20.0 documentation
- External JSON-2 API — Odoo 20.0 documentation
- European Commission: EudraLex Volume 4
- EU GMP Annex 11: Computerised Systems
- EU GMP Chapter 4: Documentation
- EMA: ICH Q9(R1), Step 5, Corr.2
- EU GMP Annex 16: Certification by a Qualified Person and Batch Release
Sources checked: .
Verify key Odoo data flows intentionally and systematically
We support you in defining the regulated scope, evaluating existing documentation, and designing appropriate tests for batch release, quality decisions, and interface validation.
- 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.

