Compliance
Practical Data Migration Validation: Annex 11 and a SharePoint Example
Plan data migration validation with mapping, reconciliation and cutover checks. Compare current and draft Annex 11 through a fictional SharePoint register example.
Data migration validation asks whether a planned transfer preserves the data and meaning needed for its intended use. As part of computerised system validation (CSV), define the source and target, assess the risks, approve the mapping, test the transfer process and reconcile the records before an authorised cutover. Matching row counts alone does not show that each record still means the same thing.
This guide relates that work to current and proposed Annex 11 provisions, then uses a realistic, fictional Excel-to-SharePoint quality-register migration to show the checks in practice. The data and decisions are illustrative: no customer migration, import or test was performed.
What do current and draft Annex 11 say about migration?
For Australian medicinal-product GMP, the Therapeutic Goods Administration (TGA) identifies PIC/S PE 009-17, dated 25 August 2023, as applying from 1 September 2025, except Annexes 4, 5 and 14. Check the product, activity, exemptions and any transition provisions with a qualified reviewer; this article does not decide whether GMP applies to your register.
| Source and status | Migration provision | Practical question |
|---|---|---|
| Australian-adopted PIC/S PE 009-17, Annex 11 section 4.8, PDF page 177 | Where data is transferred to another format or system, validation should include checks that its value and/or meaning are not altered during migration. | Can the team demonstrate that identifiers, critical values and relationships still mean what they did in the source? |
| 2025 joint PIC/S–European Commission consultation draft, section 10.3, PDF page 8 | For an ad hoc migration of critical data or a whole database between systems, the draft proposes a validated process that considers, among other things, constraints on both the sending and receiving sides. | Has the team defined and tested the transfer process and understood what each side can represent? |
The consultation draft is a proposal, not a current Australian requirement. The European Commission consultation closed on 7 October 2025; its EU GMP index lists the January 2011 revision of EU Annex 11. A newer PIC/S publication or the EU draft does not by itself change Australian adoption. These sources were checked on 17 November 2026; recheck status and applicability before relying on them.
The draft's glossary includes both raw data and metadata in its description of migration; its definition also covers format changes. Consider which source history and context need to remain available, rather than assuming that a list of values alone captures the whole record.
The steps below are practical ways to plan and evidence a migration, not an exhaustive interpretation of either text or proof of compliance. For the wider validation lifecycle, see practical risk-based computerised system validation.
A practical Excel-to-SharePoint migration scenario
Imagine a quality team moving a deviation register from an Excel workbook to a SharePoint list. The register points to controlled deviation records; it does not replace those records. A due date changed by an ambiguous format, or a reference linked to the wrong record, could affect follow-up even if both systems contain the same number of rows. The process owner, migration owner and independent reviewer agree which data and evidence are needed before cutover.
This is a fictional, illustrative worked example, not an observed customer case. It addresses a one-off transfer, not routine synchronisation, SharePoint configuration or a finding that the target is suitable for regulated use.
For the platform responsibility boundary, see Microsoft's guidance on GxP and its cloud services. Microsoft states that there is no GxP certification for cloud service providers and describes responsibilities shared by Microsoft and its customers. Supplier evidence does not establish that your configured SharePoint register or migration is validated for its intended use.
Where US FDA requirements are relevant, Microsoft's 21 CFR Part 11 offering guidance explains how supplier audit reports can support a customer's own risk analysis and qualification. It does not establish that the customer's application meets FDA requirements or replace the Australian applicability assessment above.
1. Establish the source baseline
Before any transfer, identify the source workbook and version, the records in scope, the fields that matter to the register's intended use, and the people who will approve the mapping and cutover. Preserve a controlled source snapshot and record how edits made after that snapshot will be handled. Do not silently clean or reinterpret the source while profiling it.
For a small fictional deviation register, the agreed source baseline might be:
| Record ID | Status | Due date | Controlled-record reference |
|---|---|---|---|
| DEV-2026-0001 | Open | 2026-06-29 | CR-0001 |
| DEV-2026-0002 | Open | 2026-07-03 | CR-0002 |
| DEV-2026-0003 | In progress | 2026-07-10 | CR-0003 |
| DEV-2026-0004 | Open | (blank) | CR-0004 |
| DEV-2026-0005 | Closed | 2026-06-20 | CR-0005 |
With 30 June 2026 as the example's fixed reporting date, the expected profile is five unique record IDs, four records not closed, one closed record, one overdue open record, one blank due date, and five record references. These are expected counts for a synthetic fixture, not results from an import or a live register.
The baseline should also state what the counts include: the selected worksheet or table, filters, hidden rows, duplicate records, formulas or derived values, and any excluded data. Keep the definition stable so the source and destination are compared on the same basis.
2. Approve the mapping before transfer
For every field that affects identification, status, reporting or a link to a controlled record, write down the source, target, conversion rule and check. Confirm the meaning with the process owner; a technically accepted value can still be semantically wrong.
| Source content | Proposed target and rule | Reconciliation evidence |
|---|---|---|
Record ID, such as DEV-2026-0001 | RecordID as text; preserve the complete identifier, including prefixes and leading zeroes | Compare each source ID with its destination value; list missing, duplicate or changed IDs |
| Due date and blank cells | Agreed date field; define date interpretation and keep blanks distinct from placeholder dates | Compare each value after the approved conversion; resolve ambiguous dates before acceptance |
Status, such as In progress | Agreed status value, using a documented mapping if labels differ | Compare source and destination status counts and each mapped value; do not map unknown labels to a default |
Controlled-record reference, such as CR-0001 | Agreed reference field or relationship | Check that each destination reference points to the same intended source record |
| Historical author or entry date, if in scope | Separate provenance field with a defined source and format | Compare the retained source value; do not assume target-system created/modified metadata represents the source history |
Record whether a proposed mapping is approved, pending or excluded and why. If a target field cannot represent a required value or relationship, treat that as a decision to resolve, not a reason to discard the source detail without review.
3. Test the process and reconcile beyond row counts
Before use, specify the transfer method, settings and expected results, then test with synthetic records, including invalid values and failures. Record who performed each check, actual outcomes, deviations and any retest. Check the sending workbook and receiving list constraints; do not assume that a successful import preserves original authorship, timestamps or history.
Use the same agreed scope and comparison rules on both sides. A reconciliation log can be copied and adapted for each migration:
| Check | Source baseline or rule | Destination result | Exception and disposition | Owner / approval |
|---|---|---|---|---|
| Record population | Five in-scope records; scope and exclusions documented | Not executed | — | Process owner |
| Stable identifiers | Five distinct IDs; exact text retained | Not executed | — | Migration reviewer |
| Critical values | Agreed fields and status mapping compared record by record | Not executed | — | Process owner |
| Record relationships | Five references resolve to the intended controlled records | Not executed | — | Record owner |
| Transformations and blanks | Approved conversion rules applied; blanks and rejected values accounted for | Not executed | — | Migration owner |
| Exceptions | Each mismatch is assigned, assessed and dispositioned; affected checks are repeated | Not executed | — | Named exception owner |
| Cutover and rollback | Acceptance authority, timing, rollback trigger and source retention agreed | Not executed | — | Authorised decision-maker |
The “Not executed” entries are intentional: this is a planning example, not evidence that a migration passed.
Try the checks with separate synthetic exceptions
Keep invalid fixtures separate from the five-row baseline so they do not change the expected production population.
| Synthetic fixture | Expected check result | Actual result |
|---|---|---|
Add a sixth row with duplicate ID DEV-2026-0001 | Isolate the duplicate; no silent overwrite or unexplained extra accepted record | Not executed |
Use the ambiguous source date 03/04/2026 | Confirm its intended meaning before conversion; do not guess a locale | Not executed |
| Supply a status with no approved mapping | Flag it for an authorised decision instead of substituting a default | Not executed |
| Change a controlled-record reference to a plausible but incorrect ID | Detect and resolve the relationship mismatch, not just the nonblank value | Not executed |
An exception is not resolved by editing the destination until it agrees with an assumption. Record the source value, observed destination value, decision-maker, rationale and controlled correction or disposition. Repeat the affected comparison after any approved correction.
4. Decide whether cutover is ready
- Control the source snapshot. Identify the exact workbook and data range, capture the baseline, and agree how later source changes will be handled.
- Profile and map. Identify duplicates, blanks, ambiguous formats and relationships. Obtain approval for transformations and exclusions before the transfer.
- Run a trial with synthetic records. Record the selected method, settings, rejected values and transfer identity. Do not use production records to discover how a method behaves.
- Reconcile the agreed fields. Compare identifiers, critical values, status totals, relationships, blanks and reporting outputs relevant to the register's use. Record exceptions and rerun affected checks after disposition.
- Obtain an explicit cutover decision. The authorised owner decides whether evidence is sufficient, which source is authoritative after cutover, how a final change set is handled, and what condition triggers rollback.
Unexplained missing or extra records, mismatched critical values, unresolved references, or unapproved transformations are reasons to pause the decision and refer the exception to its owner. Do not run two competing copies without an agreed source-of-truth and reconciliation plan. Retain the source and relevant history until the authorised owner decides how they should be handled under the applicable procedure.
The checks here are recommendations to adapt to the intended use; they are not a universal regulatory checklist, a fixed sampling rule, or proof of compliance. A qualified reviewer should assess applicability, evidence sufficiency and retention against the relevant requirements and procedures.
Bring the evidence together for the cutover decision
For this register example, organise the outputs into a review pack that lets the authorised owner trace each acceptance decision back to its evidence. Agree its contents before execution; a useful starting point is:
| Decision to support | Evidence to bring together |
|---|---|
| What was meant to move? | Identified source snapshot, in-scope records, exclusions and approved field mappings |
| What would count as acceptable? | Agreed acceptance criteria, comparison rules, test cases and named reviewers |
| What actually happened? | Transfer log, rejected records, completed reconciliation and retained test evidence |
| What still needs a decision? | Exception log, impact assessments, approved corrections and affected retest results |
| Can the team change over safely? | Final source-change reconciliation, cutover approval and tested rollback arrangements |
| How will the old information remain available? | Source-retention decision, responsible owner and evidence that required history can be retrieved |
If work continues in the source after the trial, identify and reconcile those later changes before the final decision. For rollback, agree how records created or changed in SharePoint after cutover would be recovered and reconciled, rather than assuming that reopening the workbook is sufficient. Approval to use the target does not itself authorise deleting the source or its history; record those retention and disposal decisions separately under the applicable procedure.
This is a suggested review structure, not an executed evidence pack or a regulator-issued template. For a broader discussion, see IntuitionLabs' cutover evidence-pack approach to GxP data migration validation. It is third-party implementation commentary, not the authority for Australian requirements.
Choose the next step
For the register's field dictionary and data checks, see SharePoint Register Validation: A Worked Example of Field Settings and Tests. For the wider question of record purpose and system use, read Validating a SharePoint Quality Register.
If you are assessing a suitable one-register move, the Excel register migration and validation sign-off pack is the relevant QikSolve service pathway. Start by confirming fit, mapping, evidence scope and authorised review; migration is not the automatic answer for every register.