ALCOA+ is easy to recite and harder to trace to the system setting that delivers each attribute. This post turns the nine attributes into concrete controls for computerized systems, so a validation lead can test them, and ties them to the data criticality question in the July 2025 draft of Annex 11. If you work on data integrity in pharma, treat it as a translation table between the acronym and your configuration.
One status note first. The January 2011 Annex 11 is in force. The July 2025 consultation draft is not adopted, and nothing here predicts when it might be. Where the draft is cited, the text says “the draft expects”. Always confirm the current status on the European Commission consultation page.
What ALCOA+ means for computerized systems
The draft Annex 11 glossary defines ALCOA+ as five core attributes (attributable, legible, contemporaneous, original, accurate) with extra emphasis on four more: complete, consistent, enduring and available. That gives nine in total, and this post uses exactly that list.
Some published lists add further attributes. I have not listed them, because which attribute set applies depends on the document your regulator, client or quality system cites. If yours is longer, add rows to the table below using the same method: one question, one control, one test.
The draft also says, in its principle on data integrity (2.4), that ALCOA+ covers many topics, including handling of data, identity and access management, audit trails, electronic signatures and security. In other words, the acronym is not a separate program. It is a lens over controls you already have to build.
A practical point: neither 21 CFR Part 11 nor the 2011 Annex 11 uses the acronym. The duties sit in the clauses on accuracy, audit trails, copies, retention and access. ALCOA+ is a convenient way to check that those clauses add up to trustworthy records.
ALCOA+ attribute by attribute: the system control
For each attribute, ask what the system must do on its own, because a procedure that relies on people remembering is the weakest control you have. The table gives the question, a typical control and the place the duty comes from.
| Attribute | Question to ask | Typical system control | Anchor |
|---|---|---|---|
| Attributable | Who did it, and can only that person have done it? | Unique personal accounts, no shared logins for anything that can change data, role recorded with each action | Part 11 11.10(d), (g); Annex 11 (2011) 12.4; draft 11.1, 12.2 |
| Legible | Can the record be read and understood for its whole retention period? | Clear printed copies, readable archives, viewer software kept available | Part 11 11.10(b); Annex 11 (2011) 8.1, 17 |
| Contemporaneous | Was it recorded when the event happened? | System-applied time stamps from a controlled clock, audit trail written at the event, clock changes themselves logged | Part 11 11.10(e); draft 12.2, 12.3 |
| Original | Is this the first capture, or a verified true copy of it? | Raw data stored in the native record, dynamic data kept dynamic, controlled copy process | Part 11 11.10(b); draft 12.9 |
| Accurate | Is it correct and unaltered in value and meaning? | Range checks on entry, validated interfaces, second check on critical manual entries, migration verification | Annex 11 (2011) 4.8, 5, 6; draft 10.1 to 10.3 |
| Complete | Is everything there, including changes and deletions? | Audit trail covering create, modify and delete; printouts that show changes since first entry | Part 11 11.10(e); Annex 11 (2011) 8.2, 9; draft 12.1 |
| Consistent | Do events line up in sequence, time zone and format? | Enforced step sequencing, one time source, time zone recorded where sites differ | Part 11 11.10(f); draft 12.2 |
| Enduring | Will it survive until the end of retention? | Tested backup and restore, archive checks, planned media and format migration | Annex 11 (2011) 7, 17; draft 16, 17 |
| Available | Can authorized people and inspectors get it when needed? | Searchable and exportable data with its audit trail, access that lasts the retention period | Part 11 11.10(b), (c); Annex 11 (2011) 7.1; draft 12.9 |
Attributable and contemporaneous: identity and time
These two attributes fail together in practice. A shared analyst login removes attribution, and a workstation clock that anyone can change removes contemporaneity. The draft expects unique, personal accounts and calls shared accounts a data integrity violation unless they are strictly read-only (11.1). It also expects any change to system time or audit trail settings to create its own audit trail entry, and to be possible only for an administrator not involved in GMP activities (12.3).
Test it, do not assume it. In qualification, log in as an ordinary GMP user and try to change the clock, switch off the audit trail and edit a record without a reason. Each attempt should be blocked or visibly logged.
Legible, original and available: what you can show an inspector
Legibility is about the retention period, not the first week. If the only way to read a ten-year-old chromatography file is a software version you retired, the record is not legible in any useful sense. Plan the viewer, the format and the hardware with the same care as the data.
Original is the attribute people argue about most. The workable rule is simple: identify, for each record type, what the first capture is (the instrument file, the electronic batch entry, the sensor stream) and make sure the controlled copy process leaves that capture intact. The draft says a complete electronic copy must include the audit trail data and must be searchable and sortable, and that flat, locked files do not qualify (12.9). A PDF printout can be a legible copy. It is not, on its own, the original.
Accurate, complete and consistent: the checks the system must run
Accuracy is where the 2011 Annex 11 is most concrete. It expects built-in checks when systems exchange data (section 5), an extra check on critical manual entries by a second person or validated electronic means (section 6), and verification that migrated data keep their value and meaning (4.8). The draft keeps the same logic: plausibility checks on manual input, validated interfaces in place of manual transcription where possible, and a validated process for ad hoc migration (10.1 to 10.3).
Completeness means deletions and changes are visible, not that the record is long. For batch release records, the 2011 text expects printouts that show whether data changed since the original entry (8.2). Consistency is the quiet one: when two sites record the same batch step, do the time stamps and time zones allow the events to be sequenced? Decide the time source and zone convention once, in the requirements, not after the first deviation.
Enduring: backup, restore and archive
Enduring is proven by restoring, not by having a backup job that reports success. The 2011 text expects integrity and restorability to be checked during validation and monitored periodically (7.2) and archived data to be checked for accessibility, readability and integrity (17). The draft goes further on archiving, expecting checksum verification before data are deleted after a move to an archive (17.2). Keep the restore test results in the validation file.
Data criticality and vulnerability in draft Annex 11 4.5
You do not need maximum controls on every field. The draft’s data integrity risk clause (4.5) tells you how to decide: use quality risk management to assess how critical the data are to product quality, patient safety and data integrity, how vulnerable they are to deliberate or accidental alteration, deletion or loss, and how likely such an action is to be detected. Three questions, then, and the answers set how hard each ALCOA+ control must work.
A simple qualitative scale works. The table shows how the three questions can read across typical records. The ratings are examples to show the method, not benchmarks.
| Data | Criticality | Vulnerability | Detection likelihood | What it drives |
|---|---|---|---|---|
| Release test results in a laboratory system | High | High: results can be re-processed or re-run | Medium, only if audit trails are reviewed | Audit trail review per batch, second-person review, locked processing methods |
| Critical process parameters in an electronic batch record | High | Medium: step overrides and late entries possible | Medium | Enforced sequencing, reason-for-change prompts, review before release |
| Calibration due dates in a maintenance system | Medium | Low: few users can edit | High: dates are cross-checked | Standard access controls, periodic access review |
Two consequences follow. First, high criticality with high vulnerability and weak detection is where you spend validation effort and review time. Second, the assessment has to be written down per system, because the same field can sit in different cells in different systems. A data criticality assessment belongs in the system’s risk file and gets revisited at periodic review.
Where to look for ALCOA+ gaps
These patterns are worth checking first when you assess an existing system. They come from how systems are typically configured, not from any statistic.
- Spreadsheets used as records. Cells can be overwritten with no history. Either lock and validate the workbook with a controlled save process, or move the calculation into a system with an audit trail.
- Hybrid records. A signed printout and an electronic file both claim to be the record. Decide which one is relied on, link them, and control the other. The draft suggests a checksum of the electronic record printed on the signature page (13.9).
- Audit trails switched on but never read. The control exists, but the detection likelihood in your 4.5 assessment is overstated. See how to review audit trails in a risk-based way.
Putting ALCOA+ into your validation file
Do not write a separate attribute document that nobody tests. Thread it through the lifecycle:
- Requirements. Write data integrity requirements next to functional ones. The draft lists data integrity as its own requirement type (6.2). Each nine-attribute question above can become one or more testable requirements.
- Risk assessment. Run the 4.5 criticality, vulnerability and detection assessment on the system’s data, and record which controls it justifies.
- Testing. Include audit trails, access privileges, calculations and restore from backup in your tests. The draft’s list of test focus areas names all of these (9.6).
- Procedures. Cover entry verification, transfer, copies, hybrid records, access and audit trail review in procedures that match the configured system.
- Periodic review. Re-check incidents, access changes and audit trail findings against the assessment. The 2011 Annex 11 expects periodic evaluation of validation status, deviations, incidents and security (section 11).
For the surrounding framework, the 21 CFR Part 11 requirements guide explains which sections matter when the records are FDA-regulated, and the EU GMP Annex 11 comparison shows the 2011 text and the draft side by side.
No system is compliant by itself. Compliance depends on how you configure, validate and use it, and a table like the one above is only as good as the tests behind it.
If you want the documents behind these controls, the 21 CFR Part 11 Toolkit includes a Data Integrity Policy (ALCOA+), a Manual Data Entry Verification Procedure, an Interface and Data Transfer Verification Protocol and a Hybrid Records Control Procedure, among 105 templates (68 Word and 37 Excel) for $99. The download is named “Computer System Validation Toolkit”.
ALCOA+ FAQ
Is ALCOA+ a legal requirement?
The acronym itself is not a regulation. Part 11 and Annex 11 set the duties on accuracy, audit trails, copies, retention and access, and the acronym is a way of organizing them. The draft Annex 11 names it in its data integrity principle and glossary, but that text is a consultation draft and not adopted.
What is the difference between ALCOA and ALCOA+?
ALCOA is the five core attributes: attributable, legible, contemporaneous, original, accurate. The draft Annex 11 glossary describes it as putting extra emphasis on complete, consistent, enduring and available, which it treats as implicit in the basic principles.
Do I need to apply every attribute to every record?
Apply all nine as a lens, then scale the controls with the data criticality assessment. High-criticality, high-vulnerability data get the full set of controls and review. Low-risk data can rely on standard access controls. Write the reasoning down.


