PDF/A Validator
Open the app: Validate a PDF/A file → (products.documentize.app)
In one sentence
PDF/A Validator checks whether a PDF really meets an archiving standard (PDF/A, or the accessibility standard PDF/UA-1) and gives you a list of every rule it breaks — with severity, the clause of the standard, the page and the internal object involved.
What makes it unique
- It only checks; it does not change or fix your file. The result is a report, not a new PDF.
- It tells you exactly what is wrong, not just "failed". Each problem carries a severity (error or warning), the specification clause (for example
6.3.4), the page and the object ID, so whoever makes the file can fix it at the source. - It can judge the file by its own promise. On Detect automatically it reads the standard the file claims for itself and tests against that. If the file claims nothing, it falls back to PDF/A-1B, the least demanding widely used level, which answers "is this archivable at all?".
- It lists twelve targets: PDF/A-1A, 1B, 2A, 2B, 2U, 3A, 3B, 3U, 4, 4E, 4F and PDF/UA-1.
When to use it / when not to
- Use it when an archive, court, regulator or customer demands PDF/A, and you must be sure before submitting.
- Use it after producing a file you believe is PDF/A, as a test.
- Not to convert a file to PDF/A. This app never changes the document. Conversion belongs in the producing tool or a converter (see Converter).
- Not to check that a PDF "opens fine". A file can open everywhere and still fail PDF/A: the standard is about still rendering identically when today's fonts and software are gone.
- Not a security check. It does not tell you if a file is signed or protected (use Verify Signature).
Before you start
- Only PDF is accepted; the page is set up for up to 10 files per run (verify on the live app).
- Know the target level if someone external dictates it, such as "PDF/A-2B". The site's FAQ explains the letters: B guarantees visual reproduction, U also requires text to be extractable as Unicode, A adds the full tagging needed for accessibility, so A is hardest to meet.
- An encrypted file fails PDF/A (encryption is forbidden). Remove the password with Unlock first if you are checking a file you can open.
Step by step
- Open the PDF/A Validator page and drop the PDF(s).
- Leave Standard to validate against on Detect automatically, or choose a level.
- Leave Group identical problems on unless you want the raw list.
- Press Validate PDF/A.
- Read the result panel: verdict, counts, problems. A JSON report is also offered for download. Validate another file resets the page.
📷 Screenshot placeholder: standard selector and a result with grouped problems.
Options explained
| Option | What it does | Default | Recommended |
|---|---|---|---|
| Standard to validate against | Detect automatically, or one of PDF/A-1A/1B/2A/2B/2U/3A/3B/3U/4/4E/4F or PDF/UA-1 | Detect automatically | Auto for files that should already be PDF/A; pick explicitly when an outside rule names a level |
| Group identical problems | Shows each distinct problem once with a count | on | On. One unembedded font may otherwise appear as dozens of identical rows |
The report keeps at most 500 distinct problems; if there are more, it says the list was truncated while still giving the full counts.
What you get
- On screen: valid or not valid, which standard was used and why (the one you requested, the one the file declares, or the fallback), error and warning counts, a summary by clause, and the problem list (severity, clause, category, page, object, description, number of occurrences).
- A downloadable report file named
pdfa-validation-report.json(machine-readable; for people it is easiest read on screen). - For several files, one report entry per file. A file that cannot be read gets a "Failed" entry with the reason; the rest are still checked.
- Your PDF is never changed.
Quality & limits
- It is a validation by the same engine used across Documentize; treat "valid" as strong evidence, but for a legally critical submission confirm with the receiving party's own validator (the book cannot promise that every validator agrees on every borderline case).
- "Not valid" is common and not a flaw in your content. Typical causes named in the site's FAQ: fonts referenced rather than embedded, colours without an output intent or ICC profile, transparency, encryption, JavaScript and external links.
- Standard limits and ~2 day retention apply.
Works well with
- Use before: Flatten and Optimize are the usual tidy-up steps; validate after them, since any step that rewrites the file can change the result.
- Use after: your PDF/A producing step (Converter or the original authoring tool). Run Compressor before validating, not after; if you compress afterwards, validate again.
- Recipes: Archive-ready (PDF/A) delivery, Digitise a paper archive.
Common mistakes & fixes
- "It fails against PDF/A-1B but my file says PDF/A-2B." Keep Detect automatically, so it is judged against the claim it makes.
- "Hundreds of identical rows." Turn Group identical problems on.
- "Which level should I pick?" The letter matters more than the number; ask the receiver which letter and part they accept.
- "I fixed one error and a new one appeared." Normal; hidden problems surface as earlier ones clear. Re-run.
- "It is valid but still not accessible." PDF/A-B does not guarantee accessibility; use the A levels or PDF/UA-1.
Same thing elsewhere
- API: the
pdfa-validateendpoint with an optionalstandardvalue. - CLI plugin and MCP tool: none found in this version of the book's sources; use the web page or the API.
Try it
Export any Word document to a normal PDF and validate it with Detect automatically: you will see the fallback standard (PDF/A-1B) and probably several errors. Re-run with the same file but choose PDF/A-2B, and compare the counts. Then switch Group identical problems off to see how a single defect repeats.