What does a veraPDF report show?

The report’s profile identifies the target tested. A PDF/A result is not a PDF/UA result. The PDF/UA version must also match the brief. veraPDF documents profiles for PDF/UA-1, PDF/UA-2 and WTPDF; check what the installed version actually supports.

Sources: [1]

For a PDF/UA-1 brief, explicitly select that profile. The documented CLI syntax supports verapdf --flavour ua1 report.pdf; ua2 selects PDF/UA-2. List installed profiles with verapdf --list and use a version supporting the agreed target.

Automatic selection can use identifying metadata and fall back to another profile when identification is insufficient. Always read the profile name in the result itself. A successful PDF/A-1b run answers a different question from PDF/UA validation, even for the same file.

Start with a small trial and confirm the report identifies the correct file and profile and contains an actual validation outcome. For a directory of files, keep a traceable record per file. One batch can contain both completed assessments and files that could not be processed.

Sources: [2]

A rule about an object may run against many objects in one file. Failed checks can therefore outnumber failed rules. Do not compare these counts as if they meant the same thing. Keep the full report, not just its green or red summary.

Imagine one rule failing on twelve objects. The report may show one failed rule and twelve failed checks. Both counts can fall after repair, but their ratio is not an accessibility score. One missing description of the central information graphic may matter more than numerous minor structural defects.

To diagnose the cause, follow the reported rule and object information to the relevant file content. Keep the original report alongside any readable summary. “55 errors” alone rarely supports reproducible investigation.

Our synthetic before-file produced 6 failed rules and 55 failed checks. The after-file produced 0 failed checks using veraPDF 1.30.2 and the PDF/UA-1 profile. These are measurements of two specific files, not an overall accessibility percentage. See the before-and-after guide for the limits of the example.

• PDF filename and SHA-256.
• veraPDF version and selected profile.
• The complete machine-readable report.
• Time, execution status and any tool errors.
• Separate findings for matters outside the automated profile.

If the tool stops or cannot read the file, record that validation could not finish. Do not treat an empty list of reported document failures as a pass in that situation.

Distinguish starting the process, reading the file, running the correct profile and passing its rules. An automated pipeline checking only for an empty failure list can mistake an interrupted run for success. Record execution status separately from document results.

A completed result with no failed checks still has limits. It cannot independently establish whether an alternative explains the right information, a heading is intelligible or a person can complete an important reading task. Keep those assessments alongside the machine report.