PDF Artifact
Content marked in a PDF as purely decorative, layout-related (page numbers, headers, rules), or pagination-related — telling assistive technology to ignore it. Artifacts are the counterpart to meaningful tagged content and are required for a fully accessible tagged PDF.
In a tagged PDF, every piece of content on a page falls into one of two categories: real content — text, images, and graphics that carry meaning and should be conveyed to assistive technology — and artifacts — content that is decorative, structural, or pagination-related and should be explicitly ignored by screen readers and other accessibility tools.
The artifact mechanism is the counterpart to tagged content. If tagged content says "this sequence of operators corresponds to paragraph 3 of section 2", artifact marking says "this sequence of operators is a page border decoration and has no semantic meaning — skip it."
Without proper artifact marking, screen readers announce visual noise alongside meaningful content — reading out page numbers mid-paragraph, announcing decorative rule lines as "graphic", or describing background watermarks before the actual document text. Correct artifact marking is one of the most common PDF/UA remediation tasks and one of the least glamorous — but it matters significantly for document usability.
Artifact types defined in PDF/UA
The <a href="https://www.iso.org/standard/51502.html" target="_blank" rel="noopener">PDF/UA specification (ISO 14289)</a> defines four types of artifacts that can be explicitly marked:
**Pagination artifacts** — Content generated by the layout engine that is unrelated to the document's information content: - Page numbers - Running heads and feet (repeated headers and footers) - Bates numbers (sequential IDs on legal documents) - Cut marks, registration marks, and colour bars (print production artefacts)
**Layout artifacts** — Visual elements that support the page design but carry no information: - Decorative rules and dividers - Background images and textures - Borders and frames around content areas - Column separator lines
**Page artifacts** — Content belonging to the physical page itself rather than the document: - Watermarks (page-level branding, "DRAFT", "CONFIDENTIAL" marks) - Background colour fills - Drop shadows and decorative page border elements
**Background artifacts** — Objects that appear behind real content and are purely visual: - Decorative background graphics - Hero images in section dividers that are purely aesthetic
In the PDF content stream, artifacts are marked using `/Artifact` as the tag in a `BMC` or `BDC` operator: ``` /Artifact BMC % simple artifact (no type specified) q 0.8 w 0 0 595 0 re S % draw a horizontal rule Q EMC
/Artifact <</Type /Pagination /Subtype /Footer>> BDC BT /F1 9 Tf 280 20 Td (Page 1 of 10) Tj ET EMC ```
Why artifact marking matters for screen reader users
To understand why artifact marking matters, consider what a screen reader does with an untagged or improperly tagged PDF:
Without artifact marking, a screen reader processing a page encounters all content in the content stream. For a typical business document, this might mean: 1. A background watermark image → "Graphic, DRAFT" 2. Page header text → "Company Name | Q3 Report | Confidential" 3. Actual heading → "Section 3: Revenue Analysis" 4. Page number in footer → "3" 5. Next page header → "Company Name | Q3 Report | Confidential" 6. Page number → "4"
A user navigating through a 50-page report hears the header text and page number announced on every page — dozens of interruptions with no informational content.
With correct artifact marking: - The watermark is marked `/Artifact <</Type /Background>>` — ignored - The header is marked `/Artifact <</Type /Pagination /Subtype /Header>>` — ignored - The page number is marked `/Artifact <</Type /Pagination /Subtype /Footer>>` — ignored - The heading is tagged as `/H2` — announced as "Heading level 2: Section 3: Revenue Analysis"
The difference in usability is substantial. This is why <a href="https://pdftheguide.com/glossary/pdf-ua" target="_blank" rel="noopener">PDF/UA</a> requires that all non-meaningful content be marked as artifacts.
How authoring tools handle artifacts
**Adobe InDesign** — InDesign automatically marks many common artifacts correctly in its PDF export: - Running headers and footers defined with InDesign's master pages are marked as Pagination artifacts - Decorative elements on master pages are typically marked as Layout artifacts - Manually placed decorative objects require the "Tag as Artifact" setting in the Object Export Options panel
**Microsoft Word** — Word's PDF export marks page headers and footers as Pagination artifacts automatically. Decorative elements inserted as images require the "Mark as decorative" checkbox in Format Picture > Alt Text.
**LibreOffice** — Headers and footers are marked as artifacts. Other decorative content requires manual configuration.
**Limitations of automatic marking** — All authoring tools struggle with complex layouts where the "decoration vs content" distinction is ambiguous. A background image might be a visual decoration (artifact) or a meaningful illustration (tagged Figure with alt text). The tool cannot determine this automatically — it requires human judgment.
**Adobe Acrobat Pro remediation** — For PDFs that need artifact marking corrected post-export: 1. Open the Tags panel 2. Right-click on tags that should be artifacts → Change Tag to Artifact 3. For page-level content not in the tag tree: use Touch Up Reading Order tool → select → mark as Background
The <a href="https://pac.pdf-accessibility.org/" target="_blank" rel="noopener">PAC accessibility checker</a> reports "content not tagged as real content or artifact" as a failure — this is the most common trigger for artifact remediation work.
Artifacts and PDF text extraction
Artifact marking affects not just screen reader accessibility but also automated text extraction. Tools that respect the PDF tag tree — including many AI document processing pipelines and high-quality OCR post-processing systems — skip artifact-marked content when extracting logical text.
This is valuable because: - **Running headers/footers** — If page headers repeat on every page, and those headers are not marked as artifacts, every page's extracted text starts with the same header string. This confuses AI chunking algorithms and degrades RAG (Retrieval-Augmented Generation) retrieval quality. - **Page numbers** — Page number artifacts in the middle of extracted text produce numbers that look like data values or sentence fragments. - **Watermarks** — Watermark text extracted as content produces repetitive "CONFIDENTIAL" or "DRAFT" strings throughout the document.
A PDF with correctly marked artifacts produces cleaner text extraction output than one without, even for non-assistive-technology use cases. This is another instance of the principle that accessibility investment benefits AI processing as well as human users with disabilities.
For the full picture of AI and PDF quality, see <a href="https://pdftheguide.com/news/pdf-and-ai" target="_blank" rel="noopener">PDFs and AI</a>.
Frequently asked questions
What is a PDF artifact?
A PDF artifact is content on a page that is explicitly marked as non-meaningful — decorative elements, page numbers, headers and footers, background images, watermarks, and other layout structures that are visual but carry no information for readers. Artifacts are marked using the /Artifact tag in the page content stream. Screen readers and accessibility tools skip artifacts, presenting only the meaningful tagged content to users.
Do decorative images need alt text or should they be marked as artifacts?
Decorative images that add no information — a coloured band, a geometric background pattern, a repeated brand element — should be marked as artifacts. Adding empty alt text (alt="") is the WCAG 2.x mechanism for the same purpose in HTML; in PDF, the correct approach is the /Artifact tag rather than an empty /Alt attribute. Images that provide informational content (photographs, diagrams, charts, infographics) must be tagged as Figure elements with meaningful alt text.
How do I check if page numbers and headers are correctly marked as artifacts?
Use the PAC accessibility checker — it shows a screen reader preview where you can see exactly what content would be announced, and flags content that is neither tagged nor marked as an artifact. In Adobe Acrobat Pro, the Tags panel shows the structure tree; content in page headers and footers that does not appear in the structure tree may or may not be correctly marked. The Touch Up Reading Order tool highlights content by category, making it easy to identify untagged, unannotated content.
Can artifact-marked content still be visible on the page?
Yes — artifact marking is purely about accessibility and text extraction; it has no effect on visual rendering. Artifact-marked content appears identically to non-artifact content when the PDF is viewed or printed. The /Artifact tag only affects whether assistive technology announces the content and whether text extraction tools include it in extracted text. Page numbers, decorative rules, and watermarks are all fully visible on the page whether or not they are correctly artifact-marked.