A PDF is a portable page description, not simply a stack of pictures. The format can contain fonts, vector paths, images, interactive form fields, annotations, links, metadata, embedded files and digital signatures. That flexibility explains why two documents that appear identical may behave differently when you search, merge, print, redact or copy text. This handbook gives you a way to diagnose that behavior before modifying the file.
Key distinction: A PDF that looks correct is not necessarily searchable, accessible, safely redacted, or digitally signed in a valid state. Each property needs its own check.
1. The parts of a PDF
At a high level, a PDF contains numbered objects describing pages, fonts, drawings, streams, references and document structure. Page content instructions tell a renderer which text glyphs, paths and images to draw. A cross-reference mechanism helps locate objects. Some documents use object streams and incremental revisions; an edited PDF may retain historical content or append new revisions rather than rewriting everything from scratch.
This matters operationally. Putting a white rectangle on top of a name is only a visual overlay. The underlying text can still exist in an earlier content stream or object. Likewise, a file renamed with .pdf is not guaranteed to contain valid PDF syntax. Open unknown files in a trusted, updated viewer and be cautious of embedded active content or attachments.
2. Digital text versus scanned pages
A digital PDF usually has text represented by positioned glyphs and associated character information. Text selection, search, copy/paste and screen reader behavior can work, though font encodings and tagging affect quality. A scanned PDF may contain nothing but a bitmap of each page. You can see words, but the computer only sees pixels until an OCR engine produces a text layer.
OCR (optical character recognition) estimates characters from the image. Its accuracy depends on image resolution, skew, print quality, language, handwriting, page layout and the chosen OCR model. OCR may confuse 0 with O, 1 with l, or reorder columns. The presence of selectable text does not establish that OCR was correct.
Quick diagnosis: try selecting a single sentence, searching for a visible uncommon word, and copying a table into a plain-text editor. If search fails but the page looks readable, it may be image-only. If copy/paste returns bizarre symbols, font encoding or OCR may be the issue. For legal or medical documents, review OCR output manually rather than trusting it without verification.
3. Merge, split and rotate without losing meaning
Consider three inputs: cover.pdf (1 page), notes.pdf (4 pages), and appendix.pdf (2 pages). A correctly ordered merge should produce 7 pages. That is an arithmetic expectation, not a claim that the result has passed content inspection. Page sizes can differ, bookmarks may be lost, and page labels (roman numerals versus normal numbers) may not map neatly to the new physical order.
- Open the PDF Merge tool. Load a small non-sensitive test set or download our two sample PDFs.
- Arrange the files deliberately and merge. Save a new file rather than overwriting your only original.
- Check the page count, first and last page, internal links, orientation and rotation. Test search on one representative page.
- Open the result in a second reader if the document is business-critical. Verify embedded signatures separately because merges and edits can invalidate them.
Splitting can break cross-document links, bookmarks and attachments. Rotating can mean changing the page's view rotation or actually transforming content coordinates; the visual result may look similar while other consumers behave differently. Use PDF Split, PDF Rotate and PDF Organizer for their stated purposes, and verify results in a viewer.
4. PDF compression: image-heavy versus font-heavy documents
A 15 MB photographic PDF and a 15 MB vector diagram can need completely different treatments. Raster images may dominate size because of pixel dimensions and embedded encodings. Pages made mostly from fonts and vector paths may already be compact. Recompressing scanned photographs or downsampling images can help, but text may become unreadable or barcodes fail.
Check whether the file includes high-resolution page scans. A 300 dpi A4 image contains many millions of pixels; converting a color scan to a lower-quality JPEG can dramatically change size, but it may remove fine marks. If you need durable records, do not blindly downsample or flatten document structure just to get below an email attachment limit.
Better workflow: choose an appropriate export profile at the source application, optimize pictures before embedding, remove unused attachments when you have permission, then visually and functionally compare the PDF. “Save as PDF” from different programs does not imply identical output sizes or accessibility. If an upload service has a hard limit, consider distributing a link rather than destroying required readability.
5. Redaction versus covering information
Redaction permanently removes sensitive content and potentially associated representations; an opaque white box only hides it from the current view. Inadequate redaction can leave copyable words, hidden layers, annotations, metadata, thumbnails, attachments or previous incremental revisions containing the original information. A “whiteout” or drawing feature should not be treated as certified redaction.
For sensitive content, use a dedicated redaction workflow that searches and removes the underlying information, sanitizes additional sensitive data when applicable, saves safely, and then verifies by text search, selection, extraction and independent inspection. Depending on the jurisdiction and stakes, professional document procedures may be necessary.
Our PDF Editor may be useful for ordinary annotations and page edits, but it is not a substitute for a validated high-assurance redaction tool. Read the PDF Redaction and OCR Guide before working with confidential material.
6. Password protection is not the same as a valid signature
PDF encryption may restrict opening or particular operations; actual effectiveness depends on the algorithm, reader compliance and password strength. Permission flags such as “do not print” are not a strong guarantee that content cannot be copied by every tool. Encryption protects a document differently from access control on the service that distributes it.
A digital signature uses cryptographic verification and an established identity/trust model to check whether signed bytes have changed. Adding a handwritten-looking signature image is not the same as creating a digital signature. Editing, merging, re-saving or sanitizing a digitally signed PDF can invalidate the original signature or affect what the signature covers. Preserve the signed original and verify signatures with a trusted validator before and after any permitted workflow.
If a PDF contains a suspicious link or attachment, opening it with a vulnerable application can carry risk. Use updated software and handle untrusted documents cautiously, especially when your conversion path relies on a third-party library.
7. Accessibility: reading order, tags and alternative text
A visually attractive PDF can still be difficult for users who navigate with a keyboard or screen reader. Tagged PDF supplies structural information for headings, paragraphs, lists, tables and other elements. A logical tag order, correct tab sequence, meaningful link labels, image alternatives and language settings make the document more understandable to assistive technology.
Bad reading order is especially visible with multi-column layouts. A naïve extraction might read across two columns rather than down one column at a time. A page that has a text layer but no semantic structure is not automatically accessible. Good authoring practices begin in the source document: meaningful headings, properly marked lists and tables, and descriptive alternatives for relevant images.
Accessibility check: navigate the PDF using only a keyboard, inspect its logical structure with an accessibility tool, verify heading order and check how a screen reader reads columns and tables. The W3C's PDF3 technique explains why reading and tab order matter.
8. Archival PDFs and the PDF/A family
PDF/A is a set of ISO standards intended for long-term preservation. It constrains features of PDF and imposes requirements on aspects such as fonts and color information so the document can remain reliably rendered. A random file with .pdf is not automatically PDF/A; changing the filename cannot make it archival-compliant.
Archiving also requires the surrounding records-management workflow: stable identifiers, secure storage, backups, checksums, an agreed retention policy and periodic validation. Even a valid PDF/A file can become lost or inaccessible if the storage system fails. For evidence or regulated data, verify the exact required conformance level using a professional validator.
9. Troubleshooting decision table
| Symptom | Likely causes | First investigation |
|---|---|---|
| Cannot select text | Scanned page / outlines | Test OCR on a copy and compare output |
| PDF opens blank in one viewer | Unsupported feature, damaged stream | Try updated trusted viewer; verify source integrity |
| Merge creates unexpected order | Input order / page labels | Recheck physical pages and first/last page |
| Text copied as gibberish | Font mapping / bad OCR | Compare rendered page to extracted text |
| File size unexpectedly high | Large images, attachments, embedded fonts | Inspect content and export settings |
| Annotation survives after “delete” | Hidden annotations/revisions | Use proper removal and verify exported PDF |
| Signature marked invalid | Modified signed bytes or trust-chain issue | Preserve original and examine validation report |
| PDF inaccessible by keyboard | Missing tags/tab order | Inspect structure and repair at authoring source |
10. A safe PDF preparation checklist
- Keep an unchanged original and work on a separate copy.
- Identify whether text is selectable and whether OCR is needed.
- Check page count, orientation, bookmarks, links and reading order after operations.
- Distinguish annotations from true redaction; verify sensitive data removal independently.
- Treat encryption, permissions, and signatures as three separate concepts.
- Reopen critical output in a second PDF reader before sharing.
- Use an appropriate accessibility or archival validator when the document requires it.
Try it now: PDF Merge · PDF Split · PDF Editor · Images to PDF · PDF sample files.
Technical references
- PDF Association: PDF specification archive — ISO 32000 and historical specifications.
- W3C: Correct tab and reading order in PDF — accessibility techniques.
- NowConverter: OCR and scanned PDFs — an accompanying practical guide.