Legal eDiscovery PDF Workflow: Redaction, Bates Numbering, and Production Sets
eDiscovery production deliverables fail review when redactions are reversible or Bates numbers don't match the load file. Here's the production-ready workflow.
Priya Patel·May 8, 2026·7 min read
Why "Just Black Out the Text" Fails Review
A common mistake in legal document production: drawing a black rectangle over sensitive text in a PDF. The visible text is hidden. The underlying text layer remains intact. Anyone with Adobe Acrobat (or a free PDF text extractor) can copy and paste the "redacted" content.
This isn't a technicality. It's a routine sanction. Federal courts have sanctioned firms for inadequate redaction practices going back to 2008. The production looks fine to a junior associate but fails any technical review.
This post covers the production-ready eDiscovery PDF workflow: proper redaction (text removed, not hidden), Bates numbering that survives load file matching, production set requirements, and the privilege log integration. For underlying PDF tools, see our PDF converter.
Text removal: the underlying text layer is removed (not just hidden)
Image removal: any embedded image data containing the text is overwritten or replaced
Metadata removal: any PDF metadata containing the text is stripped
Acrobat Pro has a "Redact" tool (Tools > Redact) that does all four. The free Acrobat Reader does not have this tool. Third-party PDF editors with a "Redact" feature usually do all four; "drawing tools" with just black rectangles do not.
Tool
Redaction
Notes
Adobe Acrobat Pro DC
Yes
Industry standard
Foxit PhantomPDF
Yes
Cheaper alternative
Nitro Pro
Yes
Comparable
iText (programmatic)
Yes
Custom redaction tools
Preview (Mac)
No
Black rectangles only
GIMP / Photoshop
Partial
Visual only, no PDF text removal
LibreOffice Draw
Partial
Visual only
For real eDiscovery work: Acrobat Pro DC, Foxit PhantomPDF, or Nitro Pro. Don't use Preview, drawing apps, or "black rectangle" tools.
Acrobat Pro Redaction Workflow
For each document:
Open in Acrobat Pro
Tools > Redact
"Mark for Redaction" > select text by dragging
Apply > Redact (this removes the text layer)
Save with new filename
After redaction, verify:
Tools > Edit PDF > select the redacted area
The text should not be selectable (no highlight on click)
Run Edit > Find and search for redacted terms; should not appear
If Find still shows redacted text, the redaction didn't apply. Re-do with the proper Redact tool.
Programmatic Redaction at Scale
For thousands of documents:
from pypdf import PdfReader, PdfWriter
import re
# Pattern for SSN
ssn_pattern = re.compile(r'\d{3}-\d{2}-\d{4}')
reader = PdfReader("input.pdf")
writer = PdfWriter()
for page in reader.pages:
text = page.extract_text()
matches = ssn_pattern.finditer(text)
for match in matches:
# Use pypdf or qpdf to overlay black rectangle
# AND remove text from text layer
# ... implementation depends on PDF library
pass
with open("output.pdf", "wb") as f:
writer.write(f)
For production-grade programmatic redaction, use specialized tools like pdftools (commercial), iText (commercial Java), or Adobe Acrobat's Acrobat Action Wizard for batch processing.
Acrobat assigns sequential numbers across all selected files. Save each PDF with its assigned number.
Production Sets
A "production set" is the collection of documents delivered to opposing counsel:
Original PDFs (or images)
Bates-stamped versions
Load file (index of documents with metadata)
Privilege log (documents withheld with explanation)
The load file is typically a .DAT or .CSV with columns:
Column
Content
BeginBates
First Bates number on the document
EndBates
Last Bates number on the document
Custodian
Person/entity who held the document
DocDate
Document creation date
FileName
Original filename
Pages
Page count
MD5
Hash for integrity
Standards for load files include Concordance (.DAT) and Relativity formats. The production agreement specifies which.
OCR Before Redaction
For scanned documents, OCR before redaction. Otherwise the text isn't searchable for the redaction tool to find sensitive content. See Searchable PDF With OCR for the OCR pipeline.
After OCR, the text layer is searchable. Use Acrobat's "Search and Redact" to find and redact patterns (SSN, account numbers, names) in bulk.
Privilege Log
Documents withheld for privilege reason are listed in a separate privilege log:
Column
Content
Privilege Type
Attorney-client, work product, etc.
Description
Brief description of subject matter
Date
Document date
Author
Sender/creator
Recipients
Recipients (To, CC)
Bates Range
If withheld in part, identify pages
Privilege logs are produced separately from the document set. They identify documents the producing party claims privilege over without producing the documents themselves.
Common Issues
Redacted text appears in search: redaction wasn't applied properly. Re-do with Acrobat Pro's Redact tool, not drawing tools.
Bates numbers don't match load file: numbering happened after load file generation. Re-generate load file from final Bates-numbered set.
Documents appear in wrong order in Concordance: load file column order doesn't match Concordance import expectations. Check the production specification.
Metadata still shows author after redaction: PDF metadata (Document Properties > Description) wasn't sanitized. Use Acrobat's "Sanitize Document" feature to remove all metadata.
Image-only PDFs can't be redacted: redaction tool needs a text layer. OCR first.
Pro Tip: Always run a "post-redaction QC" pass on a sample of redacted documents. Open in Acrobat, search for the redacted terms, and verify they don't appear. Catching one bad redaction in QC is much better than receiving a sanctions motion three months later.
Frequently Asked Questions
How do I redact a specific person's name across hundreds of documents?
Use Acrobat's "Search and Redact" feature, search for the name. Acrobat finds and marks all instances. Apply redaction to all marked instances. Verify a sample after.
Can I redact images?
Yes. Acrobat's redact tool can cover image regions. The image data within the redacted region is overwritten in the saved PDF.
What about embedded files (Excel, Word)?
PDFs can contain embedded files. These have their own metadata and content. Acrobat's "Sanitize Document" removes embedded files, or you can selectively remove them via Files panel.
Should I produce native files or PDFs?
Production agreements vary. Native (original) files preserve metadata and are required in some agreements. PDF/TIFF productions are easier to Bates-stamp and redact. Most modern productions are TIFF-converted with metadata extraction.
What about email attachments?
Each attachment is typically Bates-numbered separately. The load file links them via a "Family" or "Parent" field. The producing party documents the family relationship.
How do I handle accidentally privileged disclosure?
Federal Rule 502(d) protects against waiver if procedures are documented and the production is timely clawed back. Notify opposing counsel immediately, document the unintended disclosure, and request return.
For eDiscovery PDF production: use Acrobat Pro DC's redaction tool (not drawing tools), sanitize metadata, OCR before redaction, Bates-stamp the production set, generate load files in agreed format, document privilege withholdings in a separate log. Verify by searching for redacted terms in the output. Our PDF converter handles format-conversion steps; redaction itself requires Acrobat Pro or equivalent.
eDiscoveryPDF redactionBates numberinglegalproduction set
About the Author
Priya Patel
UX researcher and technical writer exploring document accessibility, font technology, and cross-platform compatibility.
