BASALT · JOURNAL
Redacting client names from work you want to show publicly
A portfolio piece, a case study, and a conference deck built from real work all raise the same problem. To redact client names from documents convincingly, you have to remove much more than the name, because a logo in the corner, a letterhead color, a file name in the download link, and a timestamp inside an embedded screenshot each identify the client as reliably as the word itself.
What identifies a client besides the name
Reviewers look for the name and stop. Everything below identifies without the name appearing.
Logos and brand marks. A logo on a cover page or in a header is recognized instantly and survives text redaction, because it is image or vector content. Vector logos are worse than raster ones, since the drawing instructions may carry the brand name as a path label or as live text.
Letterheads and templates. A distinctive header block, a color, a typeface pairing, and a footer layout are recognizable to anyone in the client's industry. A deck built in their own template announces them on every slide.
Product and project names. Internal code names, product SKUs, and campaign names are often more searchable than the company name, since they appear in press coverage and job listings.
Domains and URLs. A @company.com address in a signature block, a staging URL in a screenshot, or a link annotation whose visible text and target URI disagree.
Named individuals. A stakeholder, an approver in a sign-off block, a signature image, or an author name in a comment. Any of them leads to the organization in one search.
Screenshots of the client's own systems. A dashboard with the account name in the corner, a browser window with a bookmarks bar, a chat sidebar, a ticket number with a familiar prefix.
Numbers that are public. Revenue, headcount, and contract values can match a public record and identify the company in an anonymized case study.
Where the name hides after you have redacted the page
The content stream. Text is painted by operators whose operands are character codes, so a black box on top is one more drawing instruction and the codes underneath remain extractable. If the PDF is posted publicly, anyone can run a text extractor over it.
Document metadata and XMP. These typically carry the title, the author account, the producing application, and the original file path. A path like /Users/dana/Work/Clients/Northwind/Q3_deck_final.pdf names the client, the project, and you, and the title field is often the client's own, carried over from an export. See removing PDF metadata on a Mac.
Image metadata inside embedded pictures. A JPEG on a page usually keeps its EXIF block, which can carry GPS coordinates from a site visit, a capture timestamp, a camera serial number, and sometimes an owner name. Clearing the PDF's metadata does not touch these, since they live inside the image objects.
Annotations and comments. Review comments carry names, dates, and often blunt internal commentary, stored independently of the page.
Attachments, hidden layers, and earlier revisions. An embedded source file or a switched-off layer holding the un-anonymized version is fully present, and an incremental save keeps the original page reachable inside a file that displays correctly. Hidden data in PDFs covers these.
The file name itself. This sits outside the PDF and catches people constantly: a file served at /portfolio/northwind-case-study.pdf identifies the client before anyone opens it.
How to redact client names from documents without gutting the work
Decide first what the piece needs to demonstrate. Most case studies survive anonymization intact, because the value is in the reasoning, the structure, and the outcome, none of which depends on the client's identity.
Inspect the file before marking. Check metadata, XMP, annotations, attachments, hidden layers, and any OCR text layer, and clear those first so nothing you do later reintroduces them. Inspecting what a PDF contains covers it.
Build the value list: the company name in every form, including the legal entity name, abbreviations, the stock ticker, product names, project code names, the domain, employee names, and office locations. Then search and redact each value across the whole document, which is how you catch the header on page 30 and the footnote you did not read.
Handle images deliberately, since they are where most portfolio leaks live. Logos, letterhead marks, screenshots, photographs, and signature blocks need pixels destroyed and the image re-encoded, not covered. Redacting images inside a PDF explains why a drawn rectangle leaves the original intact. Then rename the file, remove the client name from any download URL, and write the PDF as one generation.
Substitution reads better than a page of black boxes
For public-facing work, removing a name is often less useful than replacing it. A case study that says "a national retail bank" reads as considered writing. The same case study with 60 black rectangles reads as a document that was not meant to be shared, and invites the reader to guess.
Substitution has one requirement: the substitute has to be applied everywhere the original appeared, with the same discipline as a redaction. An alias used in the body while the original survives in a header, a chart label, or the metadata is worse than no alias, because it advertises that there was something to hide.
Separately, confirm what your agreement with the client permits before publishing work derived from it. That is a contractual question, not a technical one.
Frequently asked questions
How do I remove a client name from a PDF portfolio?
Remove it from the page content, the metadata, the XMP packet, annotations, attachments, and the file name, then verify by extracting the text of the finished file and searching for the name. A black box leaves the characters in the content stream where any extraction tool reads them. Search the whole document rather than page by page, since headers repeat.
Do I need to remove the client's logo too?
Yes, in almost every case, because a logo identifies faster than a name and survives text redaction entirely. Logos are image or vector content, so removing them means destroying the pixels and re-encoding the image, or removing the vector object, rather than covering it. Vector logos can also hold the brand name as live text.
Can someone tell who my client was from the file metadata?
Frequently, yes. PDF metadata commonly carries the original file path, a title inherited from an export, the author account name, and the producing application, and a path containing a client folder identifies the account immediately. Embedded photographs and screenshots carry their own metadata, so strip both.
Should I use a fake client name instead of a black box?
A consistent alias reads better than a page of black rectangles and communicates the same work, provided it is applied everywhere the real name appeared. The risk is partial substitution: an alias in the body while the real name survives in a chart label or the metadata is more revealing than no alias at all.
Does renaming the file protect the client's identity?
Renaming is necessary and not sufficient. A file called northwind-case-study.pdf identifies the client from a link alone, so the name has to change, but renaming does nothing about the content stream, the metadata, the embedded images, or the annotations. Treat it as one item on a list rather than as the fix.
How do I anonymize screenshots in a case study?
Destroy the pixels in the identifying regions and re-encode the image rather than drawing boxes over the screenshot, and check the whole frame rather than the obvious label. Account names appear in sidebars, browser tabs, bookmark bars, and window titles. Also strip the screenshot's own metadata.
Will a black box over a client name hold up if the PDF is posted publicly?
No. A publicly posted PDF gets downloaded, indexed, and parsed as a matter of routine, and text under a drawn rectangle is recovered by copying it or by running a standard extraction tool. Because the file is public and permanent, a covered name is discovered eventually rather than possibly. The characters have to be removed from the file.
How do I check that a client name is gone from my PDF?
Extract the text of the finished file with a parser other than the tool that wrote it and search the output for the name, the abbreviations, the domain, and the project code names. Then extract the embedded images and confirm logos and screenshots are destroyed rather than covered, read the metadata separately, and check for more than one saved revision.
Doing it in Basalt
Basalt makes this a short job rather than a nervous one: the Inspector reports what the file still carries, and search and redact applies a client name, a domain, or a project code name across the entire document, so nothing survives in a header you skimmed. Redaction destroys content rather than covering it, with glyphs excised from the content stream, image pixels under a mark destroyed and the image re-encoded so a logo or screenshot is genuinely gone, and metadata, attachments, and hidden layers stripped in one generation. A verifier then re-opens the written bytes with an independent parser and proves the content is absent before anything is saved, and if it cannot prove that, nothing is written. Nothing is uploaded, since the engine has no network entitlement, enforced by macOS at the code signature level. It is $29 once, lifetime, up to three Macs, free for 24 hours, and removing personal information from PDF files covers the broader case.
Redaction that proves itself
Basalt destroys the content you mark, then re-opens the file it wrote and proves the content is gone before it saves anything. Your documents never leave your Mac.