We tested whether our CV anonymisation actually removes a candidate's identifying details or merely hides them, and it removes them. Across five CVs in five different layouts, every identifying field that showed in the normal branded pack, the candidate's name, email, phone number and LinkedIn, was gone from the anonymised version. We checked by extracting the text from each anonymised PDF and searching it. Zero leaks.
This matters more than it sounds. "Redaction" fails in a very common way: a tool draws a black rectangle over a name or an email, the document looks clean, but the original text is still in the file and anyone can select, copy or search it back out. For blind recruitment and for GDPR, that is not anonymisation, it is a cosmetic cover. So the real question is not "does the name disappear from view", it is "is the data actually gone". We tested that.
Key takeaways
- We tested whether our one-click anonymisation actually removes a candidate's identifying data or just hides it. It removes it. Across five CVs and five layouts, every identifying field that appeared in the normal pack (name, email, phone, LinkedIn) was gone from the anonymised one, with zero leaks.
- The important part is how. The fields are deleted from the candidate data before the document is built, so there is nothing left underneath the page to recover. This is different from the common failure where a tool draws a black box over text that is still sitting in the file.
- We confirmed it by reading the output. We extracted the text from every anonymised PDF and searched it for the candidate's name, email and phone. None of them were there.
- The tool can anonymise more than we needed to here. Alongside name, email, phone and LinkedIn, it can remove address, photo, date of birth and age, gender, nationality, marital status, other social links, and even employer and school names. These CVs simply did not carry all of those.
- This is a correctness test on five hand-checked CVs, not an independent audit. Always review an anonymised pack before you send it, the same as any CV.
A redaction tool that draws a black box over a name leaves that name in the file as live text. Anyone can select it, copy it, or search it out, and a parser at the client's end reads it straight through the box. For blind recruitment and GDPR, that is not anonymisation. The only safe version removes the data before the document is built. That is what we set out to verify.
The result
Across five CVs in five layouts, every identifying field that appeared in the normal pack was gone from the anonymised one, and reading the output PDF confirmed none of it could be recovered. Sixteen identifying fields showed across the five packs (name, email and phone on every one, plus a LinkedIn URL on the complex CV), and all sixteen were removed. Zero leaks.
Removing data vs hiding it
When you turn anonymisation on, RefineCV does not paint over the candidate's details on the finished page. It deletes those fields from the candidate's data first, and then builds the document from what is left. The name, the email, the phone number, the address, are removed before a single line of the PDF is drawn, so they never reach the page and there is nothing underneath to recover.
That is the difference between removing data and hiding it. A black box over live text leaves the text in the file, selectable and searchable. Deleting the field upstream means the finished document was built without it, so a reader, a copy-paste, or a parser at the other end finds nothing.
You control what gets removed. The fields available are name, email, phone, address, LinkedIn, photo, date of birth and age, gender, nationality, marital status, other social links, and, when you want a fully blind sift, employer and school names. For a blind-recruitment pack you typically remove the identity and contact fields and keep the experience, which is exactly what we tested.
How we tested it
We used the same five golden CVs as our parsing and speed tests: a clean single-column CV, a dense CV, a two-column CV, a CV with a sidebar, and a complex CV with five employers. For each one we produced two branded packs from the identical template: a normal pack, and an anonymised pack with the identity, contact and demographic fields switched on for removal.
Then we read both. We extracted the text from each PDF and searched it for the candidate's real name, email address, phone number (digits only, to catch reformatting) and LinkedIn URL. A field only counts if it appeared in the normal pack in the first place, so we are measuring genuine removals, not fields that were never there.
Results by CV layout
For each CV we checked the normal pack for the candidate's identifiers, then checked whether any survived in the anonymised pack.
| CV layout | Identifiers in the normal pack | Recoverable after anonymising |
|---|---|---|
| Simple (single-column PDF) | Name, email, phone | None |
| Dense (one-column DOCX) | Name, email, phone | None |
| Two-column (DOCX) | Name, email, phone | None |
| Sidebar (PDF) | Name, email, phone | None |
| Complex (five employers, TXT) | Name, email, phone, LinkedIn | None |
We publish our own tests because a checkable, first-party result is worth more than a claim. This is five hand-checked CVs, not an independent audit, so treat it as a transparent sample. See the companion tests on parsing accuracy and formatting speed.
What this means for agencies
The result was the same on every CV: everything that identified the candidate in the normal pack was gone from the anonymised one, and the text extraction found nothing to recover. Sixteen identifying fields appeared across the five packs, and all sixteen were removed.
The denominator is honest and small on purpose. These CVs mostly carried name, email, phone and, on one, a LinkedIn URL, so those are the fields we could verify. The point of the test is not the count, it is the method: we did not trust that the fields looked gone, we read the finished file and confirmed the data was not in it.
For a recruitment agency this is the thing to get right. Blind recruitment only works if the client genuinely cannot see who the candidate is, and GDPR treats data that can be read back out as not anonymised at all. A pack built without the identifying fields clears both bars in a way a black box over live text does not. For the process side, how to decide what to strip and when, see our guides to anonymising a CV for blind recruitment and anonymising resumes for client submission.
Frequently asked questions
Does CV anonymisation remove the data or just hide it?
It should remove it, and in our test ours does. RefineCV deletes the identifying fields from the candidate data before it builds the document, so the finished PDF is created without them and there is nothing underneath the page to recover. We confirmed this by extracting the text from each anonymised PDF and finding none of the candidate's name, email or phone. This is different from redaction tools that draw a box over text that is still in the file and can be copied or searched out.
What can a CV anonymisation tool remove?
In RefineCV you choose the fields: name, email, phone, address, LinkedIn, photo, date of birth and age, gender, nationality, marital status, and other social links. You can also remove employer and school names when you want a fully blind sift. For most blind-recruitment packs, agencies remove the identity and contact fields and keep the experience so the client can still judge the candidate on their record.
Is anonymised CV formatting GDPR compliant?
Anonymisation helps, but GDPR compliance is about your whole process, not one feature. The relevant point here is that GDPR does not treat data as anonymised if it can be read back out, so a cosmetic black box over live text does not count. A document built without the identifying fields, which is what we tested, is genuinely stripped of that data. You are still responsible for consent, storage and the rest of your handling.
Should I still review an anonymised CV before sending it?
Yes, always. Our test shows the configured fields are removed cleanly, but you know your client and your obligations better than any default. Check that the right fields are stripped for this submission, that nothing identifying slipped into a free-text summary or a job description a candidate wrote about themselves, and that the pack reads well, before it goes out.
Does anonymising a CV break how it parses in an ATS?
It should not, if the anonymised pack is still clean single-column text, which RefineCV's is. Anonymisation removes fields; it does not add the columns, tables or images that scramble a parser. For how layout affects parsing, see our field-by-field accuracy test.
The bottom line
Anonymisation is only worth anything if the data is actually gone. In our test the identifying fields were deleted before the document was built, not hidden on top of it, and reading the finished PDF confirmed nothing identifying was left to recover. For blind recruitment and GDPR that is the behaviour that counts. As always, review the pack before it goes to a client.
Related reading: how to anonymise a CV for blind recruitment, GDPR for candidate CVs, and our guide to the best CV formatting software for recruitment agencies.
Send blind, client-ready CVs with the data actually removed
RefineCV anonymises the fields you choose and rebuilds the CV into your branded template. Start free with 10 CVs, no credit card. Then $0.40 per CV, or $50 per month for 200.
Sources
- RefineCV anonymisation test (2026): First-party test described in this post: five golden CVs rendered as normal and anonymised branded packs, with the output PDF text extracted and searched for identifying fields.