Last updated: June 2026.
How do you anonymise a CV without losing context? Remove identity — name, photo, contact details, employer names — but keep signal: seniority markers, scope, employer descriptors ("Big-4 consultancy", "FTSE 100 retailer"), quantified achievements and dates. Over-redacted CVs don't win interviews; clients reject what they can't evaluate. Done in a template, anonymisation takes seconds, not an editing session.
This is the craft post: not why to anonymise (that's the compliance conversation) but how to do it so the blind CV still sells the candidate.
Why does over-anonymisation lose interviews?
Because clients hire on context, and clumsy redaction destroys context faster than it protects identity. A work history reading "Software Engineer — [REDACTED], 2019-2023" tells a client almost nothing: was that a two-person startup or a global bank? A blind shortlist of such CVs forces the client to either reject candidates they can't evaluate or ask for the un-redacted versions — defeating the entire exercise.
The mistake comes from treating anonymisation as deletion rather than translation. The goal is to remove who while preserving what: the seniority, scale, sector and achievement evidence a hiring decision actually rests on. A well-anonymised CV reads like a complete story about an unnamed protagonist; a badly anonymised one reads like a leaked document. Recruiters running blind processes for diversity or client-relationship reasons live with the results — interview rates on blind shortlists track directly with how much evaluative signal survives redaction.
Which details must always be removed?
The non-negotiable redaction list is short and mechanical:
- Name — replace with initials or a candidate reference number.
- Photo — remove entirely; never blur (blurred photos still signal age, gender, ethnicity).
- Contact details — phone, email, address, LinkedIn URL and any personal website.
- Employer names — current and previous, including group companies and well-known project codenames.
- Identifying combinations — the subtle one. "Head of Risk at a Dublin-headquartered airline" identifies a person as surely as a name. So can a unique qualification plus a niche sector, or a one-of-a-kind job title.
Depending on market and client policy, you may also strip graduation years (age proxy), specific universities (background proxy) and nationality or visa detail beyond a simple right-to-work confirmation. What you should not strip by default: dates of employment, role titles, achievements, or the descriptive shape of each employer — those are signal, and they're handled by translation rather than deletion, as below.
How do you preserve seniority signals?
Seniority lives in scope, not in the company logo, so anonymisation must keep scope language intact and specific:
- Team and budget: "led a 14-person engineering team", "owned a £3.2m P&L", "managed a panel of 40 suppliers".
- Reporting line: "reported to the Group CFO", "two levels below CEO".
- Progression: keep the arc visible — three promotions within the same redacted employer is a stronger signal than three employer names.
- Trajectory of scale: "joined as the 4th hire; left at 90 headcount" sells growth-stage experience with no name attached.
- Quantified outcomes: "reduced time-to-fill by 30%", "grew the desk to £1.1m NFI" — numbers carry no identity but all the evidence.
The discipline is to resist trimming sentences during redaction. Editors instinctively shorten what they're censoring; in a blind CV the opposite is needed — wherever a name comes out, a measure of scale should remain or go in.
How do you describe employers without naming them?
Employer descriptors are the core craft. The formula is sector + scale + relevant distinction, calibrated so the description matches a class of companies (say, five or more plausible candidates) rather than one:
- Deloitte → "Big-4 professional services firm"
- Tesco → "FTSE 100 grocery retailer"
- Monzo → "UK digital challenger bank, 3,000+ staff"
- A niche employer → "specialist marine-insurance MGA, ~200 staff"
- A stealth startup → "venture-backed B2B SaaS scale-up, Series C"
Two tests before a descriptor ships. The reverse-search test: would a knowledgeable person in that market shortlist one company from the phrase? "Cupertino-based consumer electronics giant" fails; "top-5 global smartphone manufacturer" passes. The usefulness test: does the descriptor still tell the client what they need — regulated vs unregulated, enterprise vs startup, domestic vs global? "A company" passes anonymity and fails usefulness. Keep a house glossary of approved descriptors per sector so every recruiter translates the same employer the same way; inconsistency across a shortlist is its own identifier.
How granular should redaction be?
Match granularity to the disclosure stage rather than applying one blunt setting:
- Stage 1 — longlist: fullest redaction. Initials, employer descriptors, no education institutions, graduation years removed.
- Stage 2 — shortlist: same identity redaction, but restore anything the client now legitimately needs — certification dates, security-clearance level, location to region.
- Stage 3 — interview: identity revealed with candidate consent; the CV un-redacts entirely.
Granularity also varies by section. Free-text profiles and achievement bullets need a careful pass — identity leaks through phrases like "rejoined my previous employer" or "the UK's first such programme" — while skills and education sections are usually mechanical. This staged approach is exactly where doing it manually breaks down: maintaining three redaction depths per candidate in Word is unsustainable. In a template-driven tool, redaction rules live in the template, so the same source CV renders at any stage in seconds — FormaCV's anonymisation works this way per template, alongside AI CV formatting automation, at $0.99 per CV.
What do clients still need to see?
A blind CV succeeds when the client can answer their real questions without knowing who the candidate is. Preserve, always:
- Complete dated career history — gaps and tenure patterns are legitimate evaluation material.
- Role titles and scope — with the seniority signals above intact.
- Employer class — sector, scale, regulatory context via descriptors.
- Quantified achievements — the evidence layer; numbers are anonymity-safe.
- Skills, methods, technologies — including depth indicators ("7 years production Python").
- Right to work and notice period — stated plainly, no documents attached.
Agree the redaction standard with the client before the first shortlist: what's hidden, what's preserved, when identity is revealed. That conversation prevents the classic failure where a client quietly googles their way to the candidate's identity — or rejects the shortlist as unreadable. For the legal side — lawful basis, candidate rights, retention — see our GDPR CV anonymisation guide; this post is the craft layer on top of that compliance floor.
Anonymisation done well is invisible: the client never feels information is missing because the information that matters was never removed — only translated. Build the descriptors into your templates once, and every blind CV your agency sends tells the full story of a candidate nobody can yet name.