How to Run Blind Resume Screening, Step by Step

To run blind resume screening, decide what to hide (name, photo, contact details, address, dates of birth, school names and anything else that is not about doing the job), have someone not involved in scoring remove it, score every anonymised CV on the same written must-haves, then reveal identities only after the shortlist is set.

What blind screening is and what it is not

Blind screening means reviewing applications with details removed that could trigger unconscious bias, so the first decision rests on skills and evidence. It tries to reduce the effect of names, photos, ages, addresses and similar signals on who gets shortlisted. It is a useful step, but it is not a cure. It does not fix a biased job description, a narrow sourcing channel, or bias in the interview that follows. It can also hide information that is truly relevant, for instance the name of an employer that is a direct competitor or a licence that is required. Treat it as one layer in a fair process, alongside written must-haves, structured interviews and a record of how decisions were made. Be honest with yourself about what it can and cannot achieve, and do not describe it to candidates or managers as a guarantee of fairness. Some teams use blind screening for the shortlist only and then use structured interviews, a common pairing, because the two address different stages. Others use it only for roles where they have seen the shortlist lack variety. Whichever way you use it, decide in advance what you will measure to know whether it helped, such as how many people from different sources reach interview, using only data you are lawfully allowed to hold. If you cannot measure it, at least write down why you chose to do it, so that a later review has something to compare against.

Step 1: write the must-haves first

Blind screening only works if reviewers know what they are looking for. Before any CV is opened, agree three to six must-haves with the hiring manager, in plain words, with what counts as evidence. For example, two years of handling customer complaints, shown by job descriptions that name complaint handling, or a recognised accounting qualification. Write a short scoring scale, such as 0 for no evidence, 1 for some, 2 for clear, and say what each means. Without this step, reviewers fall back on impressions, and removing names simply moves the bias to other clues. Keep the list short. Skills the person could learn in the first month are not must-haves.

Step 2: decide what to hide

A sensible starting list is name, photo, email, phone, home address, date of birth, age, gender or title (Mr, Ms), marital status, nationality, social media links and any statement of health or family. Many teams also hide school and university names and graduation years at the first stage, because they can signal class, age or background, but keep the qualification and level if the role requires it. Consider hiding employer names if they would cause snap judgements, but keep the job title, duties and results. Do not hide anything legally required for the role, such as a licence, or the right to work check, which you can do later. Write your list down so that every redaction is consistent between candidates and across reviewers. Think about each field in terms of whether it helps predict doing the job. Contact details do not. A university name rarely does, though a degree subject or professional qualification might. The names of previous employers sometimes do, for example where experience in a regulated industry is a must-have, so you may keep the industry but hide the name. Dates are a judgement call: you need enough to see how recent and how long the experience was, but you can show years of experience instead of birth year or graduation year. Where a CV includes a personal statement with identifying detail, redact those sentences too. Document these choices once and reuse them.

Step 3: redact the applications

The person who redacts should not be the person who scores. In a small team, one recruiter can remove details and a colleague can score. You can redact by hand with a copy of the CV, by using a redaction feature in your tools, or by asking applicants to fill a form with skills and experience only. Give each application a neutral reference, such as a number, and keep the key linking numbers to names in a separate place. Check the result: names often appear in file names, headers, footers, email signatures and a cover letter's last line. Photos can sit inside a PDF. Be aware that some details are hard to remove, such as a distinctive club, language or a long career with one well-known employer, and accept that blind is not perfect. If you collect applications through a form, you can reduce the work by designing the form to ask only for what the first review needs: work history, skills, qualifications, availability and answers to two or three job-related questions. Contact details can be collected in a separate step after the shortlist. If you accept CV uploads, build a short redaction checklist for the person doing the work: name at the top, header and footer, file name, photo, links, address, and the last lines. A second person should spot-check a few redacted files before scoring begins, because one missed name can undo the point for that file.

Step 4: score each CV against the must-haves

Each reviewer scores every anonymised CV on the same scale, and notes the evidence they found. Do not give an overall impression score before the individual scores. If two reviewers score the same CV, they should score separately and then compare, discussing only large differences against the written criteria. Avoid reading CVs in a different order or at a different time of day without noticing. Fatigue changes judgement, so set a pace you can keep, and take breaks. Keep notes factual: the candidate ran a team of six, not seems senior. The scores produce a ranked list, but a person still decides who is shortlisted. Where you have more than one reviewer, track agreement. If reviewers often disagree on a particular must-have, the wording is probably unclear, so rewrite it with an example of what counts and what does not. Keep a short log of questions that came up, since they point to gaps in your criteria. If you use any automated ranking, review how it reaches its results and keep a person responsible for the decision, because some countries restrict decisions made solely by automated means. Never let a score replace judgement about evidence you can read.

Step 5: set the shortlist, then unblind

Choose the cut-off before you look at names, for example everyone who scores at least 8 out of 10, or the top eight. Applying a cut-off set in advance prevents you from adjusting the list after you see who is on it. Then reveal the identities to the people who need them to arrange interviews. Check the shortlist against the pool for anything unexpected, using only the data you are lawfully allowed to hold and in line with your privacy rules. If the shortlist is very narrow, review your must-haves and your sourcing before you blame the screening. Carry the written scores forward to the interview stage, so that the first decision remains on record. After unblinding, watch for the temptation to undo the work. Do not move someone up because you recognise their name or down because of a detail you now see. If you spot a conflict of interest, such as a reviewer who knows a candidate personally, handle it openly by asking someone else to take over that case. Record the final shortlist, the cut-off and the date, so that you can explain later how it was formed. Give feedback to unsuccessful candidates in the usual way, based on the must-haves, not on the redacted details.

A worked example

An invented example. A team has 120 applications for a customer success role. The must-haves are: handles escalations, writes clearly, uses a support tool, and has worked with business customers. One recruiter redacts names, photos, addresses, school names and dates of birth from all 120 and gives each a number. Two colleagues each score every CV 0 to 2 on the four points, giving a score out of 8. They compare, resolving differences of more than 2 points by rereading the evidence. The cut-off, set beforehand, is 6 out of 8, and 19 CVs pass. Only then does the recruiter match numbers to names, check for duplicates and invite people to phone screens. The notes on each score are saved with the record. The numbers are only illustrative. After the shortlist is set, the team compares the outcome with their usual process on a small scale, by asking whether the blind step surfaced anyone they would have overlooked and whether anyone they expected to see was missing. They write down what they notice without drawing large conclusions from one round. Over several roles, those notes show whether the step is worth its cost. This habit matters because it is easy to feel virtuous about a process without checking what it does.

Limits, risks and common mistakes

Redacting only names but leaving a photo, a graduation year or a postcode that reveals a lot. Using blind screening, then letting bias back in at an unstructured interview. Hiding information that matters, such as a required licence. Having the same person redact and score, so they already know who is who. Changing the cut-off after seeing names. Believing blind screening alone solves a diversity problem, when the pool of applicants may be the real issue. Forgetting that some roles need contact with the candidate early, such as roles requiring a particular language. Review the results over time, using only lawfully held data, to see whether the stage is doing what you hoped. Another limit is that blind screening applies to the first look only. Once names are revealed, the interview is where bias often returns, so keep the same questions and scoring scale for every candidate, and avoid informal chats before the formal interview. Also be careful about candidates who need adjustments, such as extra time or a different format. Blind screening should not stop you from offering them, and you should invite requests in your invitation. Finally, be honest in your advert: you can say that applications are reviewed without names and personal details, but avoid claiming that the process is free of bias.

Rules differ by country: check local law

Whether and how you may collect, store and redact personal data, whether you may ask for photos or dates of birth in the first place, and what you must do about equal opportunity monitoring differ by country and state. Some places expect or encourage anonymised applications and others restrict collecting the very data you would remove. Check local rules, and take advice from a qualified employment lawyer or data protection adviser where a legal point matters. Practices also differ in culture. In some countries photos, dates of birth and marital status are routinely included on CVs, and in others employers are discouraged from asking for them at all. If you hire across borders, apply the stricter standard, or the local rule for each role, and tell applicants what information is optional. Keep any data you do collect secure, and delete the link between reference numbers and names when you no longer need it. If you are unsure about a step, ask a qualified adviser before launching.

In Rezumz

In Rezumz, CVs are ranked against the must-haves you confirmed in the job workspace, with the supporting evidence shown, so the first review can start from skills. Duplicate detection across jobs helps you spot repeat applicants, and interview scorecards keep scores in one place. Rezumz does not currently redact or anonymise CVs for you, so any blind step in your process is done by your team before or around the workspace, and people make each shortlisting decision.