Redact CVs for blind screening

Blind screening works, and the reason it is rare is that somebody has to do it. Forty CVs, forty different layouts, and every one of them puts the name somewhere else.

Open the tool Free, in your browser. Nothing is uploaded.

Before redaction. — A page before and after: the name, the contact block and the photograph are gone.

A page before and after: the name, the contact block and the photograph are gone.

Why it is usually done by hand

A CV is a design document. The name is 24-point type in a header, or white on a coloured band, or inside a graphic exported from a design tool with no text in it at all. The photograph is a picture. The address is three lines in a sidebar. A search-and-replace over a folder of PDFs finds some of that and quietly leaves the rest.

And the point of the exercise is defeated by one miss. A panel that sees one name has seen a name.

What to take out, and what the panel keeps

Type the candidate’s name once and it goes from the header, the footer, the sidebar and the lettering of any graphic. Tick the detectors for email addresses, phone numbers and street addresses. Draw a box round the photograph.

What stays is the work: the employers, the dates, the projects, the results. You decide whether the university and the school stay too – they are the usual argument in blind screening, and this takes no position on it. Type them if you want them gone.

A consistent code, so the panel can still refer to somebody

Turn on labels and each removed name becomes a stable placeholder. The panel can say "[P3] has the stronger delivery record" and the person running the process can map that back afterwards. Without it, forty anonymous CVs are forty documents nobody can discuss.

What blinding cannot do, and should not pretend to

Removing a name removes one signal. A CV still carries a career shape, a set of employers, the language somebody writes in and often a gap that has a reason behind it. Blind screening narrows the gap between candidates; it does not close it, and a process that believes it has been made objective is a process that has stopped watching itself.

It is also the easiest half. The hard half is the scoring rubric agreed before anybody reads anything, and no tool supplies that.

What this does is make the easy half cheap enough to actually do. The usual reason a panel sees names is not principle, it is that somebody had forty PDFs and an afternoon.

How to do it, in four steps

Open a CV. Everything happens in the tab.
Type the name, and tick the detectors for email, phone and address.
Draw a box round the photograph and any personal logo or crest.
Turn on labels so the panel can refer to candidates, then Redact and export.

Who it is for

Hiring managers and recruiters running a structured process, university admissions, grant panels, and anybody who has been asked to score work without knowing whose it is.

Open the tool

Questions

Do the CVs get uploaded?

No – which matters more here than almost anywhere, because a CV is somebody’s personal data and you did not ask their permission to send it to a third party. Nothing leaves the machine.

Does it find the name in a designed header?

Yes. Pages are read in the tab, so lettering that exists only as artwork is searched too, and words the reading cannot make out are looked for by shape.

Can I do forty of them?

One at a time. The terms stay as you go from one file to the next, but each CV names a different person, so each needs its own name typed – which is the part that cannot be automated without guessing who somebody is.

How do we unblind at the end?

Keep the originals. The labels are consistent within a document, and whoever runs the process holds the mapping – not this tool, which keeps nothing.

Also

Open the tool