PDF work for HR teams

Blind shortlisting means removing the name, contact details, addresses and often the dates that reveal age before a panel sees the document. Doing that by drawing boxes in an annotator does not work — the text stays in the file and anyone can select it out.

Scanning the document for personal identifiers first is faster than reading forty CVs for them, and it finds the ones people forget: an email address in a footer, a phone number inside a hyperlink, a photograph's metadata.

A CV exported from a word processor usually carries the author name, the organisation the software was registered to, and sometimes the full history of when it was edited. A candidate's document can therefore reveal their current employer through nothing that appears on the page.

Strip metadata from anything being circulated internally, and be aware that it is also a route by which your own organisation's information leaves in documents you send out.

A joiner's file arrives as a dozen separate documents — offer letter, signed contract, right-to-work evidence, bank details, policy acknowledgements. Merging them into one document in a consistent order makes the file navigable, and page numbers make it referenceable in a later dispute.

Keep the sensitive evidence separable. Right-to-work documents and bank details are the parts most likely to need sharing with a third party in isolation, and extracting a page from a bundle is easier than un-merging one.

A contract completed in form fields can render blank in a phone mail app or a browser preview, which is how a signed document arrives looking empty. Flattening writes the answers into the page so every viewer shows the same thing, and stops them being edited afterwards.

A password stops a personnel file being opened by whoever finds the laptop, but note that PDF permissions — restricting printing or copying — are advisory and many viewers ignore them. A password to open is enforced; a permission is a request.

Under the DPDPA, the GDPR and equivalent regimes, sending a CV to an online converter is a transfer of personal data to a processor, and one you would have to be able to justify and document.

Processing in the browser removes the question rather than answering it: no third party received the data, so there is no processor to assess, no transfer to record and no retention policy to check. For routine work that is a considerably cheaper position than doing the paperwork.

A request for everything an organisation holds about a person turns into a PDF exercise very quickly, and it has a specific difficulty: the file must contain everything about the requester and nothing about anybody else. Third-party names in an email thread, another candidate's details on a shared shortlist, a colleague's comments in a review — all of it has to come out.

Scanning the bundle for personal identifiers first is considerably faster and more thorough than reading for them, and it finds what a reader skims past. Then verify by extracting the text of the finished file and searching it for the names that should not be there.

Page numbering across the bundle matters here too, because a response is usually accompanied by a schedule that refers to it, and the schedule has to be able to point at something.