Back to Blog
Recruitment Tools

How to Generate Structured CSV Shortlists From Raw CVs

CVSense® InsightsCircle
5 views
0 comments
Share:
How to Generate Structured CSV Shortlists From Raw CVs

Clients do not want twelve PDF attachments, they want one comparable table. This workflow guide covers turning raw CVs into a structured CSV shortlist with evidence-based rationales attached to every row.

A client does not want twelve PDF attachments. They want one table they can sort, filter, forward to a hiring manager and make a decision from. Working out how to generate structured CSV shortlists from raw CVs is therefore less about export functionality and more about deciding what belongs in the columns, because a spreadsheet of names and phone numbers is not a shortlist, it is a contact list.


This workflow guide covers the column design that makes a shortlist actually useful, how to get there without manual retyping, and the data protection steps that a spreadsheet of candidate data obliges you to take.

Why the Spreadsheet Still Wins

Every recruitment platform has a candidate view, and clients keep asking for a spreadsheet anyway. It is worth understanding why, because the reason shapes the design.


A spreadsheet is comparable, portable and annotatable. A hiring manager can sort by whichever column they care about, add their own notes, forward it to a colleague who has no account on your system, and read it on a train. No platform view competes with that on convenience, and asking a client to log in to review a shortlist adds friction at exactly the wrong moment.


The failure is not the format. It is that most exports dump contact details and a CV link, which leaves the client doing the comparison work you were engaged to do.

Column Design That Earns Its Place

A useful shortlist export has four groups of columns.

Identity and Logistics

Kept deliberately minimal: candidate reference, name, current role and employer, location and work arrangement, notice period, availability, and rate or salary expectation. Contact details are often better withheld from a first-pass shortlist, both because clients rarely need them at comparison stage and because it limits how far personal data travels.

One Column per Must-Have Requirement

This is the heart of it, and the part most exports omit. One column per genuine must-have, with a consistent, sortable value. Three values work better than a numeric score:

  • Evidenced: the requirement is supported by a described application of the capability.
  • Partial: mentioned or contextual, without depth.
  • Not evidenced: nothing in the document addresses it.

A client can sort and filter on that immediately. A composite score out of one hundred invites the question "what does 73 mean?", and you will not have a satisfying answer.

The Rationale Column, per Requirement

This is what distinguishes a shortlist from a list. For each must-have, a short rationale quoting or closely paraphrasing the actual evidence.


Not "strong Python experience" but "built and maintained the ETL pipeline processing nightly reconciliation, about 2m records, in Python; rewrote it when memory limits were hit". The client can now evaluate your reasoning rather than taking your word for it, which is the entire basis of the relationship.


Where a requirement is not evidenced, say so plainly and say what you propose: "no evidence of team leadership in the CV; candidate confirmed at screening call that they led a team of four for eighteen months". That sentence demonstrates you did the work.

Assessment Metadata

Who assessed the candidate, when, against which version of the criteria, and any override with its reason. Clients rarely read these columns and their presence changes how the document is received: it shows a process rather than an opinion. It is also the record you will want if a decision is ever questioned.

The Workflow

  1. Agree the must-haves with the client first, and keep the list short. Three to five. This conversation is also where you discover that two of the twelve requirements on the brief were aspirational.
  2. Define the schema and version it. The columns are the criteria. Fix them before assessing, because criteria that evolve mid-process produce a shortlist you cannot defend as consistent.
  3. Batch intake the CVs. Upload the full set at once rather than handling records individually.
  4. Extract evidence per criterion, with the source passage retained against each finding.
  5. Review by criterion, not by candidate, confirming or overriding each proposal with a reason. This is faster and considerably more consistent than reading each CV end to end.
  6. Resolve the gaps where a strong candidate simply wrote a thin CV, with a short call, and record what you learned in the rationale.
  7. Export to CSV with the rationale columns populated from the retained evidence, not retyped.
  8. Sense-check the export before it leaves: read three rows as the client will read them and confirm the rationales say something.

Step seven is where most of the time is saved. If rationales are retyped from a system into a spreadsheet, a twelve-candidate shortlist costs an hour of copy-paste and introduces transcription errors. If the evidence is structured when captured, the export is instant and accurate.

The Data Protection Part, Which Is Not Optional

A CSV of candidate assessments is a file of personal data, including your opinions about identifiable people, and it is about to be emailed. Six things follow.

  • Minimise the columns. Export what the client needs for this decision. Date of birth, full address, nationality and anything from an equality monitoring form should not be in the file.
  • Transfer it securely. A shared link with access controls beats an email attachment, which gets forwarded, downloaded and stored in places nobody tracks.
  • Be clear on roles. Where you assess candidates for your own placement purposes you are a controller. Where you process to a client's instruction the position needs to be set out in your agreement rather than assumed. Sharing candidate data with a client is a disclosure to a third party, and your privacy information should tell candidates it happens.
  • Write rationales you would be content for the candidate to read. They can request their personal data, and your assessment notes are their personal data. This is a useful constraint: it produces better-evidenced, more professional rationales.
  • Version and date the file. Multiple spreadsheet copies in circulation is how stale assessments get acted on, and how your retention position becomes untrue.
  • Set a retention position for the export, shorter than your main record, and actually enforce it.

The rationale point deserves emphasis. Knowing a candidate may read the reasoning tends to eliminate the throwaway remarks that cause the real problems, and improves the quality of the shortlist as a document.

Frequently Asked Questions

What Columns Should a Client-Ready Shortlist Contain?

Minimal identity and logistics; one column per must-have requirement with a consistent evidenced, partial or not-evidenced value; a short rationale column per requirement quoting the actual evidence; and assessment metadata showing who assessed it, when and against which criteria version. The rationale columns are what make it a shortlist rather than a contact list.

Should a Shortlist Include a Single Overall Score?

Generally not. A composite number hides the reasoning and invites questions you cannot answer precisely. Per-requirement values are sortable, transparent and far easier to defend. If a client insists on ranking, rank on the must-haves they identified as most important and say that is what you have done.

How Do You Avoid Retyping Rationales into a Spreadsheet?

Capture the evidence in a structured form at assessment time, with the source passage retained against each criterion, so the export composes rationale columns from data you already hold. Retyping is only necessary when the assessment was recorded as free-text notes.

Is Emailing a Spreadsheet of Candidate Data a Problem?

It is a disclosure of personal data to a third party and should be handled deliberately: minimised columns, secure transfer rather than an open attachment, clarity about controller and processor roles, candidate-facing transparency that data is shared with clients, and a retention position for the copy. None of it is difficult; all of it is easier before the file is sent.

Should Contact Details Be Included?

Usually not at comparison stage. Clients are deciding who to interview, not contacting candidates directly, and withholding contact details limits how far personal data travels while keeping the process running through you. Release them when they are needed.

The Point of the Exercise

A shortlist is an argument, and the columns are the evidence for it. A client who can see why each candidate is on the list and what remains uncertain about them can act immediately. A client who receives twelve names and twelve CV links has been handed the work back.


CVSense captures evidence against each criterion as structured data with the source passage retained, which is what allows a client-ready CSV with populated rationale columns to be produced directly rather than assembled by hand. If your shortlists currently cost an hour of copy-paste before they can go out, that hour is the one this removes.


Sources

Information Commissioner's Office. Employment Information and Guidance.
https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/employment/

Information Commissioner's Office. Right of Access.
https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/individual-rights/individual-rights/right-of-access/

Chartered Institute of Personnel and Development.
https://www.cipd.org/uk/


InsightCircle

Comments

Start the discussion

Be the first to comment on this article.

Loading comments…
Powered by CVSense

Follow @CVSense on LinkedIn

Get recruitment best practices, career guides, and insights that can help you succeed.

Tags
#csvexport#clientshortlist#recruitmentreporting#structureddata#agencyworkflow
CI

About CVSense® InsightsCircle

At CVSense, we have built technologies that help you present your skills most compellingly, in addition to helping recruiters ensure that they get the right candidates.

Supercharge Your Job Search with CVSense

Apply to jobs faster and smarter with CVSense's browser extension. Autofill applications, get AI-powered recommendations, and track your progress - all from your browser.

Start Landing Job Interviews

More Articles You Might Like