Shortlisting: getting from fifty applications to five fairly

Shortlisting is the step between receiving applications and deciding who to interview. It gets a fraction of the attention that interviewing does, and it determines far more: an interview can only choose among the people the shortlist produced. If the filter is arbitrary, everything downstream is arbitrary too, however carefully the interviews are run.

It is also the stage where bias operates most freely, because judgements are quick, unrecorded, and made from documents that carry a great deal of irrelevant information — names, addresses, schools, photographs, dates that imply age.

What a shortlist needs to be defensible

• Written criteria, agreed before applications are opened — the essential requirements from the job description, nothing added later.
• A scoring scale per criterion, even a simple one: does not meet, meets, exceeds.
• The same order of assessment for every application, so nobody is judged against whoever came before.
• A record of the score per criterion, not just the outcome. A shortlist without reasoning cannot be reviewed or defended.
• A threshold decided in advance — everyone meeting all essentials gets an interview — rather than a target number that shifts to fit the day.
• At least two assessors on borderline cases, scoring independently.
• A note of anything set aside, such as a missing formal qualification balanced by demonstrated capability, so exceptions are visible rather than invisible.
• A retention rule for the scores, since they are the record if a decision is ever questioned.

Strip what you do not need before scoring. Removing names, addresses, dates of birth and educational institutions from what the assessor sees is not complicated with a form-based application, and it removes the strongest triggers for the associations people make without noticing. Nobody scores fairly by deciding to.

Doing it in an hour

1. Build the scoring sheet before the advert closes: one row per applicant, one column per criterion, a total and a notes column.
2. Collect applications through a form with structured questions rather than CVs alone, so the same information exists for everyone in the same place.
3. Score one criterion across all applicants at a time, rather than one applicant across all criteria. It is faster and it anchors far less.
4. Do a first pass on essentials only. Anyone missing an essential requirement is out, and that pass usually removes half.
5. Score the remainder on the desirable criteria to rank, not to exclude.
6. Review the borderline group with a second assessor, scoring independently before discussing.
7. Check the shape of the shortlist before sending invitations: if everyone shortlisted resembles everyone already in the team, look again at the criteria rather than at the candidates.
8. Keep the sheet. It is the answer to "why was I not shortlisted", and it is what makes the next round faster.

When the numbers are wrong

Two failure modes need different responses. Fifty applications where five meet the essentials means the advert reached the wrong people, and the fix is sourcing rather than a stricter filter. Fifty applications where forty meet the essentials means the essentials are not essential — the criteria are describing a general professional rather than this job, and tightening them after seeing the applicants is exactly the thing that lets preference in through the back door. Better to interview more people this time and write sharper criteria next time.

Ettex Sheets is the natural home for the scoring: import an existing XLSX with formulas intact, dropdown validation so scores stay on the same scale, conditional formatting to highlight anyone meeting every essential, formulas for weighted totals, sorting and filtering to rank without re-sorting by hand, and cell comments with threads so a second assessor can question a score against the cell it sits in. Version history means the sheet as it stood when you decided is recoverable. The applications themselves come from Ettex Forms with structured questions and CV uploads landing timestamped in one inbox, the criteria come from the job description in Ettex Docs, and the candidates you take forward can be tracked in Ettex CRM.

Said plainly: Ettex does not screen candidates for you. There is no CV parsing, no automatic matching or ranking, no anonymisation feature that hides fields from an assessor, and no diversity reporting. Building the sheet, choosing what to show the assessor, and doing the scoring are all manual. That is workable for tens of applications and not for thousands.

Where shortlisting goes wrong

• Criteria written after reading the applications, which reverses the whole point of having them.
• Reading CVs in the order received, so early applicants set the standard everyone else is judged against.
• A target number of interviews rather than a threshold, so the bar moves to fit the diary.
• One assessor deciding alone with no record of why.
• Names, schools and dates visible during scoring, which is where most unconscious filtering happens.
• "Culture fit" used as a criterion, which in practice usually means similarity.
• Gaps in employment treated as a negative without asking, which filters out carers and people who have been ill.
• Scores discarded afterwards, leaving nothing to answer a challenge or improve the next round.

Two documents make the decision defensible rather than instinctive. A job description template fixes what is actually being assessed, and a structured interview keeps the questions the same across candidates so the comparison means something.

Shortlisting decides what the interview can possibly produce. Write the criteria first, score one criterion at a time, hide what you do not need, keep the sheet — and fix sourcing rather than the bar when the numbers disappoint.