Pre-Interview Scorecards: Deciding Who to Talk to Before Any Call
Most recruitment problems that show up late in a hiring process started early, at the moment someone decided which applications deserved a conversation. When that decision rests on a vague sense of who 'seems good,' the shortlist reflects the reviewer's mood as much as the candidate's fit for the work. A pre-interview scorecard is a simple tool that moves that decision onto paper before any call is scheduled. It does not guarantee a good hire, and it does not replace a conversation. What it does is make the reasoning behind each screening choice visible, checkable and repeatable, so that a second reviewer can see why one applicant moved forward and another did not. This article walks through how to build one, what to put on it, and which questions to ask before you start scoring.
Start From the Brief, Not From the Applicants
A scorecard is only as useful as the scope it measures against. Before you write a single criterion, reread the task brief you published and extract the parts that describe the work itself: the deliverable, the inputs the worker will need, the conditions that mark the work as finished, and any constraints such as location, availability window or tools the person must be comfortable using. Each of these becomes a candidate criterion. If the brief says the worker will produce a set of cleaned contact records from a spreadsheet and confirm each entry against a source list, then experience with spreadsheet review and a habit of documenting exceptions are relevant. Experience with unrelated design software is not, however impressive it looks on a résumé.
This step also protects you from a common trap: writing criteria that describe the ideal candidate in general terms, such as 'excellent communicator' or 'self-starter.' Those phrases feel meaningful but give two reviewers no shared way to decide. Translate each one into something observable. 'Excellent communicator' becomes 'has described a past task in writing with a clear outcome and a named next step.' 'Self-starter' becomes 'has completed work without daily check-ins, which the application or references can show.' Where you cannot make a criterion observable, drop it or mark it as something to explore in conversation rather than something to screen on.
Choose Few Criteria and Define Each One Precisely
Three to five criteria usually works better than ten. Every extra criterion adds scoring effort and gives reviewers more room to drift. For each criterion, write three things: the name, a one-sentence definition, and a simple scale. A three-point scale is often enough: 'clearly shown,' 'partly shown,' and 'not shown in the application.' Avoid a ten-point scale, because it implies a precision that a short written application cannot support. Attach to each point on the scale a short example of what it looks like, so that a reviewer on a different day can calibrate against the same reference.
Include at least one criterion about evidence quality. A useful question is whether the applicant's claims are supported by something checkable, such as a described task, a named type of deliverable or a reference the employer can contact. Note that evidence is not the same as volume. A long list of past roles with no detail about what was delivered scores lower on evidence than a single well-described task. Keep that distinction written on the card so reviewers do not reward length by default.
Separate Screening Notes From Decisions
Record what you observed on the scorecard, then record the decision separately. An observation says 'application describes two completed data-cleaning tasks with stated error rates.' A decision says 'proceed to a short call.' Keeping these apart makes it possible to audit the process later. If a candidate is rejected and questions come up, you can point to the observations and see whether the decision followed from them. It also prevents a common failure in which a reviewer writes a conclusion like 'not a fit' without recording any of the reasoning behind it.
Where more than one person reviews applications, have each reviewer score independently before comparing notes. If scores diverge sharply on the same application, that is useful information. It usually means a criterion is ambiguous or that one reviewer has read something into the application that the others did not see. Resolve the divergence by returning to the written evidence, not by splitting the difference.
A Short Checklist Before You Use the Card
Before a scorecard goes into use, check it against these questions. Does every criterion trace back to a line in the published brief? Can a reviewer who has never seen the job score each criterion from the application alone? Is each point on the scale tied to an example? Are the criteria limited to what the work requires, rather than what would be pleasant to have? Are you screening on observable evidence rather than on tone, formatting or how closely an applicant's background resembles your own?
Also ask what the scorecard cannot tell you. It cannot confirm that a candidate is honest, available on the dates you need, or capable of the work under real conditions. It cannot guarantee that anyone will accept an offer, complete a task or stay engaged. Treat the scorecard as a way to decide who merits a conversation and what to test in that conversation, and keep the final judgement tied to direct evidence from the interview or a short work sample.
Keeping the Card Useful Over Time
A scorecard should be revised after each cycle. Look at the applications you advanced and the ones you did not, and ask whether the criteria separated them in ways you can explain. If every application scored 'clearly shown' on one criterion, that criterion is not doing any work and may be worth replacing. If the people you advanced turned out to be a poor match for reasons the card never measured, add a criterion that would have caught it. Date each version and keep the old ones, so that changes in screening practice can be traced back to the reasons behind them. A plain, consistent card used honestly will serve a small employer or agent operator far better than an elaborate framework that nobody reads twice.