How to screen candidates fast without lowering the bar
Speed in screening is not the enemy. Unstructured speed is. A reviewer going fast with clear criteria is more consistent than one going slowly with none. This page is the method, whether or not you use any software for it.
Step one: separate eliminating from scoring
Before reading anything, write down the two or three requirements that genuinely disqualify. Legal right to work, a mandatory certification, a language the job is impossible without. Be honest: most things people list as mandatory are preferences.
Everything else is a scoring criterion with a weight. This single split is what stops a strong candidate being cut for missing something optional, and it is what makes fast reading safe.
Step two: read for evidence, not impression
For each scoring criterion, look for something in the CV that would convince a sceptical colleague. Three years listed under a job title is a claim. A described outcome, a scale, a tool, a responsibility is evidence.
Write the evidence down next to the score, even in three words. If you cannot write it, you did not have it, and you were about to score on impression.
Step three: handle doubts as doubts
Gaps, career changes and unfamiliar employers are ambiguity, not negative evidence. The fast and fair move is to mark them as questions for the phone screen rather than silently lowering the score.
This is where most speed-related damage happens: ambiguity gets read as risk, and everyone with a non-linear career quietly disappears from the shortlist.
Step four: review the borderline group, not the top
The obvious top candidates need little of your attention, and neither do clear misses. Your judgement is worth most in the middle band, where a five minute second look changes the outcome.
If you are using a tool that ranks for you, spend your review time exactly there. That is the correct division of labour between automation and a person.
What this looks like automated
Parthumanally implements this method directly: mandatory criteria that eliminate, weighted criteria that score, evidence quoted next to each score, doubts flagged for the interview, and a human confirmation step before anyone is rejected.
The method is worth using even on paper. The tool mainly removes the part where you do it two hundred times by hand.
Frequently asked questions
How fast is fast enough?
With clear criteria, a structured screen of a CV takes about a minute of human attention, and less on applications that fail a mandatory requirement. What matters more than the clock is that the same criteria are applied to the first and the last application.
Should I screen blind, without names?
Removing names and photos reduces certain biases and is worth doing when your tooling supports it. It does not remove signals like university, employer prestige or the way a CV is written, so treat it as one measure among several rather than a solution.
What do I do with the candidates I reject?
Tell them, quickly, and keep the reasoning you recorded. Under the GDPR they can ask what data you hold, and under the EU AI Act you should be able to explain how an automated evaluation contributed to the decision. Recording the reasoning as you go is much cheaper than reconstructing it later.
See the method running
The demo shows the eliminating and scoring split on a real job structure, with fictional candidates. No account needed.
See a demo