Hundreds of applicants. The same serious read.
To screen hundreds of applicants, agree the role requirements first, review everyone against them and investigate the unanswered questions that could change your shortlist. Cernor does that review and shows who deserves interview time and why.
No credit card required.
Filtering the obvious mismatches is not the same as screening the people who remain.
Check essential requirements first. Then examine what each person has actually owned and delivered. A keyword or job title alone cannot establish that experience.
A known requirement, answered yes or no
Work authorisation. A mandatory licence. A genuine non-negotiable. These have factual answers, and software has been able to check them for years.
Evidence, answered with strength and uncertainty
Whether the record demonstrates the capability, the ownership and the scope the role needs. This has no yes-or-no answer, which is why filtering harder does not solve it.
The Standard has to exist before the pile starts influencing it.
Agree what matters, what evidence would satisfy it and which requirements are essential. Cernor calls these agreed role requirements the Standard.
The shortlist should follow the Standard.It should not quietly rewrite it.
Volume should change throughput. It should not change the depth of the read.
The first promising resume is easy to read carefully. Nobody skims it. The harder problem is applying comparable rigour to the forty-seventh, and to the ninety-third, at the point where attention has run out but the pile has not.
The pressure is measurable. Across more than 6,000 companies and more than 640 million applications, Greenhouse records applications per job rising from 116 to 244 between 2022 and 2025, while annual applications per recruiter rose from 146 to 746 and recruiters per organization fell from 10.43 to 4.62. Those are platform measures rather than per-employer averages, so they do not describe any one team. The direction is the problem this page addresses: substantially more volume moving through a smaller reading capacity. Our research note The Decision Starts Where the Application Stops reports the full series and the scope of each figure.
No candidate is read against a bar that moved while they were waiting.
A resume can look right and still leave the important question unanswered.
Four states, so that missing information becomes neither certainty nor rejection. A claim the record contradicts and one it simply never makes are different findings, and they lead to different actions.
The record carries evidence that meets the criterion.
Some evidence, short of what the criterion asks for.
Nothing in the record speaks to it either way.
The record’s own account works against the claim.
Career strength and support for the important claims are separate judgments. See how Cernor separates them.
Screening 500 people isn’t finished when you have 500 scores.
A score for every applicant is the point at which the pile is measured, not the point at which it is decided. A low view may reflect genuinely thin evidence or a material question nobody has asked. Those findings deserve different next actions.
A material open question can become targeted Follow-Up built from that candidate’s unresolved evidence. Cernor evaluates what comes back against the same Standard, then strengthens, weakens, holds or leaves the recommendation unresolved with the reason attached.
Ask what would resolve the decision.Not everything you could ask.
See examples of candidate screening questions and what each one settles.
The workflow is simple. The hard part is keeping the judgment intact.
Use the same process for the first application and the three-hundredth. Agree the requirements before reviewing candidates, and record any later change.
- 1 Remove explicit mismatches
The gates your ATS can already answer.
- 2 Lock one Standard
What matters, what counts as proof, what is consequential.
- 3 Screen the plausible cohort
Every remaining candidate, against that version.
- 4 Keep uncertainty visible
Open stays Open rather than collapsing to yes or no.
- 5 Follow up where it matters
Only where an answer could change the decision.
- 6 Send a defensible shortlist
With the evidence and the Standard attached.
Automate the repetition. Keep the judgment inspectable.
These are principles rather than product claims. Whether any given tool meets them is a question to ask it directly.
- Repeat the same screening contract on every application
- Read the whole record, not the first page of it
- Preserve the evidence behind each judgment
- Surface what needs a human, rather than hiding it
- Treat keyword overlap as professional meaning
- Treat a job title as proof of capability
- Change the bar without saying that it changed
- Turn unknown into failed because unknown has no column
You do not need screening software because you have many resumes. You need it because several plausible candidates have to be compared against the same bar.
Cernor fits teams with more applications than time to assess them carefully. Start instantly with a job description and resumes. From $199 per live role, with up to 300 candidates and targeted Follow-Up included.
A spreadsheet can hold candidates.It cannot make the judgment.
The choice is obvious and the evidence is already sufficient
Manual review may be enoughThe requirement is a yes-or-no gate
Your ATS already does thisYou need proof someone can do the task
An assessment toolSeveral plausible candidates need comparing and the evidence is uneven
This is where Cernor becomes usefulCompare the five kinds of screening tool See how evidence-driven screening works
The scalable part is not rejecting faster. It is preserving the quality of the decision as the pile gets larger.
Common questions.
What is candidate screening?
Deciding which applicants move forward. In practice it is two jobs: removing applicants who do not meet an explicit requirement, and judging the ones who remain against what the role actually needs. The first is a filter. The second is a judgment, and it is the part that gets harder as volume grows.
How do you screen a large number of applicants?
Resolve the explicit requirements first, because they are cheap to answer and they shrink the pile. Then decide what the bar is and apply it to every remaining candidate the same way. At volume, the last candidate often gets a shallower read than the first.
What should happen before resume screening begins?
The criteria should exist. Which requirements matter, which are non-negotiable, and what counts as proof for each one. Deciding that after reading the first fifty applications means the applications have shaped the bar rather than the other way round.
Can AI screen hundreds of applicants?
It can perform the read on all of them, which is the part that does not scale by hand. Whether it should decide is a separate question. The useful version applies the same criteria to every application, shows who leads and why, and keeps the evidence behind each judgment so a person can check the calls that matter.
How do you avoid keyword matching when screening resumes?
Judge claims rather than terms. A resume that uses different words for the same work should not lose to one that happens to match the job description, and a resume that matches every keyword can still leave the important question unanswered. That means asking what the record supports, not what it contains.
Should missing resume information disqualify a candidate?
Not automatically. Missing evidence and contradicted evidence are different findings. If a criterion is unanswered and the answer could change the decision, ask a targeted question rather than rejecting the candidate or sending a generic questionnaire.
When should screening questions be used?
For requirements with factual yes-or-no answers, asked at application: authorisation, a licence, a genuine non-negotiable. They are a poor instrument for judging depth, because a candidate answering a question about their own experience is not the same as evidence of it.
Do I need candidate screening software if I already have an ATS?
Not necessarily. An ATS holds the applications and can answer explicit gates well. Screening software becomes useful when several plausible candidates have to be compared against the same bar and the evidence is uneven. Higher volume makes that consistency problem harder, but it exists in a small cohort too.
Give the last candidate the same serious read as the first.
Set the Standard once, run the cohort through it, and look at what comes back before you take our word for any of this.
No credit card required.