How do I screen hundreds of job applications quickly?

The instinct with 400 applications is to cut. Filter for a first-class degree. Search the CVs for the tool name. Drop anyone without five years. It feels like progress because the pile shrinks in minutes, and it is the single most expensive habit in Nigerian hiring.

Keywords measure how well someone wrote a CV, not how well they will do the job. The candidate who spent two hours tuning their CV to your job ad ranks above the one who has actually run the thing you need run, because the second person described it in their own words. Career changers, people from smaller companies, and anyone whose experience is real but oddly labelled get cut first. Meanwhile the CV-optimised applicant sails through to your interview, and you find out in week three.

There is a second problem specific to a market where a single Lagos job post can pull a thousand applicants: everyone is filtering the same way, so the same people are shortlisted everywhere and the same people are invisible everywhere. If your filter is a keyword, you are competing for a small, well-marketed slice of the market and never seeing the rest.

Speed is not the enemy here. Unstructured speed is. A fast process that evaluates everyone against the same defined thing is both quicker and better than a slow one where you skim CVs and trust your gut, because gut in a pile of 400 is mostly fatigue.

You cannot screen quickly against a job description that lists twelve responsibilities and no outcomes. You can screen very quickly against three to five things the hire must achieve in the first year.

Write them as results with a number and a date where you can: reduce delivery failures below one in twenty by June, take the monthly close from fifteen days to five, sign the first ten enterprise accounts. Then ask what a person must be able to do to produce that result. Those become your criteria, and there should be four to six of them, no more. Each one has to trace back to an outcome, or it does not belong on the list.

Three criteria are worth assessing on almost every role, at very different bars: whether the person sees the situation clearly, whether they act without being pushed, and whether they carry other people along. A dispatch supervisor and a finance lead both need all three. The role sets the bar; the score is read against that bar, not floating on its own.

This is the step people skip because it feels like paperwork before the real work. It is the only step that makes the rest fast. With a scorecard, screening is a mechanical comparison anyone on your team can do consistently. Without one, every application is a fresh argument with yourself, which is why 400 CVs takes two weeks and still ends in a coin toss.

Do not let the application be a CV upload. Put three or four written questions in front of it. Not competency-test questions, and not the ones every applicant has a prepared answer for. Ask who they are and why this role: what drives them, what they value, why they want this specific job at this specific company, what they would want to be true about their work in three years.

This is cheap for you, since nobody on your side reads it until it is scored, and it is cheap for a serious candidate, who has the answer already. It is expensive for the person mass-applying to sixty roles this week, and that is the point. The drop-off between people who upload a CV and people who write three honest paragraphs is the first real filter, and it filters for self-awareness and genuine interest rather than for CV-writing skill.

Keep it short enough to be honest. Four questions, a few sentences each, no essays. Long forms punish the busy candidate who is currently employed, which is usually the candidate you want. Tell people up front roughly how long it will take and what the stages are.

Generic answers are easy to spot and are themselves a signal, whether they came from a template or a chatbot. What you are looking for is specificity: a named situation, a real constraint, an opinion the person actually holds. Someone who cannot say why they want this job in their own words will not suddenly become clear in the interview.

Once someone clears the written stage, stop asking generic questions. Read what they claimed and interrogate that. They say they rebuilt the dispatch schedule: ask what it looked like before, who resisted, what broke first, what they would do differently. They say they left because the company changed direction: ask what specifically changed and who told them.

Questions generated from a person's own claims do two things at once. They are impossible to prepare for in the way a standard question list is, and they tell you whether the story holds together. A real story gets more detailed under pressure. An invented one gets vaguer.

Say early that finalists will be asked to connect you with a former manager. You do not have to call everyone, and often you will call nobody. Saying it changes what people write, because honesty gets cheaper than the alternative. It is one line in your application page and it works before you make a single call.

This is also where you catch the mismatch between what someone says and what they have done. A candidate who describes themselves as a systems builder but whose history is entirely execution against someone else's plan is not lying; they are just not the person your outcome needs. You only see that by reading their claims against their history, which is why both stages exist.

Screening degrades badly when it is done in the gaps between other work. The applicant reviewed at 9am and the one reviewed at 4pm are not being judged by the same person. Close the applications, then screen the whole set in dedicated blocks.

Two passes beat one long pass. The first pass is fast and only decides whether there is evidence worth reading. The second pass scores the survivors properly against the scorecard. Most of your pile dies in a first pass that takes well under a minute each.

1. Set a closing date and screen after it closes, not as applications trickle in.
2. Pass one: for each application, ask only whether there is any evidence against your top two criteria. Two minutes maximum, usually far less. Yes or no, no scores yet.
3. Pass two: score everyone who survived, criterion by criterion, on a 1-3 scale, writing the one piece of evidence behind each score.
4. Rank by total, then read the top fifteen properly, including the ones with an uneven profile - a 3, 3, 1 is often more interesting than a flat 2, 2, 2.
5. Take eight to ten to a short call or a demonstration task, and only then to a real interview.
6. Keep the scores. When the hire works or does not, you learn which criterion actually predicted it.

You do not need software to do this well, and if you hire twice a year you probably should not buy any. A Google Sheet is enough. Collect applications through a form so the answers land as rows instead of as attachments, then build the scorecard as columns.

The honest limit of the manual version is attention. Two passes over 400 applications is still several hours of concentrated reading, and the quality of hour five is not the quality of hour one. That is the specific thing a tool can fix: not judgment, but consistency across the whole pile.

Tezera is one way to run this shape of process. You define the outcomes and the criteria, candidates answer the written stage, and the AI reads every application against your scorecard with the evidence quoted underneath, so a human reads deeply only at the top of the ranking. The limitation worth stating plainly: it is hiring-only, it has no marketplace reach of its own the way Jobberman or MyJobMag do, so you still promote the role, and it cannot score anything until you have defined what the role must deliver. If you skip the scorecard, no tool saves you.

• One row per candidate, one column per criterion, plus a column for the evidence behind each score.
• Score 1-3 only. Wider scales invent precision you do not have and slow every decision down.
• Write the anchor for each score before you start: what a 1 looks like, what a 3 looks like.
• Two people score the first twenty independently, then compare. Where you disagree, the criterion is badly worded, not the candidate ambiguous.
• Hide names, schools and photos in pass one if you can. It costs nothing and it changes results.

The same question, phrased the other ways people ask it. Every one of them lands on this page.

• How do I shortlist 500 applicants without reading every CV?
• What is the fastest way to screen job applications in Nigeria?
• How do I filter hundreds of applications fairly?
• How can I reduce 300 applications to a shortlist of 10?

• →What is a hiring scorecard and how do I make one?
• →Tezera vs Google Forms for screening applications
• →Google Forms alternatives for hiring
• →Software engineer interview questions (Nigeria)

Everything else we have written like this sits on the answers index.