How to screen hundreds of applicants with a small team

You have 600 applications waiting, interviews tomorrow and a shortlist due Friday. Reading CVs tonight feels like the only way through.

But another 120 applications arrive each day. The question is not just how fast you can review the 600. It is whether your review capacity can catch up with the queue.

The worked plan below counts new arrivals, candidate replies and time for notes. It shows how much help you need, what faster screening would change and what to tell the manager when the deadline does not fit.

First, find the work that is actually unfinished

Start with the oldest applications and read their latest notes. An application still marked “New” may already have an interview recommendation buried in the messages. Reviewing it again won't move it forward.

Separate applications that need screening from those that need a decision, a document or a booking:

Use your existing stages or notes to identify work already completed. A first review should leave a reason and a next step: recommend an interview, request important missing information or identify a requirement the applicant does not meet. Follow the team's approval process before acting on that recommendation.

For example: “Complaint-handling experience listed; investigation responsibility unclear. Ask for one case and the applicant's part in it.” That gives the next reviewer a question to resolve rather than another CV to reread.

A first review is finished when the requirements are checked, the reason is recorded and the next step is clear. The application may still need a reply, interview or decision. Track that remaining work separately.

How many CVs can one recruiter review in a day?

Divide the minutes genuinely available for CV review by the time needed for one complete review, including its note. Five hours at two minutes per CV gives you 150 completed first reviews. These timings are planning assumptions, not a speed target for recruiters.

First agree what Friday’s deadline means: an initial shortlist, or a completed first review of every application. Those are different workloads. Keep any promise to consider applications received by the advertised closing date.

To finish the first review of 600 waiting applications in five working days, while 120 more arrive each day, you need 240 reviews a day. The current queue and new arrivals both need space in the plan:

That is eight hours of first reviews in a five-hour block: a shortage of three hours each day, before replies and interviews. Take that gap to the manager. The estimate should describe the work, not become a demand to read every CV faster.

Why reviewing the original 600 is not the same as clearing the queue

At 150 reviews a day, you can finish the original 600 in four working days, working oldest first. But another 480 applications arrive during those days.

The whole queue gets smaller by only 30 a day: 150 reviewed minus 120 arriving. At those steady rates, it takes 20 working days to catch up completely.

Four days covers the original applicants. Twenty days covers the growing queue. Keep those deadlines separate, and do not let newer applications repeatedly push older ones back.

Calculate your catch-up date

Working days to clear the queue = applications waiting ÷ (daily reviews − daily new applications).

Daily reviews minus daily arrivals tells you how much the queue shrinks. Equal numbers mean it stays the same size. More arrivals than reviews mean it grows; there is no catch-up date at those rates.

This estimates when the unread queue will be cleared, not when every applicant will receive a final decision. It assumes steady arrivals and review capacity. If applications will stop arriving at an advertised closing date, calculate the work up to that date and the remaining backlog separately. Round up to a whole working day when booking the time.

What if applications close after five working days?

Keep the same starting figures, but stop new arrivals after day five. By then, another 600 applications have arrived and you've reviewed 750. That leaves 450 unread applications from a total of 1,200.

With no more arrivals, three further days at 150 reviews a day would clear them. You would finish on working day eight, rather than day twenty. Use the advertised closing date in your calculation; don't bring it forward without agreement.

How do candidate replies change the workload?

The first-review calculation does not include reading the answers to the questions you send today. Those replies become new work even though the applications have left the unread queue.

Now add 30 replies taking four minutes each. They use two hours of the same five-hour block. The remaining three hours allow 90 first reviews at the original two-minute pace.

With 120 new applications arriving, the unread queue grows by 30 that day. The earlier plan had it shrinking by 30. The difference is the time now needed for replies, not a change in the recruiter's speed.

Reserve time for replies, candidate messages and shortlist preparation before setting the first-review target. Moving a task to another list does not create time to do it.

Ask follow-up questions only when the answer could change your decision. Finding out whether an applicant investigated complaints may help you choose whom to interview. Asking which version of software they used adds little when you have already agreed to train them.

Keep unread CVs, replies awaiting review and manager decisions distinct in the existing hiring system. Also flag the people owed an update. Do not add those counts as though they were different candidates: one person can be awaiting the manager and an update at the same time.

Agree what counts as relevant experience

Before dividing the queue, have the manager and everyone helping with reviews compare a few applications. Resolve questions they would otherwise ask repeatedly. For a support vacancy, start with these descriptions:

“Investigated hotel complaints and followed up until they were resolved.” The setting is different, but the applicant describes investigating a customer's problem.

“Logged software-support tickets for the specialist team.” The setting is familiar, but the applicant hasn't described doing the investigation.

“Customer service duties.” You still need an example before you can compare the responsibility.

Ask which experience the job requires: investigating customer problems, working in software support, or both. A familiar industry doesn't establish the first skill, and a detailed complaint-handling example doesn't establish the second.

Ask the manager:

Would you interview someone who handled customer complaints in a hotel? Could we teach them our software?

Have reviewers make their recommendations independently, then compare the reasons with the manager. Save an agreed example of what counts and what still needs clarification. Use it across the queue; revisit earlier applications if the discussion changes an essential requirement.

Ask for one example when the CV doesn't explain the applicant's responsibility:

Your CV mentions handling customer complaints. Did you also investigate what had gone wrong? Please describe one case and your part in it, or let us know if a colleague handled that work. A few sentences are enough.

Can AI screening reduce the workload enough?

Compare the total time you spend reviewing applications with and without the tool. Reading its report, checking the CV and correcting mistakes all count. Add setup time and any extra administration too.

Try ordinary CVs as well as difficult ones. Check some the tool recommends declining, not only its favourites. Stop the timer when you have checked the application and written your recommendation—not when the AI report appears.

Don't time your manual review and then immediately reread the same CV with AI. You already know the application, which gives the second attempt a head start. Compare similar groups of applications and account for who reviews each group. Check disagreements separately from the timing.

Screened's resume screening checks a CV against your job requirements and explains the assessment. Test whether it reduces the time needed for a complete, checked review.

Our first interview can ask about experience the CV leaves unclear. That addresses a different task: count the time to review those answers and any human call still needed separately.

For the original five-day deadline, the target remains 240 completed first reviews a day. Compare the pace your trial achieves with the five hours available for that work:

All three scenarios reserve five hours for first reviews, including checks and notes. They are not Screened speed claims. Deduct time for replies or other tasks if those must share the same block.

At 90 seconds per review, you still need an extra hour of help each day. The tool reduces the gap; it does not eliminate it. Take both figures to the manager.

What should you tell the manager when the deadline does not fit?

Agree a shortlist review date with the manager and ask them to set aside interview times. Faster CV review won't get candidates to an interview while their applications are waiting for the manager's approval.

Ask the manager to resolve the decisions only they can make. For the support role:

Can you confirm by Thursday whether we can teach the software after hiring? Several applicants have investigated customer complaints in other industries. We need to decide whether that experience is enough to interview them.

Include the application examples that explain the question. Continue any follow-ups you're already authorised to send; don't create a new approval step for every candidate.

Until a trial demonstrates a faster pace, plan with the original two-minute review time. That leaves 90 reviews, or three hours of work, uncovered each day. Ask your lead:

Can we arrange three extra hours of CV-review help each day for the next five working days?

That would cover the 90 reviews a day we're short of. Without more capacity, catching up takes about 20 working days if applications keep arriving at the current rate.

We also need separate time for candidate replies and manager decisions.

Add time to brief anyone helping. Reduce the request only after the complete, checked review becomes faster; quicker report generation alone does not close the gap.

Update the backlog count at the end of each day

At the end of the day, account for everyone still waiting for a first review:

Waiting tomorrow = waiting today + new applications − completed first reviews − applications that no longer need review.

Keep the last two figures separate. A withdrawn application or an application closed because the vacancy was cancelled can leave the queue without a review. That is not screening work completed.

For example, start with 600 waiting, receive 120 new applications and complete 150 reviews. Ten other applicants withdraw before their first review. That leaves 560 awaiting first review. Report 150 reviewed and ten withdrawn, not 160 reviewed. A withdrawal after a completed review must not be subtracted a second time.

Count an application as still waiting if nobody has read its AI report or its CV is unreadable. If you have reviewed it and sent a question, remove it from the unread list but keep track of the reply. Different lists do not make that follow-up work disappear.

Recalculate when arrivals, review time or staff availability change. Check the oldest waiting applications too: a higher daily total can hide difficult cases that everyone keeps postponing.

Leave the day with a plan, not just a smaller number

Give your lead three things: the work still waiting, the hours available and the decision needed to close the gap. Ask for help, remove duplicate work or agree a later review deadline while there is still time to act. Changing the advertised closing date needs separate approval; quietly tightening rejection criteria is not a capacity plan.

Then send the authorised updates, assign the remaining decisions and protect the next review block. A smaller unread queue is not the whole result. Reserve time for the replies, interviews and promised updates it creates.