How many candidates should you interview in the first round? The number, worked back from your funnel

Interview enough candidates in the first round to leave two or three finalists who clear every must-have; with published pass rates, that comes to roughly 5 to 15 first-round interviews per opening. Your own number comes from your funnel: openings ÷ (share of first-round candidates who reach the final × share of finalists who get offers × offer acceptance rate). Using Gem's 2024 benchmark rates for final rounds and offers, and assuming 35% to 60% of first-round candidates reach the final, the arithmetic runs from about 5 per hire for finance roles to about 15 for engineering roles.

The ratio behind this, measured on your past hires, is the interview-to-hire ratio; the stage-by-stage rates come from pipeline conversion rates. This page turns them into a number to plan the first round with, and a rule for when to stop booking.

How many candidates should you interview in the first round?

Work backward from the hire, one stage at a time:

Offers needed              = openings ÷ offer acceptance rate
Finalists needed           = offers needed ÷ final-round offer rate
First-round interviews     = finalists needed ÷ first round-to-final pass rate

The pass rates come from your own history if you have it. If you do not, published rates for the last two stages give a starting point, and the first-round pass rate is the one assumption you have to make yourself.

What does published data suggest?

Gem's 2025 Recruiting Benchmarks Report covers January 2021 to December 2024 across employers using its software, with data on over 140 million applicants and 1.3 million hires. For 2024 it reports onsite-to-offer rates of 26% for engineering and 36% for finance candidates, offer acceptance of 76% for engineering and 88% for finance, and, for inbound applicants overall, 35% of those brought onsite receiving offers, with an aggregate offer-to-hire rate of 84%. Turning those into finalists per hire, and then into first-round interviews under three assumed first-round pass rates:

The pass-rate columns are our assumptions, not Gem's data, so treat the table as arithmetic, not a benchmark. Gem's own stage before onsite includes recruiter screens as well as hiring-team interviews: for inbound applicants it reports 20% of that stage reaching onsite in 2024, which implies about 1 ÷ (0.20 × 0.35 × 0.84) = 17 candidates entering the screening stages per inbound hire.

One number in the report is easy to misread: hiring teams conducted 20 interviews per hire in 2024, up from 14 in 2021. That counts interviews, including every round each candidate had, not candidates. A loop of four sessions for one finalist is four interviews.

Worked example: your own funnel

If your history is your interview-to-hire ratio, the shortcut is openings × ratio: two openings at 9:1 means about 18 first-round interviews. The hours behind that number, and how a tighter screen reduces it, are worked through in interview-to-hire ratio.

How many phone screens does that take?

Take the same formula one stage further back. Continuing the example, if 40% of screened candidates pass the recruiter screen, 18 first-round interviews need 18 ÷ 0.40 = 45 screens. At 30 minutes each plus 10 minutes of notes, that is 45 × 40 minutes = 30 hours of recruiter time for two hires, before any sourcing. Two checks keep the plan honest:

• Recalibrate after the first few interviews. If the hiring manager passes one of the first five first-round candidates when you planned on 45%, the screen and the brief disagree. Stop and fix the brief with the intake questions rather than booking more of the same.
• Watch the screen's own pass rate. A screen that passes nearly everyone moves the work, and the cost, onto the hiring manager's calendar.

How many finalists do you need?

Two or three who clear every must-have, as a minimum. With one, there is no comparison, and the decision is really whether to keep searching. The US Office of Personnel Management's structured interview guide lists "pressure to hire" among common interviewing mistakes: when interviewers feel they must decide quickly, they tend to decide on limited information or a small number of candidate interviews.

The upper limit is cost. A final round of four sessions can take seven or so interviewer hours per candidate once write-ups are counted (see the worked loop in how to design an interview loop), so five finalists for one opening is a week of someone's time. If you regularly have more finalists than that, the earlier rounds are not filtering; the interview-to-offer ratio shows which managers that applies to.

When should you stop booking first-round interviews?

• Stop when the pipeline covers the finalists you need. Book interviews until the number in progress, multiplied by your first-round pass rate, reaches your finalist target.
• Use a batch window where you can. Interviewing the first round within one or two weeks makes candidates easier to compare. OPM's guide warns that the order in which candidates are interviewed can affect their ratings, which rolling interviews spread over months make worse.
• Keep an exception path. A strong candidate with another offer should not wait for the batch to finish; move them forward and keep comparing.
• Do not leave early candidates waiting. Tell first-round candidates when they will hear back; long gaps show up as withdrawals, which candidate drop-off rate measures.
• Reopen sourcing, not the bar, if the first round produces fewer qualified finalists than planned.

Does the 37% rule tell you how many candidates to interview?

The "37% rule" comes from a well-known puzzle in mathematics, often called the secretary problem: candidates arrive one at a time, you must accept or reject each on the spot, you can never go back, and you win only if you pick the single best. The best strategy is to reject roughly the first 37% (1 in e) and then take the next candidate better than everyone seen so far. Hiring breaks almost every assumption: you can compare a slate, call back an earlier candidate, and you need someone who clears the bar, not the best person in the world. Use the funnel instead.

How does the number change by kind of role?

The proportion of applicants who end up hired is the selection ratio, and why a low one raises the payoff of a good assessment is explained in selection ratio in hiring.

Common mistakes