Quick answer: There is no magic number. Interview enough qualified people that you are choosing, not settling. For most SMB roles that means about three to five solid candidates after filtering, and that is a floor, not a finish line. CareerPlug’s 2025 Recruiting Metrics Report (2024 data from 60,000+ small businesses) found about 27% of interviewed candidates get hired, roughly 3.7 people interviewed per hire. Keep each call short, take notes on every one, and filter hard before anyone lands on your calendar.
I learned this the expensive way. At my previous company we interviewed too few people. A role opened, two or three candidates came through, one felt fine, and we hired. Everyone was busy. More interviews felt like a luxury.
Some of those hires could not do the job. Not bad people. Just not ready for the seat we put them in. It took months to get them up to speed, and the rest of the team carried the gap the whole time.
Here is what I got wrong. I thought the interviews were the cost. They were not. The bad hire was the cost. A few extra calls would have taken hours. The cleanup took months.
So this is for founders and hiring managers on small teams asking how many candidates to interview. In my opinion: stop hunting for a number and fix the process around it.
What do “interviews per hire” numbers actually count?
They count two different things, and mixing them up leads to bad plans. Some reports count people interviewed. Others count interview events, meaning every conversation with every interviewer.
CareerPlug counts people. About 3% of applicants get invited to interview. About 27% of interviewed candidates get hired (roughly 3.7 interviews per hire). Average applicants per hire sit around 180.
Ashby’s State of Startup Hiring report (published February 2026, mostly 2025 data) also counts people. For every hire, about 15 applicants receive an interview (about 18 in tech, 13 in business roles).
Gem’s 2025 Recruiting Benchmarks report counts events. Hiring teams ran about 20 interviews per hire, up from 14 in 2021. If one candidate meets four interviewers, that is four events and one person.
So when someone quotes “3 to 5 interviews” or “20 interviews,” ask which unit. For the question in the title, the unit that matters is people. How many qualified humans did you actually compare?
| Source | What it counts | Number | Year |
|---|---|---|---|
| CareerPlug | People interviewed per hire (SMB) | ~3.7 (27% interview to hire) | 2024 data / 2025 report |
| CareerPlug | Applicants invited / applicants per hire | 3% / ~180 | same |
| Ashby | People who receive an interview, per hire | ~15 (18 tech / 13 business) | 2025 data / 2026 report |
| Gem | Interview events per hire | 20 (was 14 in 2021) | 2025 report |
Is interviewing 3 to 5 candidates enough?
Sometimes, but only if all of them were genuinely qualified. Three to five is a sensible floor. It is not a box you tick.
The lazy version goes like this. Interview the first three who replied, pick the least worrying one, done. That is not choosing. That is sampling whoever happened to apply early. You can hit a tidy number and still make the same mistake we made.
CareerPlug’s ~3.7 and Ashby’s ~15 both describe real hiring. The gap is the point. The right count depends on your role, your market, and how well you filtered before the first call.
My test is simple. After your interviews, can you write down why your top pick beats your second pick, in a sentence each? If you cannot, you have not seen enough people. Or you did not take notes. Usually both.
Why is a bad hire more expensive than a few extra interviews?
Because the interview costs hours and the bad hire costs months.
In CareerBuilder’s 2017 Harris Poll survey, 74% of employers said they had hired the wrong person. The average cost of one bad hire was about $14,900. SHRM’s 2025 benchmarking reports put average cost per hire for nonexecutive roles at about $5,475 (executive about $35,879).
Different years and different surveys, so do not mash them into one precise formula. The direction is still obvious. Getting it wrong costs far more than one more short call with a solid candidate.
My own story fits. We saved a few hours of calls and paid for it with months of onboarding, rework, and patience.
Should you add more interview rounds instead?
No. More people early beats more rounds for each person.
Teams burned by a bad hire often respond by stacking panels. Four rounds becomes six. Gem’s event count shows where that leads: about 20 interviews per hire. SHRM’s 2025 benchmarks put screening and interviewing at about 8 to 9 days each in the posting to offer timeline. Every extra round stretches that, and good candidates take other offers while you schedule.
Google’s Rule of Four (People Analytics / re:Work, summarized by CNBC in 2019) is about rounds per candidate, not how many candidates to see. Four interviews predicted the hire decision with about 86% confidence in their data, and four interviewers matched larger panels most of the time. Extra rounds added little. Go wide early, then save depth for the few who earn it.
How do you interview more people without losing your week?
Keep the calls short and take notes on every one. My rule now is to interview as many solid people as I can, as long as each call stays tight.
- Filter before you dial. Only people who meet the must-haves get a call.
- Run a short first call. A tight phone screen can check salary, logistics, and one proof question without turning into an hour of small talk.
- Ask everyone the same core questions. Same questions make people comparable.
- Write notes right after each call. Pass, fail, or unclear, with one line of evidence.
- Score the real interview properly. Use an interview scorecard so every interviewer rates the same things.
In Canvider, I keep those notes in Internal Comments on the candidate so the team sees them in one place. Schedule Interview can send the invite with Inform invitees and a .ics. There is no automated reminder feature to lean on. The calendar hold and a human follow up still matter.
Short calls with notes are cheap. Long, loose calls without notes are expensive, and you forget half of them by Friday.
Where should the filtering happen?
Before the interview, not during it.
CareerPlug’s 3% figure tells you most applicants were never going to get a call. The job is not to interview more of the pile. It is to find the solid people in the pile faster.
If you are drowning in applications, start with too many applications and not enough real candidates. Knock out hard requirements first, then rank what is left.
That is where I use Canvider. Criteria Match checks each CV against up to five must-haves you define. AI Score compares the CV against the job description and puts it in a band (excellent, good, fair, or weak), with the score rounded to the nearest 10. Bulk Candidate Analysis can run across a whole role.
Be clear about what that is. It helps you decide whose application to read first. It is not a rejection machine. Never reject on the score alone. A human decides. A thin job description also makes scores noisy, so lock the JD and skills before you analyze.
When have you interviewed enough candidates?
You have interviewed enough when you have two to four finalists you would genuinely be happy to hire, and your notes let you compare them fairly.
If you have that, stop. Put the finalists side by side with the team. Canvider’s Decision Helper is built for that moment: comparing two to four finalists, not screening hundreds. The top pick is an input, not the hire decision.
If you do not have that, do not settle for the best of a thin batch. That is the mistake that cost us months. Widen sourcing and run a few more short calls. A week of extra interviews is cheaper than a quarter of cleanup.
Canvider will not hand you a magic interview number. The product has no recommended N. What it does is keep the filtering, the notes, and the final comparison in one place, so seeing more solid people does not eat your week.
Frequently asked questions
How many candidates should you interview before hiring?
Interview enough qualified people to compare, often three to five after hard filtering. CareerPlug's 2025 report (2024 SMB data) shows about 27% of interviewed candidates get hired, roughly 3.7 people interviewed per hire. Treat that as a floor, not a finish line.
What is a good interview-to-hire ratio?
For SMB hiring, CareerPlug's interview-to-hire rate of about 27% works out to roughly 3.7 people interviewed per hire. Gem's 2025 report counts interview events, not people, and lands near 20 events per hire. Always check whether a source means people or conversations.
Is interviewing only 3 candidates enough?
Only if all three were genuinely qualified and you can still explain why your top pick beats the second. Three weak early applicants is not a shortlist. If nobody clearly meets your bar, interview more people instead of hiring the best of a thin batch.
Does interviewing more candidates improve hire quality?
More short, noted interviews of solid people usually help. Endless panels for the same person usually do not. Google's Rule of Four is about rounds per candidate, not shortlist size, and it showed diminishing returns after about four interviews per person.