The Published Hiring Statistic Reference, One Row per Interview Number

Recruiters write hiring benchmarks like "44 days to hire", "5 rounds", or "1 in 4 interviews end in an offer" for other recruiters, and each one counts something different. This reference is for job seekers who want to use those numbers to judge their own interview loops and timeline. Each row covers one published figure from the candidate's side: where its clock starts and stops, whose interviews it counts, which population it came from, how it misleads you, and how to convert it into something you can compare with your own search or else reject.

The five fields every hiring statistic carries

Every published hiring number is defined by five fields: clock start, clock stop, unit, population, and statistic type. If you can't name all five, you can't compare the number with your own search.

• Clock start. The event that starts the count. It might be requisition approval, your application, or your first interview. Get this wrong and you misread a figure by weeks.
• Clock stop. The event that ends the count, which is either offer accepted or offer-or-rejection. If the clock stops at a rejection, the figure includes candidates who were turned down.
• Unit. Days, people interviewed per hire, rounds the hired candidate sat, or individual sessions. A count of people and a count of sessions can differ by a factor of two or more for the same hire.
• Population. Whose hiring was measured: small businesses, campus programs, startups, enterprises, or a broad sample of new hires. Offer rates swing by a factor of two across these groups.
• Statistic type. Median or mean. A median is the middle value. A mean is the arithmetic average, and a handful of very long searches pull it up.

There's a sixth thing to check, the collection method, but it isn't a field you can convert. Glassdoor's figures are self-reported by job seekers. SHRM's come from a survey of HR teams. SmartRecruiters, Gem, and Ashby use timestamps from their applicant tracking systems (ATS), the software employers use to move candidates through stages. Nobody has publicly established which way each method biases the result, so note the method and don't adjust for it.

Which clock is it: time to fill vs time to hire for job seekers

Time to fill starts when the employer approves the requisition. Time to hire starts when the candidate applies. Both stop at offer acceptance. Glassdoor and NACE use shorter clocks that stop at either an offer or a rejection, so their figures aren't measuring the same span.

"39 days" or "44 days" (SHRM time to fill)

• Counts: Calendar days from requisition approval to offer acceptance, with weekends and holidays included. It comes from a survey of HR teams. The current brief gives medians of 39 days for nonexecutive roles and 45 for executive roles. An older SHRM Talent Access report gave a nonexecutive median of 44.
• Misleads by: Including the days before the posting existed for you. People read "44 days" as interview time, but it isn't.
• Convert: In an older SHRM report summarised by AIHR, time to fill was 36 days and the gap between screening candidates and offer acceptance was 24, which implies about 12 days before screening (derived). Subtract about 12 days to get a rough screening-to-acceptance figure, so 39 becomes roughly 27. Since that gap comes from a different, older sample, treat the result as approximate.

"30", "40", or "41 days" (ATS time to hire)

• Counts: Ashby's medians run from application to acceptance: 30 days for business roles and 40 for technical roles. Gem's mean runs from pipeline entry to acceptance and went from 33 days to 41 over its reported period.
• Misleads by: Mixing a median (Ashby) with a mean (Gem). Gem's pipeline entry can also come before a formal application for sourced candidates.
• Convert: These are the closest match to a candidate's own clock. Measure from your application date to the day you accept. If you haven't accepted yet, measure to today.

"23.8 days" (Glassdoor interview duration)

• Counts: Days from application submission to an offer or a rejection letter, self-reported in 83,921 interview reviews. The US figure was 23.8 days in 2017 and 22.9 in 2014.
• Misleads by: Being old and including rejections. A quick rejection ends the clock early.
• Convert: Compare it with your application-to-decision days, counting rejections as well as offers. Don't compare it with acceptance dates.

"27.3 days" (NACE interview lag)

• Counts: Days between a student's first interview and an offer or a rejection, averaged across campus recruiting at NACE member employers. The 2019 survey figure was 23.5.
• Misleads by: Starting at the first interview and sampling only campus pipelines.
• Convert: It only applies to campus searches. Measure from your first interview to the decision.

Industry and seniority figures

• Counts: Glassdoor's industry split, from January to June 2017 data: Government 53.8 days, Aerospace and Defense 32.6, Energy and Utilities 28.8, Restaurants and Bars 10.2. SHRM's seniority split is executive 45 and nonexecutive 39. An older SHRM release gave 60 and 44.
• Misleads by: Age. The industry figures are nine years old. A healthcare figure of 27.0 shows up only in a secondary source, and no director-level figure has been published.
• Convert: Use the industry figures for ranking only. Government is slow and restaurants are fast. Don't use them as day counts.

How many interview rounds is normal, and whose interviews a count includes

In Ashby's data the hired candidate sits about 4 rounds for a business role and 5 for a technical role. Bigger figures like 11.7, 17.6, 20, or 35 count everyone interviewed for the seat or every individual session, not the rounds you sit.

For this guide, a round means a distinct scheduled stage you attend, with any same-day block counted once. The sources don't state this definition. I inferred it from how they count, and it's the definition to use when you log your own loop.

"11.7" and "17.6" interviews per hire (Ashby)

• Counts: How many candidates are interviewed for each hire, across all candidates.
• Misleads by: Sounding like the length of your loop. The technical figure rose from 11.6 to 17.6 while the hired candidate still sat about 5 rounds. The cause was more applicants interviewed per hire, as the share of applications that got an interview fell from 7 to 8% to between 3.6 and 4.7%.
• Convert: Read it as a measure of competition for the seat. For your loop length, use 4 or 5 rounds.

That falling interview rate is the part of these numbers you can actually act on. If fewer than one application in twenty gets an interview, sending a generic resume costs you more than it used to. Refolk tailors your resume to each posting from your own history, so a carefully targeted application doesn't take an evening's rewrite.

"14 to 20" and "35" interviews per hire (Gem)

• Counts: Sessions, meaning individual interviews conducted, across all candidates. Gem's figure rose from 14 to 20 per hire over its reported period.
• Misleads by: Getting reported as rounds. Pin's write-up calls Gem's 14 to 20 figure "average interview rounds per hire", even though Gem counts interviews conducted. A finalist who meets 5 people across 3 rounds adds 1 to Ashby's count and several to Gem's, which explains most of the gap between 17.6 and 35.
• Convert: Don't convert it. No published figure turns sessions into rounds per candidate, so this number tells you about interviewer workload and nothing about your loop.

Rising counts often reflect pipeline volume rather than longer loops. Célia Sauthier, Director of TA at Moss, reported interview volume "up by about 30%" without more hires.

"5 stages" and "7 or more stages" (Pin)

• Counts: Distinct interview stages configured for a role. Pin found that the median corporate hire passes through 5, based on more than 4,000 ATS-connected jobs, and 38.5% of roles are configured with 7 or more.
• Misleads by: Measuring configuration, not what candidates experienced. A role set up with 7 stages can skip some of them for a strong candidate. Same-day panels count as one stage here but as several in Gem's session count.
• Convert: This is close to the candidate's definition of a round. Compare it with your own stage count, counting same-day blocks once.

"5 interviews to 2 offers" (ZipRecruiter)

• Counts: New hires reported completing 5 interviews to secure 2 job offers over their whole search, across all employers. An earlier survey found 4 interviews across all companies but a median of 2 for the job respondents actually took.
• Misleads by: Getting read as one employer's loop. The survey also never defines "interview" for respondents.
• Convert: This is the closest public figure to "how many interviews per job offer" from the candidate's side. Compare it with your own count of interviews and offers across the whole search, and remember the unit isn't defined.

Hours per hire

Ashby's interviewer hours per hire (12.2 business, 23.3 technical) measure the employer's workload. The hired candidate spends about 2.4 and 3.5 hours, roughly a sixth of the interviewer time. If an "hours per hire" headline leaves you dreading the time you'll spend, it's describing the interviewers' time, not yours.

Interview-to-offer rate averages, by sample

Published interview-to-offer rates range from 22% at startups to between 46 and 49% in campus hiring. The spread comes from who was sampled and from whether declined offers are counted. None of these is a universal rate.

The conversion that matters most is: interview-to-hire equals interview-to-offer times offer acceptance. NACE's average interview-to-offer rate of 47.5%, multiplied by its 69.3% acceptance rate, gives about 33% interview-to-hire (derived). To go the other way, divide a hire rate by acceptance. Across published acceptance rates of 69% to 84%, that step moves the figure by up to a third. Nobody has publicly established how much of the spread in the table comes from population and how much from excluding declined offers.

A "1 in 4" headline usually lands between Gem's 22% startup interview-to-offer rate and CareerPlug's 27% small-business interview-to-hire rate. Those two figures measure different things, so check which one the headline is quoting before you use it. A third figure gets relayed secondhand: 72.2% for enterprise, 16.6% for mid-market, and 7.0% for small business. It hasn't been verified at the primary source, so I leave it out.

What each rate proves about your search

• Campus rates prove that a pre-filtered pipeline at large member employers converts well. They don't tell you anything about an experienced-hire search.
• Startup rates are the nearest match for an experienced candidate applying to small companies. Gem's sample is limited to companies under 500 employees. A second Gem segment shows 14%, but its label wasn't captured, so don't use it.
• Small-business rates move with applicant volume. CareerPlug's interview-to-hire halved between reports while applicants per hire rose from 85 to 180, so a year-old figure can't be carried forward. The current report covers more than 60,000 small businesses and more than 10M applications.
• Onsite-to-offer starts at a later stage, so it runs higher than interview-to-offer for the same employers. Compare it only with your own final-round results.

Every benchmark rate blends candidates who fit the role closely with candidates who barely fit, so it can't tell you whether one particular loop was a long shot. Refolk scores how well you fit each posting before you apply, which lets you separate the loops where you'd expect to beat the base rate from the ones where a rejection was always likely.

Who runs the loops behind the benchmarks

In Refolk's index of professional profiles, US Technical Recruiters outnumber UK ones by about 30 to 1, coordinators are a fraction of the recruiter headcount, and campus recruiters are a small group concentrated at a few employers. Each of these facts narrows which hiring a benchmark actually describes.

Here's how each count changes how you read a benchmark. I've marked where I'm inferring rather than reporting.

• Country. None of the timing or offer-rate sources above says its sample covers the UK. With the UK recruiter population this much smaller in Refolk's index, UK job seekers should treat these figures as describing a different market until a local source turns up. The top UK Technical Recruiter employers in a 25-profile sample are Monzo, Meta, and Capgemini.
• Scheduling. There are 6,663 Recruiting/Interview Coordinators in the US against 21,394 Technical Recruiters. The ratio suggests, but doesn't prove, that the person who schedules your rounds is often shared across several recruiters. It's one possible reason gaps between rounds run longer than the interviews themselves.
• Campus. Only 888 Campus/University Recruiters appear in the US, and the campus sample concentrates at Jane Street, Meta, and Eli Lilly. Together with NACE's 24.9% response rate (197 of 790 member organizations), that's why campus offer rates describe a narrow, large-employer slice of hiring.
• Agencies. The staffing agencies K2 Partnering Solutions and Experis are among the top US Technical Recruiter employers in a 25-profile sample. My inference, which nobody has established publicly, is that loops run through agencies may not show up in employer ATS benchmarks at all. If an agency placed you, published loop counts may not include searches like yours.

How to convert a hiring statistic to your own search

Work through seven fields in order, then convert or reject. It takes about 15 minutes per number, and most numbers fail at step 3 or step 4.

For step 5, SHRM's older nonexecutive time-to-fill distribution (n = 840) shows why the statistic type matters:

• 25th percentile: 28 days
• Median: 44 days
• 75th percentile: 73 days
• Mean: 54 days

The mean sits 10 days above the median because a few long searches pull it up. Wherever percentiles are published, compare yourself with them rather than with a single average.

Figure as quoted:
Source and collection method (self-report / HR survey / ATS timestamps):
Clock start (requisition / application / pipeline entry / first interview):
Clock stop (offer accepted / offer or rejection):
Unit (days / people per hire / rounds the hire sat / sessions):
Population (small business / campus / startup / enterprise / broad new hires):
Data year:
Statistic type (median / mean / percentile):
Conversion applied (e.g. minus ~12 days pre-screening; x acceptance rate; replaced with rounds):
Converted figure:
Verdict (compare / direction only / reject):

How published hiring statistics mislead job seekers

Most misuse comes from eight specific mix-ups, each of which can shift a figure by weeks or by a factor of two or more. Run each one as a check before you trust a number.

1. "44 days" read as interview time. It's time to fill, so it includes the days before you applied. Check: does the definition start at the requisition? If it does, subtract roughly 12 days (derived from the older SHRM data) or reject the comparison.
2. "5 interviews" read as one loop. ZipRecruiter's figure covers the entire search across all employers. The earlier per-job median was 2. Check: does the wording say the interviews were for the respondent's current job?
3. Gem's "20 interviews per hire" read as rounds. It counts sessions across all candidates. The hired person sits 4 to 5 rounds. Check: is the unit people, sessions, or stages?
4. A campus offer rate applied to an experienced-hire search. NACE samples are pre-filtered pipelines at member employers that chose to answer the survey, about a 25% response rate. Startup interview-to-offer is 22%. Check: does the sample hire students or experienced people?
5. Interview-to-offer confused with interview-to-hire. The gap between them is offer declines, and acceptance rates between 69% and 84% move the figure by up to a third. Check: does the rate stop at the offer or at the accepted hire?
6. Glassdoor's 2017 industry numbers quoted as current. One candidate-facing explainer calls the 23.8-day figure "outdated" and then quotes a 63 to 68 day "national average" without naming any source for it. Check: the data year, and whether the replacement figure has a named source at all.
7. Median and mean swapped. SHRM's older nonexecutive median of 44 days sits against a mean of 54. Check: which one is quoted, and whether percentiles exist.
8. Offer-or-rejection clocks read as time to offer. Glassdoor's and NACE's clocks stop at a rejection as well as an offer, so fast rejections may shorten the average. Nobody has measured this effect. It's my inference. Check: does the stop event include rejections?

A benchmark you can't define is worse than no benchmark, because it feels like evidence.

Two more limits apply across the whole table. First, nobody has publicly established whether self-reports, HR surveys, or ATS timestamps run fast or slow, so any adjustment for collection method is guesswork. Second, vendor explainers that start from metric definitions, such as Klipfolio's, use illustrative numbers rather than measured ones. Don't cite them as benchmarks.

Is my interview process taking too long? Holding your loop against a benchmark

Your loop is taking too long only when your own application-to-today count is past a converted median for a matching clock and population. Even then, it means you're behind half of hires, not that the process has stalled. To make that comparison you need your own dates logged the same way the benchmarks log theirs.

Log every search event against the same fields the sources use:

• Application date, which is your time-to-hire and Glassdoor start
• First interview date, which is NACE's start
• Each distinct scheduled stage, with same-day blocks counted once
• Decision date, whether offer or rejection
• Acceptance date, if you accept

Then compare like with like. For an experienced-hire business role, hold your application-to-acceptance days against Ashby's 30-day median and your round count against about 4. For a technical role, use 40 days and about 5 rounds. For a campus search, hold your first-interview-to-decision days against NACE's 27.3. If you're already past the median, the useful next step is to ask the recruiter how many stages remain. Treat that number as information, not a verdict.

Employer | Role type (business/technical/campus) | Applied | First interview | Rounds sat | Decision date | Offer or rejection | Accepted date | Days applied-to-decision | Benchmark used | Verdict

To keep this reference current, re-check the mechanism rather than memorising values. Each source publishes in editions. When a new one appears, confirm three things before you update your card: the data year, whether the definition page changed its start or stop event, and whether the sample changed size or type. CareerPlug's small-business rate halving between reports shows how fast a volume-driven figure can move, so any small-business figure older than one report cycle should go back through the steps.