Roughly 8 applications per offer. That’s the honest number, and it comes from two independent surveys built a decade apart that land in the same range: ZipRecruiter’s quarterly panel of people who just got hired, and our own analysis of the New York Fed’s public job-search microdata going back to 2013. It is not 100. It is not 200. The “100 to 200 applications,” “250 resumes per opening,” and “2% callback rate” claims that circulate every time someone posts about a hard search are not measurements of anything. Each traces to a specific, nameable place where a real number got replaced with a guess.
That matters beyond pride. Believe you need 150 applications, and a search that’s going fine will feel like a failure at application forty. The real data is more useful than a bigger target. It shows which kind of effort converts.
Where do these numbers come from?
“100 to 200 applications”
It comes from a single 2019 blog post by a now-defunct auto-apply startup, which multiplied two assumed rates together and called the result an expectation. It was never a measurement.
The post is Part IX of a series called “The Science of the Job Search,” published by TalentWorks (talent.works) on February 6, 2019. The relevant sentence, verbatim: “Given an interview rate of 10% … and a job offer rate of 10% … you’d expect it to take 100 applications before you have a job offer in hand.” Both rates are stated assumptions, not findings from the post’s own data. The one number in that piece that was actually measured, 92 applications, came from 50 users who self-reported success when they cancelled a paid auto-apply subscription, on a tool whose users submitted a median of six applications a day, one of them 125 in a single day. That is a small, self-selected sample using a product built to inflate the count, about the opposite of a representative measurement. The same company had published a separate post four months earlier admitting it had no answer: “we don’t have the hard science right now to give you a straight answer on that topic.”
“250 resumes per opening”
The “250 resumes per job opening” figure has an even thinner foundation. It traces to Dr. John Sullivan, writing on ERE.net on May 20, 2013, in a listicle of more than 35 recruiting statistics. Every other number in that piece carries a source in parentheses. The 250 does not. It sat uncited until Glassdoor’s employer blog reprinted it in January 2015, in a post its own landing page describes as a curation of “favorites from a variety of sources.” Glassdoor never ran that number. It just became attached to Glassdoor’s name because Glassdoor said it loudest, and the internet has cited “Glassdoor” for a stat Glassdoor never produced ever since.
“2% get an interview”
The “2% get an interview” line is a denominator swap on the same 2013 article. In a separate paragraph, Sullivan cites a hiring funnel attributed to Talent Function Group LLC: out of 100 completed applications, 4 to 6 people get interviewed, a real, attributed 4-to-6% rate. Glassdoor’s 2015 post welded it to the unrelated 250 with the phrase “of these candidates, four to six will be called for an interview,” silently swapping the denominator from 100 to 250. Divide 4 to 6 by 250 instead of 100 and you land on 1.6% to 2.4%, rounded to “2%,” a figure built by re-basing one real, attributed rate onto an unrelated, uncited number rather than by measuring anything new.
The “BLS” number that never existed
One more number is worth naming because it borrows a specific agency’s credibility: a stat claiming the Bureau of Labor Statistics found a 30.89% offer rate for 21 to 80 applications, and 20.36% for 81 or more. No such BLS release exists. The BLS Current Population Survey does not even count applications sent; it only records whether someone used at least one active search method, a checklist that includes “submitting a resume or application” alongside networking and contacting employers directly. There is no BLS series that could produce a number like 30.89%, and the two-decimal precision is the kind of false specificity that shows up when a made-up figure gets passed along enough times to start sounding official.
So what does the real data say?
The current best instrument is ZipRecruiter’s Survey of New Hires. In its Q2 2026 wave, more than 1,500 U.S. adults who started a job in the past six months reported a median of 16 applications submitted, 5 interviews completed, and 2 offers received, over a median five-week search.
We ran our own analysis of the New York Fed’s public Job Search Supplement microdata, an annual survey fielded every October since 2013, pooled across all available waves through 2021. Among 5,085 people who reported the number of applications they sent before accepting the job they currently hold, the median is 2, with 29.4% reporting zero. Restricting to the 3,491 people who reported both an application count and an offer count, the median is 3 applications and 1 offer, and the average is 13.34 applications against 1.54 offers, a ratio of about 8.7 applications per offer among people who applied at all (7.2 if the zero-application group is included). Either way, it converges with ZipRecruiter’s independent figure of roughly 8, from a completely different sample and decade.
Two caveats matter more than the number itself. Both surveys only include people who got hired; a search that stalled, or is still running, is not in either sample, so 8-per-offer describes what a successful search looked like, not what a search requires. And the New York Fed figure is a career-long recall question about the job someone currently holds, which can be from any point in their working life, not a snapshot of this year’s market. Read it as “people who landed a job reported roughly this many applications,” never as “you need roughly this many.”
One more data point backs the same conclusion. Faberman, Mueller, Şahin, and Topa, the economists who designed the New York Fed survey, published tables (working paper, 2017; peer-reviewed in Econometrica, 2022) showing employed job seekers received offers at a noticeably higher rate per application than unemployed job seekers, despite sending roughly half as many applications a month, a pattern that lines up with the referral findings later in this piece more than with any theory about effort or volume.
Why do you also see “244 applications per job” everywhere?
Because that is a real, measured number too. It just answers a different question: how many applications land on one job posting, not how many a person sends before getting hired.
Greenhouse, an applicant tracking system used by thousands of companies, measured 244 applications per job posting in 2025, more than double the 116 it measured in 2022, based on 640 million-plus applications across 6,000-plus companies. Ashby, a competing ATS, measured 291 applications per hire in the first quarter of 2026, drawing on 109 million applications and 247,000 jobs since 2021, after peaking above 300 through most of 2025. CareerPlug’s 2025 report, built from 10 million-plus applications across 60,000-plus small businesses, found about 180 applicants per hire and a 3% applicant-to-interview rate. Read that last one as small-business hiring specifically; CareerPlug’s customer base skews retail, restaurants, healthcare, and trades, and does not generalize to corporate or tech hiring.
Line up the employer-side number against the job-seeker-side number and the gap looks enormous, roughly 15 times bigger. It’s not a contradiction. Each person applies to many postings at once, so one application sits inside dozens of employers’ denominators simultaneously, while a given posting’s 244 applications came from 244 different people, almost none of whom are the person who eventually gets hired.
The trend makes the two-sided story clearer still. Applications per posting have more than doubled since 2022. Over the same stretch, the median applications behind a successful new hire has been falling: 30 in the second quarter of 2025, down to 24, then 19, then 16 a year later, per ZipRecruiter’s quarterly series. Both real, both measured, and moving in opposite directions. The market got more crowded on the posting side while individual successful searches got shorter. ZipRecruiter’s own reading is that new hires are settling faster, taking the first solid offer rather than holding out for a better one.
Why do people confuse this with a different NY Fed survey?
The New York Fed runs two different surveys about jobs, and only one of them measures applications. The one that generates news headlines every four months, the Labor Market Survey, asks whether someone searched in the past four weeks and how many offers they received in the past four months. It does not ask how many applications they sent. The string “applicat” does not appear anywhere in its questionnaire.
The survey that does ask about applications is a separate annual module, the Job Search Supplement, fielded every October since 2013, the source behind the 8.7 figure above. It asks a retrospective question about the job someone currently holds, not a snapshot of the last few weeks. Its public microdata runs through the October 2021 wave; nothing newer has been published as of this writing, and whether the module stopped or the data simply have not been released is genuinely unclear from the New York Fed’s own materials, so we are not going to guess.
The well-known survey adds its own confusion on top: it asks whether someone searched in the past four weeks, and separately, how many offers they received in the past four months, two different windows in the same instrument. Dividing one by the other is not a ratio anyone designed.
What predicts an offer without an application?
Referrals and direct contact convert without a formal application far more often than searching job boards does. That’s the finding underneath both the survivorship caveats above.
In the New York Fed sample, 29.4% of the jobs people reported holding required no formal application at all. Cross-tabbing how those people learned about the job against how the people who did apply learned about theirs turns up a stark inversion: 38.2% of the zero-application group learned about the job through a friend or family referral, versus 23.7% of the people who applied; 15.5% came through unsolicited contact from the employer, versus 6.8%. Online job search ran in reverse: 18.2% of people who applied found the job that way, versus 2.0% of people who never applied at all. On the same list of channels, referrals and direct outreach swing nearly twelve points one way while job boards swing the other. That’s a reversal, not a rounding difference.
The employer side of the market shows the same pattern from the opposite direction. Ashby’s 2026 report found that 52% of referred candidates pass an employer’s initial screen, against 35% overall. Two unrelated datasets, one built from job seekers describing how they got hired and one built from an ATS logging who passed a screen, arrive at the same conclusion: the channel someone comes through does more of the work than the number of applications sent.
So what should you do differently?
Not chase a number.
None of the figures in this piece (8, 16, 244, 29.4%) are a personal quota, and treating any of them as one is exactly the mistake this piece exists to correct. They are medians and averages computed over people who had already succeeded by the time anyone asked them a question. Your search does not come with a scoreboard showing how close you are to someone else’s average.
The constructive read isn’t “apply less.” Most people still need to send applications, and online postings remain a real channel into a real share of jobs. But most searches underuse one thing: the time spent on channels that skip a formal application entirely, asking someone who already works at a company for an introduction, reaching out directly to a hiring manager, following up on a connection instead of letting it go cold. The data above puts these channels in a different range, not a small step up from a cold application.
If you’re already applying regularly, it also helps to see your own numbers instead of guessing at them against someone else’s median. That’s part of why CareerPigeon’s application tracker exists: log an application in one click as you send it, from the confirmation email you already have open, and over a few weeks you have your own ratio of applications to responses to look at, instead of importing a stranger’s average and treating it as your own.
How we know this: the 8.7-applications-per-offer figure is our own weighted analysis of the New York Fed’s public Job Search Supplement microdata (2013–2021 pooled waves, n=5,085 for the application count, n=3,491 for the joint sample), reproduced independently before publication. The current-market figure, 16 applications and 2 offers, is ZipRecruiter’s Survey of New Hires, Q2 2026, read directly off the survey’s own published page; note its accompanying prose calls the figure an “average” when the data tiles are labeled “median.” The employer-side figures come from three applicant tracking systems’ own published benchmark reports: Greenhouse, Ashby, and CareerPlug (small-business-only). We traced the “100 to 200 applications,” “250 resumes,” and “2% interview rate” claims to their origin documents directly, and checked the circulating “BLS 2020” statistic against BLS’s own Current Population Survey definitions, which do not support it. Where a figure carries a scope limit, career-long recall, a small-business-only sample, a single vendor’s panel, we have said so in the sentence that uses it.