Three statistics carry almost every argument for investing in an employer brand, they come from the same company, and two of them can no longer be reached at their source.

That is worth establishing before anything else, because a mid-size company deciding how much to spend on this deserves to know that the business case circulating in the market has one issuer.

“A strong employer brand cuts cost per hire by up to 50 percent.” The origin is a whitepaper published by a hiring platform in March 2012. That paper does contain a real survey, of 7,250 members worldwide, with a stated margin of error. But the cost-per-hire claim is not from it. It sits in a figure introduced with the words “In previous research conducted with corporate recruiters, we determined that companies investing in their employment brand enjoy lower cost per hire”, followed by a shortened link.

The figure itself is an index with no units: 125 for a poor or moderate employer brand, 67 for a good or strong one. Sixty-seven divided by 125 is a 46 percent difference, rounded up in every retelling since. No currency, no sample size, no date, no method. And the link to the underlying research now redirects to a product page.

“A strong employer brand reduces turnover by 28 percent.” Same paper, same unnamed prior research, same absence of method.

“A bad reputation costs at least 10 percent more per hire.” This one is usually attributed to a consulting firm. It is not. It comes from an article published in March 2016 by a person the article itself identifies as the global vice president of the same platform’s talent solutions division.

Its method, from the archived version before the article went behind a paywall: a survey of 1,003 full-time professionals in the United States, conducted with an external research firm. So the 10 percent is what a thousand people said they would require, not a wage anyone paid.

The derived figure is worse. The article computes an extra $4,723 per hire from three inputs: an average salary of $47,230, an assumed annual turnover of 16.4 percent attributed to a private data vendor, and the 10 percent premium. Note that $4,723 is exactly ten percent of $47,230. The per-hire number is a restatement of the survey answer, not a second finding.

And the same platform’s own post the following day attributes the turnover input differently, to “anecdotal evidence from friendly HR professionals about turnover rates.”

“86 percent of job seekers read reviews before applying.” This one appears on an employer-review site’s own statistics page with no citation attached to it. Its citations block lists only generic site surveys of that platform’s own users. Asking people who are already on a review site whether they read reviews is not a population estimate.

Provenance of the three statistics most used to justify employer brand investmentProvenance of the three statistics most frequently used to justify investment in an employer brand, all of which originate from the same commercial issuer. The first claim, that a strong employer brand cuts cost per hire by up to fifty percent, originates in a whitepaper published by a hiring platform in March two thousand twelve. That whitepaper does contain a genuine survey of seven thousand two hundred and fifty members worldwide with a stated margin of error of plus or minus one point one five percent at ninety five percent confidence, but the cost per hire claim does not come from that survey. It appears in a figure introduced with the words that in previous research conducted with corporate recruiters the authors determined that companies investing in their employment brand enjoy lower cost per hire, followed by a shortened link. The figure itself is an index with no units, showing one hundred and twenty five for a poor or moderate employer brand against sixty seven for a good or strong one, which is a forty six percent difference rounded up in every subsequent retelling, with no currency, no sample size, no date and no method stated, and the link to the underlying research now redirects to a product page. The second claim, that a strong employer brand reduces turnover by twenty eight percent, comes from the same paper and the same unnamed prior research with the same absence of method. The third claim, that a bad reputation costs at least ten percent more per hire, is commonly attributed to a consulting firm but in fact comes from an article published in March two thousand sixteen by a person the article itself identifies as the global vice president of the same platform’s talent solutions division. Its method was a survey of one thousand and three full time professionals in the United States conducted with an external research firm, so the ten percent is what respondents said they would require rather than a wage anyone paid, and the derived figure of four thousand seven hundred and twenty three dollars per hire is exactly ten percent of the forty seven thousand two hundred and thirty dollar average salary used as an input, making it a restatement of the survey answer rather than a second finding. The turnover input for that calculation is attributed to a private data vendor in one version and to anecdotal evidence from friendly human resources professionals in the version the platform published itself the following day.Three claims, one issuer”Cuts cost per hire by up to 50 percent”An unitless index in a 2012 whitepaper: 125 against 67, which is 46 percent. Attributed to”previous research” with no sample, no date, no method. The source link is dead.”Reduces turnover by 28 percent”Same paper, same unnamed prior research, same absence of method.”A bad reputation costs 10 percent more per hire”Not a consulting firm. An article by the same vendor’s global VP of talent solutions, 2016.A survey of 1,003 professionals saying what they would demand. Not a wage anyone paid.And $4,723 per hire is exactly ten percent of the $47,230 salary used as an input.
One issuer, two dead links, one indexed chart with no units, and one survey answer restated as a dollar figure. Source : Vendor whitepaper of March 2012; Harvard Business Review article of 29 March 2016 by that vendor's global VP of talent solutions, archived version; the vendor's own blog post of 30 March 2016 (2016)

What no United States institution publishes

Having removed the vendor numbers, the reasonable next question is what the government measures instead. On the two figures every hiring business case wants, the answer is nothing.

Cost per hire. The Bureau of Labor Statistics does not publish one. Its closest program, Employer Costs for Employee Compensation, measures hourly compensation, not recruiting expenditure. The Department of Labor publishes none either. The only American standard on the subject, approved in 2012, is a calculation formula, internal plus external recruiting costs divided by hires, published by a trade association. It defines how to compute the number. It supplies no benchmark.

So every dollar figure you have seen for cost per hire comes from a self-reported association survey or from a software vendor. The nearest thing to a rigorous measurement in the academic literature used Swiss administrative firm data and found average hiring costs of 10 to 17 weeks of wages depending on firm size, with no fixed cost component detected. That is a real result, and it is Swiss.

Time to hire. The only recurring public series was discontinued in 2018. Its academic basis measured average vacancy durations of 14 to 25 days on data from 2001 to 2009. Citing either as current is citing something that no longer exists.

There is one recent institutional measure, and it is a one-off. A 2022 survey of establishments found that 15.8 percent filled open positions in 30 days or less and 7.0 percent took more than 30 days, meaning roughly a third of establishments that were hiring exceeded a month. Its usable response rate was 26.7 percent, which the agency publishes rather than hides.

What United States federal statistics measure about hiring costs and duration, and what they do notWhat United States federal statistics measure about hiring costs and hiring duration, and what they do not measure. On cost per hire, the Bureau of Labor Statistics publishes no figure, since its closest program, Employer Costs for Employee Compensation, measures hourly compensation rather than recruiting expenditure, and the Department of Labor publishes none either. The only American national standard on the subject, approved in February two thousand twelve, is a calculation formula defining cost per hire as internal plus external recruiting costs divided by the number of hires, published by a professional trade association, and it supplies no benchmark data. Consequently every circulating dollar figure for cost per hire originates in a self reported association survey or in human resources software vendor material. The nearest rigorous academic measurement used Swiss administrative firm data and found average hiring costs of ten to seventeen weeks of wages depending on firm size, with a convex structure reaching twenty four weeks at the margin and no fixed cost component detected, but those data are Swiss rather than American. On time to hire, the only recurring public series was discontinued in two thousand eighteen, and its academic basis measured average vacancy durations of fourteen to twenty five days on data covering two thousand one to two thousand nine, so citing either as current describes something that no longer exists. There is one recent institutional measure and it is a one off survey conducted between the first of August and the thirtieth of September two thousand twenty two and released in March two thousand twenty three, which found that fifteen point eight percent of establishments filled open positions in thirty days or less while seven point zero percent took more than thirty days, meaning roughly a third of establishments that were hiring exceeded a month, and that twelve point three percent of establishments had at least one position open for more than thirty days. That survey drew a stratified sample of about three hundred and forty thousand establishments from a universe of more than eight point nine million and reported a usable response rate of twenty six point seven percent, which the agency publishes openly.The two figures nobody official publishesCost per hireNo federal figure. The one Americanstandard is a formula, not a benchmark.Every dollar figure is vendor or association data.Time to hireThe only recurring series wasdiscontinued in 2018.Its data covered 2001 to 2009.The one recent institutional measure, a 2022 one-offFilled open positions in 30 days or less15.8%Took more than 30 days7.0%Usable response rate 26.7 percent, published by the agency rather than hidden.
No cost per hire. No current time to fill. One discontinued series and one 2022 snapshot with a published response rate. Source : BLS Employer Costs for Employee Compensation; ANSI/SHRM-06001-2012; BLS 2022 Business Response Survey on Telework, Hiring, and Vacancies, released 22 March 2023 (2023)

What the government does publish, and what has happened to it

The recurring measure that exists is the Job Openings and Labor Turnover Survey. For July 2026 it reports 7,271,000 job openings, a rate of 4.4 percent, with 5,054,000 hires, 3,056,000 quits and 1,666,000 layoffs and discharges. Openings a year earlier stood at 7,089,000.

The method is published in full: a stratified random sample of about 21,000 nonfarm business and government establishments, stratified by ownership, region, industry sector and size class, drawn from a frame of roughly 9.4 million establishments covering about 95 percent of nonfarm payroll employment, with confidence intervals published at 90 percent.

And here is the detail nobody in hiring commentary mentions. The survey’s unit response rate has collapsed: about 67.0 percent in July 2016, 34.5 percent in January 2025, 37.9 percent in July 2025, a series low of 29.7 percent in September 2025, and 35.0 percent in December 2025.

The most quoted labor market series in the world now rests on roughly a third of its intended sample. The agency publishes that fact openly, which is exactly the behavior the vendor statistics above do not exhibit. But anyone building a hiring plan on national openings figures should know the denominator.

United States job openings and turnover figures alongside the survey response rate behind themUnited States job openings and labor turnover figures alongside the survey response rate behind them. For July two thousand twenty six the survey reports seven million two hundred and seventy one thousand job openings at a rate of four point four percent, five million and fifty four thousand hires at a rate of three point two percent, five million and seventy two thousand total separations, three million and fifty six thousand quits at a rate of one point nine percent, one million six hundred and sixty six thousand layoffs and discharges at a rate of one point zero percent, and three hundred and fifty thousand other separations. Job openings a year earlier in July two thousand twenty five stood at seven million and eighty nine thousand at a rate of four point three percent. The method is published in full: a stratified random sample of approximately twenty one thousand nonfarm business and government establishments, stratified by ownership, region, industry sector and establishment size class, drawn from a frame of roughly nine point four million establishments covering about ninety five percent of nonfarm payroll employment, with collection by computer assisted telephone interviewing for about five months followed by web or email self report, a birth and death model added to sample estimates, alignment to the payroll survey, and confidence intervals published at ninety percent. The detail rarely mentioned in hiring commentary is that the survey’s unit response rate has collapsed over the past decade, standing at about sixty seven percent in July two thousand sixteen, thirty four point five percent in January two thousand twenty five, thirty seven point nine percent in July two thousand twenty five, a series low of twenty nine point seven percent in September two thousand twenty five, and thirty five point zero percent in December two thousand twenty five, with no data available for August two thousand twenty five owing to a lapse in appropriations. The most quoted labor market series in the world therefore now rests on roughly a third of its intended sample, a fact the agency publishes openly.The official series, and its denominatorJuly 2026, seasonally adjusted, thousandsJob openings7,271rate 4.4%Hires5,054rate 3.2%Quits3,056rate 1.9%Layoffs and discharges1,666rate 1.0%Unit response rate for the same surveyJuly 2016about 67%July 202537.9%September 202529.7%
The series is published with its own method and its own response rate. That rate has fallen from about 67 percent to under 30. Source : BLS, Job Openings and Labor Turnover, July 2026, USDL-26-1432, released 1 September 2026; BLS establishment survey unit response rates (2026)

What your job advertisement cannot say

While the statistics are unreliable, the rules are precise, and they are cheaper to comply with than to litigate.

Federal law makes it unlawful to print or publish any notice or advertisement relating to employment “indicating any preference, limitation, specification, or discrimination” based on race, color, religion, sex or national origin, with a narrow occupational qualification exception. The age statute imposes the same prohibition without that exception in the equivalent subsection.

The implementing regulation on job advertisements is unusually specific about wording, and this is the paragraph worth pinning above a hiring manager’s desk:

“Help wanted notices or advertisements may not contain terms and phrases that limit or deter the employment of older individuals. Notices or advertisements that contain terms such as age 25 to 35, young, college student, recent college graduate, boy, girl, or others of a similar nature violate the Act unless one of the statutory exceptions applies. Employers may post help wanted notices or advertisements expressing a preference for older individuals with terms such as over age 60, retirees, or supplement your pension.”

Note that the asymmetry is deliberate: preferring older applicants is expressly permitted. And asking applicants to state their age is not itself a violation, but the regulation says such notices “will be closely scrutinized”.

One further point from the enforcement agency’s own guidance: discriminatory recruiting is unlawful even with no advertisement at all, and the example it gives is word-of-mouth hiring inside a homogeneous workforce. For a mid-size company that hires mostly through referrals, that is the relevant exposure, and no employer brand campaign addresses it.

Job advertisement wording named as violating by federal regulation and wording expressly permittedJob advertisement wording named as violating by United States federal regulation, and wording that the same regulation expressly permits. Federal law makes it an unlawful employment practice to print or publish, or cause to be printed or published, any notice or advertisement relating to employment indicating any preference, limitation, specification or discrimination based on race, color, religion, sex or national origin, subject to a narrow bona fide occupational qualification exception for religion, sex or national origin. The age discrimination statute imposes the same prohibition in its equivalent subsection without any such exception. The implementing regulation on help wanted notices states that such notices may not contain terms and phrases that limit or deter the employment of older individuals, and it names specific violating terms, namely age twenty five to thirty five, young, college student, recent college graduate, boy, girl, and others of a similar nature, which violate the Act unless a statutory exception applies. The same regulation expressly permits employers to post notices expressing a preference for older individuals using terms such as over age sixty, retirees, or supplement your pension, so the asymmetry is deliberate. It further states that notices asking applicants to disclose or state their age do not in themselves violate the Act, but that because such requests may tend to deter older individuals from applying, notices including them will be closely scrutinized. The enforcement agency’s public guidance adds that discriminatory recruiting is unlawful even where no advertisement exists at all, giving as its example word of mouth recruiting within a homogeneous workforce, which for a mid size company hiring mainly through referrals is the relevant exposure and one that no employer brand campaign addresses.The words the regulation namesNamed as violating”age 25 to 35""young""college student”, “recent college graduate""boy”, “girl”Expressly permitted”over age 60""retirees""supplement your pension”The asymmetry is deliberate.And the exposure with no advertisement at allDiscriminatory recruiting is unlawful without any notice. The agency’s own example is word-of-mouthhiring inside a homogeneous workforce. No campaign addresses that.
The regulation lists the offending terms by name. The asymmetry in favor of older applicants is deliberate. Source : 29 C.F.R. 1625.4, Help wanted notices or advertisements; 42 U.S.C. 2000e-3(b); 29 U.S.C. 623(e); EEOC prohibited employment policies and practices (2026)

What a mid-size company can actually do

Strip out the vendor arithmetic and what remains is unglamorous, cheap, and defensible.

Measure your own numbers, since nobody else has any. Your applications per opening, your days from posting to signed offer, your offer acceptance rate, your twelve-month retention by hiring source. Every one of those is a fact about your company, obtainable from your own records, and more useful than a national figure that does not exist.

Stop importing benchmarks. There is no published American cost per hire and no current published time to hire. A target imported from a vendor deck is not a target, it is a decoration. Your own trailing twelve months is a target.

Fix what candidates can verify before you fund what they cannot. Response time to an application. Whether the salary range is published. Whether the interview process is described before someone applies. Whether the person who interviewed them ever follows up. Those are the parts of an employer brand that a candidate experiences directly, and none requires a campaign. None of them survives a company that says one thing to buyers and another to candidates either, which is why the promise and the tone are worth fixing once, at the level of the brand rather than the careers page.

Audit the wording. The regulation names the offending terms explicitly. Reading your own job postings against that list costs an hour and removes a documented category of risk.

Handle employee advocacy properly. If you encourage employees to write about the company publicly, the federal endorsement guidance says the relationship must be disclosed clearly, that listing an employer in a profile is not sufficient, and that a company that actively encourages such posts becomes responsible for monitoring them. A one-page policy and a required disclosure line covers it.

And say where your numbers came from. The whole problem described in this article exists because a set of plausible figures circulated without provenance for a decade. The fix is not more figures. It is a sentence naming the source next to each one, including when the source is your own spreadsheet.

Internal hiring measures a company can compute itself in the absence of published national benchmarksInternal hiring measures that a company can compute from its own records in the absence of published national benchmarks, since no United States federal institution publishes a cost per hire and the only recurring public vacancy duration series was discontinued in two thousand eighteen. The first measure is applications per opening, which indicates whether the company is reaching enough candidates at all and can be tracked by role and by channel. The second is days from posting to signed offer, which is the company’s own version of the time to hire figure that has no current national equivalent, and which the one recent institutional survey suggests exceeds thirty days for roughly a third of establishments that are hiring. The third is offer acceptance rate, which is the closest internal proxy for whether the company’s reputation and terms are competitive at the point of decision. The fourth is twelve month retention broken down by hiring source, which reveals whether referrals, job boards, agencies or direct applications produce people who stay, and which is the only one of these measures that connects hiring practice to a business outcome. The fifth is time to first response after an application, which is the part of the candidate experience most directly observable by the candidate and the cheapest to improve. Each of these is a fact about the company rather than an imported figure, is obtainable from existing records at no cost, and is more useful for planning than a national statistic that does not exist. The accompanying discipline is to name the source next to every number used in a hiring business case, including when that source is the company’s own spreadsheet, since the traceability problem described in this article arose precisely because plausible figures circulated for a decade without provenance.Measure these, because nobody else doesApplications per openingAre you reaching anyone at all? By role, by channel.Days from posting to signed offerYour own version of a figure with no national equivalent.Offer acceptance rateThe closest proxy for reputation at the point of decision.Twelve-month retention by hiring sourceThe only one that connects hiring to an outcome.And name the source beside every number, including when the source is your own spreadsheet.
Each one comes from your own records, describes your own company, and beats a benchmark that was never published. Source : Derived from the absence of published federal cost-per-hire and time-to-hire data documented in this article (2026)