Evidence Observatory
Where Job-Search Statistics Come From
Where job-search statistics come from, what they actually measured, and what they will not support. Every claim here has been traced as far toward its original source as we could get.
12 claims reviewed · 9 sources · Each review names its population, sample and measure
Most job-search statistics circulate without a source. Some have one and have been stretched past what it measured; a few are exactly right and are quoted without the qualification that makes them useful. This library records which is which, and shows the working.
A verdict here is a statement about evidence, never about the world. “Unverified” means we could not find support for a claim — not that it is false.
Check a claim
Paste a statistic you have seen and we will tell you whether it has been reviewed here, and what we found.
Unsupported or unverified
Claims we could not trace to a primary source, or where the evidence points the other way. These are the figures most likely to be quoted at you as fact.
75% of resumes are rejected by an ATS before a human ever sees them.
A figure with no traceable source, describing a quantity no available study measures.
Applicant tracking systems · 2 sources · Reviewed 2026-08-22
Unverified80% of jobs are found through networking.
No traceable source for the 80% figure, though the underlying idea has real research behind it.
Networking and referrals · 1 source · Reviewed 2026-08-22
Unverified70–85% of jobs are never publicly advertised.
No primary source located for any figure in this range; the phenomenon is real but unmeasured.
Networking and referrals · 1 source · Reviewed 2026-08-22
UnverifiedIt takes 100 to 200 applications to get a job.
No study measures this. The quoted range comes from informal sources, not measurement.
Applications · 2 sources · Reviewed 2026-08-22
Unverified
Recently reviewed
Most recently checked against their sources.
75% of resumes are rejected by an ATS before a human ever sees them.
A figure with no traceable source, describing a quantity no available study measures.
Applicant tracking systems · 2 sources · Reviewed 2026-08-22
UnverifiedMore than 90% of employers use software to filter or rank candidates before a human reviews them.
Supported for large employers: over 90% use software to filter or rank before human review.
Applicant tracking systems · 1 source · Reviewed 2026-08-22
SupportedApplicant tracking systems automatically screen out resumes that show an employment gap.
Supported: just under half of surveyed employers auto-exclude resumes with a gap over six months.
Applicant tracking systems · 1 source · Reviewed 2026-08-22
SupportedMost resumes never reach a human being.
Depends on an unstated definition. No source measures how many applications a human actually sees.
Applicant tracking systems · 2 sources · Reviewed 2026-08-22
Context dependentRecruiters spend only six seconds looking at a resume.
A real but small vendor eye-tracking study, describing an initial glance rather than total review time.
Resumes · 1 source · Reviewed 2026-08-22
Partially supported
By topic
Each area has its own measurement problem.
Topics appear here once they hold enough distinct reviews to be worth reading as a group.
Applicant tracking systems
4 reviews
A figure with no traceable source, describing a quantity no available study measures.
Resumes
2 reviews
A real but small vendor eye-tracking study, describing an initial glance rather than total review time.
Networking and referrals
3 reviews
No traceable source for the 80% figure, though the underlying idea has real research behind it.
Applications
3 reviews
Supported by two large randomised field experiments two decades apart.
Every claim reviewed
75% of resumes are rejected by an ATS before a human ever sees them.
A figure with no traceable source, describing a quantity no available study measures.
Applicant tracking systems · 2 sources · Reviewed 2026-08-22
UnverifiedMore than 90% of employers use software to filter or rank candidates before a human reviews them.
Supported for large employers: over 90% use software to filter or rank before human review.
Applicant tracking systems · 1 source · Reviewed 2026-08-22
SupportedApplicant tracking systems automatically screen out resumes that show an employment gap.
Supported: just under half of surveyed employers auto-exclude resumes with a gap over six months.
Applicant tracking systems · 1 source · Reviewed 2026-08-22
SupportedMost resumes never reach a human being.
Depends on an unstated definition. No source measures how many applications a human actually sees.
Applicant tracking systems · 2 sources · Reviewed 2026-08-22
Context dependentRecruiters spend only six seconds looking at a resume.
A real but small vendor eye-tracking study, describing an initial glance rather than total review time.
Resumes · 1 source · Reviewed 2026-08-22
Partially supportedIncluding a cover letter increases your chance of getting an interview.
One real field experiment supports tailored cover letters, but publishes no base rates.
Resumes · 1 source · Reviewed 2026-08-22
Partially supported80% of jobs are found through networking.
No traceable source for the 80% figure, though the underlying idea has real research behind it.
Networking and referrals · 1 source · Reviewed 2026-08-22
Unverified70–85% of jobs are never publicly advertised.
No primary source located for any figure in this range; the phenomenon is real but unmeasured.
Networking and referrals · 1 source · Reviewed 2026-08-22
UnverifiedEmployee referrals make a candidate several times more likely to be hired.
Referrals genuinely matter, but the popular multiplier is not what the academic evidence measures.
Networking and referrals · 1 source · Reviewed 2026-08-22
Context dependentIdentical resumes get different responses depending on the applicant's name.
Supported by two large randomised field experiments two decades apart.
Applications · 2 sources · Reviewed 2026-08-22
SupportedOnly 2% of applicants are invited to an interview.
No single rate exists. Published figures differ mainly in what they count, not in what they found.
Applications · 3 sources · Reviewed 2026-08-22
Context dependentIt takes 100 to 200 applications to get a job.
No study measures this. The quoted range comes from informal sources, not measurement.
Applications · 2 sources · Reviewed 2026-08-22
Unverified
Evidence status
What each verdict commits us to.
- Contradicted
- Identifiable evidence of reasonable quality points the other way. Reserved for cases where we can name that evidence, not for claims that are merely unsupported.
- Source withdrawn
- The claim did have a traceable source, and that source has since been retracted, corrected, taken down, or disavowed by whoever published it.
- Unverified
- We could not trace the claim to any identifiable primary source. This is a statement about the evidence, not about the world: an unverified claim is not thereby false, it is unsupported, and we say so rather than guessing.
- Partially supported
- A real source underlies the claim, but the claim as circulated overstates it — by generalising a narrow population, dropping a qualifier, rounding a range to a single number, or restating a measure as something it did not measure.
- Context dependent
- The claim is true or false depending on a condition the claim itself omits — the industry, the seniority, the labour market, or which of several different quantities is being counted. There is no single correct answer to report.
- Supported
- An identifiable primary source supports the claim as it is normally stated, and the claim does not overreach that source's population, period or measure.
These definitions are part of the methodology, alongside the source standard every review is held to.