1. Two statistics to stop repeating
Before the trends, the two figures that dominate CV advice — including advice written in 2026, quoting research from 2012.
"Recruiters spend six seconds on your CV"
This traces to an eye-tracking study by the careers site Ladders. The six seconds is the 2012 figure; Ladders' own 2018 follow-up measured 7.4 seconds, up from six. Most of the internet is quoting a fourteen-year-old number as though it were current.
Three caveats matter more than the decimal. The published methodology states no sample size, no participant count and no résumé count — so the precision of "7.4" is not supported by anything disclosed. It measured the initial screen only, not total review time, so "recruiters spend seven seconds on your CV" is a stronger claim than what was actually observed. And it has never been replicated; the underlying data was never released.
What survives is the direction, not the measurement: first impressions form fast, and clear structure is what survives that first pass. Which is the same advice the six-second version was being used to support, minus the false authority.
"75% of CVs are rejected by software before a human sees them"
This one has no traceable primary source at all. It is repeated constantly and attributed loosely to "studies", but no study with a published method or sample has been produced for it. Treat it as folklore.
The verifiable picture is narrower and more useful. A Jobscan analysis of Fortune 500 careers pages (November 2024) found 98.4% of Fortune 500 companies — 492 of 500 — use a detectable applicant tracking system. Read that precisely: it is a Fortune 500 careers-page analysis, not a statement about all employers, and "uses an ATS" is not the same as "auto-rejects". Software is near-universal for receiving and organising applications; automatic rejection is a much smaller and more specific claim.
What the systems mostly do at scale is sort, not reject. Where applications do get removed without a human, it is usually a knockout question — work authorisation, a required licence, a location — not a formatting judgment. That distinction matters, because it changes what you should optimise: get the eligibility facts unmissable, and stop worrying that a two-column layout is being silently binned.
2. AI on both sides of the process
The real change since 2024 is not that candidates use AI. It is that employers now assume they do, and have started hiring defensively.
Greenhouse's 2025 AI in Hiring Report (November 2025; 4,136 respondents across the US, UK, Ireland and Germany) found 74% of US job seekers use AI somewhere in their job hunt, and 49% apply to more roles specifically to get past automated filters. On the employer side, 91% of US hiring managers said they had caught or suspected AI-driven misrepresentation by a candidate, and 39% had added more in-person interviews specifically to verify authenticity.
Those two facts together are the trend. Applications are being generated faster and in greater number; employers have responded by trusting the written document less. The CV is not becoming irrelevant, but it is becoming a claim that the interview now exists to test.
One number from the same report deserves to be better known, because it undercuts the confident version of the automation story: only 21% of US recruiters said they were very confident their AI or ATS systems were not rejecting qualified candidates. The people operating the filters do not trust the filters.
The practical consequence for your CV is not "avoid AI". It is that specificity is now a signal. When a reviewer assumes a document may have been generated, the details that could only have come from actually doing the work — the number, the tool, the awkward constraint, the thing that went wrong — are what read as human. That was always good advice; it is now load-bearing.
3. The fear gap
Candidate anxiety about screening is now measurable, and it is larger than the thing being feared.
The 2026 State of Resumes report from Monster (January 2026; 1,001 US job seekers) found 77% worry their CV will be filtered out before a person reads it, while only 6% believe CVs are read thoroughly. The same survey found 68% spend less than 30 minutes tailoring a CV per application.
Read those three together. Most people are afraid of being filtered, most do not believe anyone reads carefully, and most respond to that by spending half an hour per application. The fear is rational; the response is the one thing that makes it self-fulfilling. If a document is being skimmed quickly, half an hour spent making the top third answer the posting is worth more than a week spent polishing page two.
4. Application volume
The competition is real and it is measurable in pipeline data. iCIMS, which processes applications for a large number of employers, reported an average of 33 applicants per opening in October 2025, up from 30 a year earlier — while hiring ran about 5% below the prior year.
More applications, fewer hires, per role. That is the arithmetic behind everything else in this article, and it is why tailoring beats volume even though volume feels more productive.
There is also a caveat worth carrying: applications-per-role varies enormously by industry and seniority. A figure quoted for "corporate jobs" in general is close to meaningless for any specific application you are about to send.
5. Skills-based hiring: the adverts changed before the hiring did
The most consequential structural change is supposed to be that employers have moved away from the degree as the primary filter, toward demonstrated skills. The advertising did change. The hiring mostly did not.
The clearest evidence on that gap is Skills-Based Hiring: The Long Road from Pronouncements to Practice (Burning Glass Institute and Harvard Business School, February 2024), which matched 769,980 roles over time at the same large employers. It found the share of postings requiring a college degree fell by 3.6% between 2014 and 2023 — a relative fall, not 3.6 percentage points, and confusing the two turns a small number into a large one. Within the roles that did drop the requirement, non-degree hiring rose by about 3.5 percentage points. Because that applies only to the small share of roles affected, the economy-wide effect is roughly 0.14 percentage points: about 97,000 workers a year out of 77 million hires. The report's own summary is that the opportunity promised by skills-based hiring "has borne out in not even 1 in 700 hires".
A second dataset measures the same thing from the opposite direction, and the contrast is the point. LinkedIn's Economic Graph research (March 2025, 56 countries) estimates that skills-based hiring would expand the available talent pool 6.1 times globally and 15.9 times in the US, with workers without bachelor's degrees gaining slightly more than those with them. That is an estimate of who could do the job, not of who gets hired — potential, not practice. LinkedIn also sells recruiting software, and it states explicitly that these figures are not comparable to its own 2023 report, so "skills-based hiring potential has shrunk" is a misreading of a change in method.
What this means for a CV is more useful than the optimistic version. The degree still does work in practice, so leaving education off is not the safe default it is sometimes presented as. But in the roles where the filter genuinely has loosened, demonstrated skill becomes the deciding evidence — which makes the Projects section and the achievement bullets carry more weight than they used to, and makes a link to work somebody can actually open worth more than a claim about capability.
6. Listing AI as a skill — and why most of it fails
"AI" is now the most name-dropped item on CVs and job postings alike, and most of it is noise.
Indeed's Hiring Lab analysed several hundred thousand postings (October 2025) and found that of the ones mentioning AI, 52% genuinely involved building or directly using AI models — but roughly a quarter gave no clear context at all, describing the company as "AI-first" and nothing more. Seventy-four per cent said only "AI"; just 2% named a specific tool like ChatGPT.
The same shallow tendency on the candidate side produces a line like "familiar with AI tools", which tells a reader nothing and reads as filler. What works is the same thing that works everywhere else on a CV: name the tool, name what you did with it, name what changed. "Used Claude and a Python script to summarise 400 customer interviews into a tagged themes list, cutting a two-week analysis to two days" is a skill. "AI-literate" is an adjective.
Worth knowing if you work outside technology: Lightcast found (August 2025) that 51% of postings requiring generative-AI skills are outside IT and computer science, up from a time when the large majority were technical roles. AI fluency is no longer a technology-sector skill.
7. What has not changed
Almost all of the format advice. That is the least interesting section of every trends article and the most reliable one.
Ladders' 2018 eye-tracking work — the same study the six-second myth comes from — found that layouts with clear section headings and visible job titles held attention, and that cluttered layouts, multiple columns, long sentences and keyword stuffing lost it. The one nuance usually dropped: the study notes keyword stuffing can help with automated screening, but that a document still has to be read by a person, so keywords have to appear in context to be worth anything.
And the two-page question has an answer in the source most people cite without reading. Ladders' own recommendation is that the two-page rule holds for experienced candidates — that an engaged reader will spend as much time on page two as page one, but that time on page two is strongly predicted by how compelling page one is. Length is not the variable. Whether page one earns the second page is. That is covered in more detail in CV format.
One genuine drift: Monster's 2026 research found 49% of US job seekers now use CVs longer than one page, against guidance that keeps pushing shorter. Candidates are voting for length. The research does not obviously support them.
8. What to do about it
- Make the top third carry the application. The evidence says it is skimmed fast; the response is not a shorter CV, it is a front-loaded one.
- Trade volume for fit. With around 33 applicants per opening and most candidates spending under 30 minutes tailoring, careful tailoring is the cheapest available edge.
- Be specific, because specificity now reads as authenticity. Concrete numbers, real constraints and named tools are harder to generate convincingly than adjectives — which is exactly why they work.
- Show work, not credentials. Given the shift to skills-based shortlisting, a project someone can click on outperforms a qualification they have to take on trust.
- Ignore the six seconds, and act as though it were true. The number is unreliable; the behaviour it implies is not.
Where Merit fits
The consistent theme above is that CVs fail for want of specifics — the number, the tool, the constraint — and that most people cannot supply them because nobody recorded them at the time. That is the gap Merit exists to close: it interviews you about the work behind the lines, keeps the facts you confirm in a private vault, and drafts a CV from that record rather than from a document you paste in. You confirm every fact before it is used, and you can export or delete everything. One interview and one CV are free; after that it is $20 a month.
If you would rather do it yourself, the worked method is in how to write a CV and it costs nothing.