How hard the job market is in 2026
The market isn’t collapsing. It’s stuck. Employers aren’t laying people off, and they aren’t hiring either.
Indeed’s Hiring Lab calls it a low-hire, low-fire market, and says it is “starting to look less like a phase and more like the new normal”. In July 2026 the US hires rate was 3.2%, the lowest since February. The layoff rate was 1.0% and the quits rate 1.9%. People who have a job are staying in it, so fewer seats come free.
Openings are not the problem on paper. There were 7.3 million in July, but the number that matters to a job seeker is how many of those lead to a hire. Indeed’s June 2026 snapshot puts overall postings at 101 against a pre-pandemic baseline of 100, and says marketing, data analysis and software development postings have fallen “by roughly 30% or more” since February 2020. That is the field I’m in.
The wait is longer too. In July 2026, 1.77 million Americans had been unemployed for 27 weeks or more, 25.5% of everyone out of work.
India, where most of my applications went, looks different. Naukri’s own hiring index, the JobSpeak report for September 2026, shows white-collar hiring about 2% higher than a year earlier. That’s growth, but slow growth, and it hasn’t made my replies easier to come by.
The flood of AI applications
The second reason is on the other side of the desk. Greenhouse, one of the applicant tracking systems I applied through most, says about 254 people apply for each job ad, and that applications per recruiter are up 412%. Its chief executive says tools that apply to every Greenhouse job automatically can be bought for about $20, and that recruiters now get “hundreds or thousands of applications in just a day or two”. He calls it an AI doom loop. These are Greenhouse’s own platform figures, not an independent count.
Candidates say the same thing from the other side. In Greenhouse’s Future of Hiring Candidate Report, published 20 October 2026, 77% of US job seekers had used AI in their search, and 81% had been rejected in the past year without speaking to anyone. 60% had changed, or thought about changing, how they write so that they don’t sound like AI. The US sample was 1,200 people. A third of those who tried to beat automated screening had added hidden text aimed at an AI reviewer, which tells you how far this has gone.
A separate Greenhouse release from 15 October 2026 says 86% of US recruiters think at least half of the applications they see are AI-written, and that fewer than 7% of applicants get an interview. It gives no sample or method, so I treat those two numbers as a direction and not a measurement.
What 1,438 applications looked like
Between 29 July and 4 November 2026 I used Claude to screen postings, check each against my rules, write the answers and keep the count. The guide explains how the system works. Here’s what it produced.
Most of the work was saying no. I skipped 3,758 of 5,196 postings, 72%, before applying. Grouping the reasons logged for the skips by hand, about 28% were the wrong kind of role, 20% asked for more years than I have or sat at the wrong level, 10% needed a tool I haven’t used and won’t claim, 9% paid too little, and 8% were in places I couldn’t work from. The rest were duplicates or carry other reasons. I sent about 14.5 applications a day.
A further 398 postings stalled before I could finish. Roughly 110 stopped at an account, login or verification step, and about 80 at a CAPTCHA. Nobody mentions that cost when they say “just apply”.
The six kinds of response
“Heard back” hides a lot. These are the kinds of response I got, in rough order of how often.
Where the replies came from
The biggest difference was where I applied. Counting only applications at least a week old, so that recent ones aren’t counted as silence:
| Where | Applications | Replies | Reply rate |
|---|---|---|---|
| Greenhouse | 160 | 58 | 36.2% |
| Ashby | 87 | 27 | 31% |
| Workable | 41 | 5 | 12.2% |
| Wellfound | 43 | 5 | 11.6% |
| Lever | 57 | 5 | 8.8% |
| Indeed | 92 | 4 | 4.3% |
| Naukri | 423 | 4 | 0.9% |
Two things stand out. Company career pages that run on Greenhouse and Ashby answered roughly a third of the time, and the largest India job site answered less than once in a hundred, from by far the most applications.
But a reply isn’t progress. Of Greenhouse’s 58 replies, 57 were rejections. Of Ashby’s 27, none were still open. Naukri replied four times, and three of those were live conversations. The sites that answer the most are often answering with a quick no, and the ones that answer rarely are sometimes answering with a person. If I’d judged channels by reply rate alone, I’d have chosen the wrong one for getting interviews.
| Country | Applications | Reply rate | Note |
|---|---|---|---|
| United States | 82 | 29.3% | 23 of the 24 replies were rejections |
| Singapore | 17 | 41.2% | small sample |
| UAE | 27 | 7.4% | |
| India | 1106 | 5.1% |
Applying to US roles got replies far more often than applying inside India, but I wouldn’t read that as the US being friendlier. Most of the US roles were on company career pages, and those answer more often whatever the country, so place and platform are tangled together. Some of these samples are also small.
One finding surprised me. I’d assumed job titles with “Manager” in them would get answered less. They didn’t: 8.4% of 854 manager roles got a reply, against 9.1% of 372 others, and 15.8% of the 57 Head, Director or VP roles. Titles mattered far less than where I applied.
Why most applications get silence
I can’t see inside any company, so these are the explanations the evidence supports, strongest first.
1. Nobody has time to read them
With about 254 applicants a job and recruiter workloads up several times over, the average application is competing for seconds of attention. A small survey of 25 US recruiters, run by a résumé company, found 23 who said their system doesn’t auto-reject for formatting or content. The software ranks and sorts. Humans then run out of time. I’d treat that as a hypothesis rather than a result, but it fits what I saw.
2. The job may not be real
Greenhouse has reported that between 18% and 22% of postings in a quarter are never filled. I couldn’t open the original report, so take it as indicative. Higher figures circulate too, up to 47%, and I couldn’t trace those to anyone’s data.
3. A knockout question ended it
Location, work authorisation and notice period are often answered in a form field, and a wrong answer ends the application without a human involved. I answered every one of those honestly, and I suspect that cost me replies from some roles abroad. I’d do it again.
4. Someone else had a referral
CareerPlug’s 2024 data on more than 60,000 small and mid-sized US businesses found job boards produced 61% of applications but 42% of hires. Referrals were 2% of applicants and 11% of hires. Company career pages did about four times better per applicant than boards. An open posting is often the last place a company looks.
5. I applied where answers are rarest
Most of my applications went to roles in India, and the largest platform there answered less than 1% of the time. That was my own targeting mistake, and the numbers above show it.
What the evidence says works
The research is thin, and most of it is made by vendors. These are the points with something behind them.
I could find no independent study showing that auto-apply tools raise interview rates. Vendor blogs say they don’t, and my own data says volume bought me replies but very few conversations.
Am I part of the problem?
Yes, a little. 1,438 applications is 1,438 more things for someone to open, and I’d be a hypocrite to write about the flood from outside it.
The rules I set myself were these. Claude may only state facts from a file I wrote, and the answers were checked against it. It skips any role I can’t honestly take, which is why 72% were dropped. It never claims a tool I haven’t used, and it doesn’t tell an employer anything untrue about where I can work. I also don’t think anyone should send a thousand applications. If I did it again, I’d send far fewer, aim most of them at company career pages and spend the saved time on people. The guide shows how to build the careful version.
Questions people ask
What is a normal job application response rate?
There is no reliable single figure. In my own search, 127 of 1,438 applications (about 9%) got any reply, and 113 of those were rejections. Rates ranged from under 1% on one large job site to about a third on company career pages that use Greenhouse or Ashby. Surveys put the share of candidates rejected without speaking to anyone at 81% in the past year (Greenhouse, October 2026, 1,200 US respondents).
Why am I not hearing back from jobs I apply for?
Volume is the main reason. Greenhouse reports around 254 applicants per job ad and a 412% rise in applications per recruiter. Other causes are postings that were never going to be filled, knockout questions on location or work authorisation, and roles that go to referrals or internal candidates first.
Are ghost jobs real?
Yes, though the size is disputed. Greenhouse has reported that roughly 18 to 22% of postings in a quarter are never filled. Higher figures, such as 47%, circulate widely but I could not trace them to a primary source.
Do applicant tracking systems automatically reject résumés?
Rarely. In one survey of 25 US recruiters, 23 said their systems do not auto-reject for formatting or content. Knockout questions, such as work authorisation, are common. The bigger problem is that hundreds of applications compete for a recruiter’s time.
How many applications does it take to get an interview?
In my search it was about 144 applications per interview, call or assessment (about ten from 1,438). Greenhouse says fewer than 7% of applicants get an interview, but its release gives no method. Your number will depend heavily on seniority, field and where you apply from.
Does auto-apply software work?
I could find no independent study showing that auto-apply tools raise interview rates. In my own search, volume produced replies, mostly rejections, and very few conversations. Tailored, checked applications and referrals have better evidence behind them.
Is the job market bad in 2026?
Hiring is slow rather than collapsing. The US hires rate was 3.2% in July 2026 and the layoff rate was 1.0%, a pattern economists call low-hire, low-fire. Indeed’s data shows marketing postings roughly 30% or more below their February 2020 level.
Which applications got an answer fastest?
Rejections came quickly: the median was about five and a half days, and 14% arrived within a day. Replies that led to a conversation were slower and far rarer.
Method, sources and what I couldn’t verify
How I counted
Every application, skip and block went into a spreadsheet the day it happened. The totals here are generated from it by a script. An application counts as “heard back” if it was rejected, is live, or lapsed after a reply. Silence is everything else. Reply rates only include applications at least seven days old. “Interview, call or assessment” is my count of rows with one of those on record, about ten, and I’ll correct it here if the tracker changes. This is one person’s search, in one field, mostly in one country, so use it for direction and not as a forecast.
Sources
What I couldn’t verify, and so didn’t use
If you spot an error, or have a primary source that settles one of these, email me at adityamg96@gmail.com.