Three years ago, Andrew Stockwell, head of people at the software-buying company Vendr, had his hiring routine down to a science: post a listing, wait a few days, then review a few dozen applications — up to 100 if he was lucky. He read résumés, conducted interviews and passed along strong candidates to hiring managers.

Then everything changed. The process starts the same way, but within a day or two Stockwell is not looking at 100 candidates. Hundreds arrive, sometimes topping a thousand. A good chunk are total bogus — fake candidates. Many more appear written with AI, stretching the truth and making it difficult to distinguish one applicant from another.

Where that leaves him: "Paying my talent-acquisition professionals — who are high-paid, high-quality individuals — just to look through applications all day long."

How the friction disappeared

In 2026, job hunting has become as easy as online shopping. AI can produce tailored résumés and cover letters in seconds. Browser extensions autofill applications with no human involved. LinkedIn serves users dozens of jobs a day, many with the option to apply in just a few clicks.

This is the result of a deliberate design choice. For years, recruiters worked to remove every obstacle in front of applying. The logic was reasonable: the more qualified candidates apply, the better the chance of finding the right fit. After the pandemic, employers struggling to fill open roles pushed that effort into overdrive.

Ophir Samson, head of voice AI at the recruiting platform Greenhouse, describes the reversal: "A year ago, every recruiter would tell me: We want to make it as easy as possible to apply for jobs," offering candidates a seamless experience. "What they got was 2,000 applicants in 24 hours for a job. That is a shitty experience for everyone."

How we got here

Today's picture has a recent history, and that history ran in the opposite direction. Jane Curran, chief transformation officer at the real estate firm JLL, describes the post-pandemic period: "Everyone was job hopping, because you literally could have three offers in an afternoon." Now, she says, "it is the polar opposite."

In that period, making things easy was a necessity. Employers could not find candidates, so every obstacle in front of applying meant a candidate lost. One-click applications, autofilled forms and fast in-platform submission spread as an answer to that need.

The problem is that the infrastructure stayed in place when conditions reversed. A system designed for a candidate shortage kept running at the same speed through a candidate surplus. The tools were built to solve a problem; when the problem disappeared, they became one.

Both sides lose

The strange part of the resulting picture is that nobody wins. Good applicants cannot stand out, while bad candidates slip through the cracks. Recruiters face the tortuous task of judging thousands of candidates every day. Instead of taking time to search for dream employees, they spend hours in applicant tracking systems tweaking filters to determine whose applications are worth a 30-second skim.

Tessa White, who worked in HR for two decades, sums it up: "We're currently in a place where employers are complaining that they can't find good people, and people are complaining that they can't find jobs." For her, this is the result of the quest for speed and ease: "Every time we seem to strive for efficiency, we seem to give up quality."

An advice economy grew around the broken process

There is indirect evidence of how dysfunctional the process has become: an advice economy formed around it. Tessa White, who left corporate life in 2018 after two decades as an HR executive, began posting about the job market during the pandemic and built a following of 800,000.

That number is itself a diagnosis. When people have to watch content to learn how to apply for a job, the process has stopped being self-explanatory. Thirty years ago applying was too obvious to require advice: see the listing, send the résumé, wait for a reply.

What the numbers say

Two trends are running at once in the background:

  • Openings contracted. After peaking at a record 12.3 million in March 2022, openings declined for two years. Since mid-2024 they have hovered around 7 million.
  • Applications rose. LinkedIn says submissions per applicant on the platform are up 46 percent compared with February 2020. Since ChatGPT launched, applications are up 22 percent.

So the pool shrank while the number of applications thrown into it grew. An industry sprang up to make that easier: some companies promise applicants they will submit "10x as many applications with less effort than one manual application." If you wanted to, you could apply for dozens of jobs a day.

LinkedIn has added limits against the pile-up and is rolling out a feature that tells seemingly underqualified applicants they probably are not a good fit, suggesting other jobs instead.

What recruiters now want: difficulty

According to Samson, recruiters are telling him something different now: "Actually, we kind of want friction. The friction is good. We want to make it harder."

That reverses a decade-long trend. Several approaches are being tried:

  • AI first-round interviews. Companies have used automated systems to screen résumés for at least a decade; now agents conduct first-round interviews.
  • Knock-out questions. Strict criteria that cut down the pool.
  • Early skills testing. Moving assessment to the front of the process.

Samson says these are not friction for friction's sake, but acknowledges a bonus: they weed out bots and halfhearted candidates. One line he reports shows the scale of the problem: "We frequently hear from recruiters: They have 1,000 applications, but they know that only 30 of them are serious."

The picture varies by sector

It is worth noting the problem is not the same everywhere. In skilled trades, where employer demand consistently exceeds employee supply, an avalanche of résumés is welcome; there, plentiful applications are a blessing rather than a problem.

The squeeze happens in white-collar knowledge work where, as JLL's Curran puts it, not enough new roles are entering the market. That distinction matters in practice: for a job seeker the real question may not be "is my application good" but "which market am I in." In a field where supply is scarce a mediocre application draws a reply, while in a saturated one even an excellent application can vanish unseen.

Why this may not be enough

White is not convinced these changes fix the underlying problem. In her view, the frictionless résumé drop has rendered the first round of the application process essentially worthless. New AI screening layers may do nothing more than process that worthless round faster.

The real issue is structural rather than technical. On one side, a candidate can produce a tailored application in seconds. On the other, an employer can scan thousands of applications in seconds. Both sides accelerate and the information between them never increases — because what got faster was production and elimination, not evaluation.

The recursive problem: AI against AI

The strangest consequence here is that the same technology sits at both ends of the problem. The candidate writes the application with a model. The employer screens it with a model. The amount of text a human actually reads keeps shrinking.

This arrangement has two known consequences. First, each side starts optimizing against the other's tool: applicants format résumés to pass the screening system, and employers tighten criteria to catch that formatting. The contest drifts away from measuring fit and toward beating the tool.

Second, measurement error grows. An application written with a model looks polished even when it stretches the truth; screening done with a model can mistake polish for quality. A tool that writes well ends up confused with a candidate who works well.

What can be done

The practical conclusions that follow from this account: on the candidate side, sending a hundred applications is not more effective than sending five good ones, because quantity is not what separates you. An automatically generated text that resembles every other becomes precisely part of the indistinguishable pile. On the employer side, the first round needs to be made meaningful rather than fast; a knock-out question or an early skills test yields more information than a 30-second résumé skim.

The lesson is not confined to job applications. The same pattern repeats in any system that makes an action nearly free: when writing a review gets easy, fake reviews multiply; when sending mail costs nothing, spam explodes. What disappears each time is not an obstacle but the information that obstacle carried.

The broader lesson is this: removing friction from a process does not improve it, it only makes it cheaper. Friction was often not a cost but a signal — it showed how much time an applicant was willing to spend on that job. When the signal disappeared, so did the thing doing the filtering.