AI job matching is software that compares your profile with open job postings and decides which ones you see, and in what order. Every tool that offers it has to settle two things: where the postings come from, and how they are ranked. This post gives ApplyEze's answers, in detail. Its board ranks with rules, not AI: ApplyEze does not rank by text similarity. It ranks on your target titles and the requirements it reads out of each posting. We make ApplyEze, so this is a product post, but most of what follows is something you can check against the board yourself: which employers it reads, how it keeps closed jobs out, how your own list of roles sets the order, and why it will show you a zero rather than a flattering number.
What AI job matching usually means
"AI matching" covers several different designs, and vendors describe theirs in their own words:
| Design | How one vendor describes it |
|---|---|
| Job titles plus filters | JobCopilot: "Quickly select the job titles you want and use various filters," with a setting to "Adjust how selective your copilot should be with job matching" (JobCopilot) |
| Match score with an explanation | Tsenta: it "surfaces roles that fit your résumé and preferences, including location, salary, experience level, and role family," and "Matches include an explanation" (Tsenta) |
| Vectors that learn from feedback | ApplyPass: "AI/ML based vector based matching from your resume and live job descriptions" that "retrains nightly on your likes and dislikes" (ApplyPass) |
Each design has a real strength. Filters are predictable, explanations let you argue with a match, and learning from your likes adapts to taste you cannot put into words. The complaint users raise most often is the same across designs: roles that do not fit. A Tsenta App Store review from September 2026 says only a small fraction of the jobs shown matched what the reviewer wanted, a Scale.jobs Trustpilot review complains of mismatched jobs, and a FastApply Trustpilot review says its matching searched only by job title. Single reviews prove little on their own, but they point at the two places matching goes wrong: a thin or stale pool of jobs, and a title match that catches the wrong role. Those are the two places this post spends the most time.
Where the jobs come from
ApplyEze watches the career boards of more than 100,000 employers directly: 109,237 enabled boards on 5 November 2026. It reads the boards employers post to themselves, on applicant tracking systems such as Greenhouse, Workday, Lever, Ashby and SmartRecruiters, rather than relying only on aggregators that copy listings from them. New postings keep being matched to your profile through the day.
On the same day the pool held about 900,000 active jobs. ApplyEze applies automatically on systems holding roughly 62% of them with a connected application inbox, and roughly 23% without one. Those systems are Greenhouse, Lever, Ashby, Workday, Workable, SmartRecruiters, Oracle Recruiting Cloud, JazzHR, BambooHR, Rippling, Pinpoint and Recruitee (Workday and Oracle are the two that need the inbox). Those numbers change daily, in both directions.
Each posting is then parsed for what a match needs: the role, seniority, remote policy and similar fields. Our own rules read it first. When the rules cannot read a posting well enough, an AI model reads it on some application systems.
Keeping the pool fresh
A big pool is only useful if it is current. We drop closed and stale postings, so the queue is built from roles that are still open.
Reading the employer's own board is what makes that possible. When a role comes down from the board an employer runs, there is no copy elsewhere for us to keep showing you. That matters more than it sounds. Greenhouse's recruiting benchmarks, drawn from over 6,000 companies and over 640 million applications, put the average at 244 applications per job in 2025, up from 116 in 2022, and time to fill at 59.67 days in 2025, up from 43.64. A posting can sit open for weeks while a pile of applications grows behind it. Arriving while the role is new, and not after it has closed, is the part software can control.
What freshness cannot fix is a posting the employer leaves up without hiring for it. Our guide to ghost jobs covers the signals to check.
Your target roles, in your order
You tell ApplyEze the roles you want, in your own order of priority. The board then ranks matches by those target roles first and by fit second, not newest first.
Say your list reads:
- Data Analyst
- Business Intelligence Analyst
- Operations Analyst
Every strong Data Analyst match ranks above every strong Business Intelligence Analyst match, which ranks above every Operations Analyst match. Within each tier, fit decides the order: a score from your role, skills, industry, level, education, experience and pay. Hard filters run before any of this: work authorization, salary floor, blocked companies and location. Two more rules keep the order honest. A posting whose title matches one of your roles but fits you poorly does not jump the queue on its title alone. And postings that match none of your roles rank below the ones that do, however recently they were posted.
The practical effect: the order of your list is a real instruction. Put the role you most want first, and keep the list to roles you can plausibly win. Our job search strategy post covers how to choose them.
Matching on the real role
Title matching is where a lot of matchers go wrong, because job titles are full of words that sound specific and are not. ApplyEze matches on the role you list, not on a stray word in the title. In practice that means:
- Generic words do not anchor a match. "Analyst", "engineer" or "manager" appear in nearly every title, so a match needs the word that says which kind, such as "data" in Data Analyst. A technology qualifier like "IT" does not carry a match on its own either.
- Qualifiers must agree. A Technology Risk role does not match a Credit Risk posting because both say "risk".
- Acronyms are understood. GRC, TPRM and similar abbreviations match their spelled-out forms.
- Slash titles are split. A target like "IT Auditor / Senior IT Auditor" is read as two roles.
- Different jobs that share words stay apart. An Automation Controls Engineer is not a Controls Automation Analyst.
- Level is respected. Director, head-of and internship titles do not match a role that does not name them, so an analyst is not shown a director opening just because the domain words line up.
None of this is visible as a feature. What you notice is its absence: fewer postings in your queue that make you wonder why they are there.
Location
Location works the way most people search. You can set a metro or a radius, and add several locations if you would move to more than one place. Remote jobs count under every filter, so setting Chicago does not hide a remote role you could do from Chicago. A Strict toggle limits the board to local roles only, for when you want an office or cannot work remote.
An honest match count
The first check you can do yourself happens at sign-up. Onboarding runs the matcher against your profile and shows your real match count, even when it is zero. A zero gets its own screen, headed "No strong matches yet", rather than a padded number dressed up as partial success.
Zero is a useful answer. It tells you your roles, seniority or locations are narrower than the open market, before you spend credits on applications nobody would read. The usual fixes are adding a neighbouring role to your list, widening a radius, or allowing remote work.
What matching does not do
Being specific about the limits is part of describing the product honestly.
- It does not predict interviews. Fit is a ranking signal, not a forecast of whether an employer will call. ApplyEze publishes no interview or offer rate for people using it on their own.
- It does not reach every job. ApplyEze applies automatically on systems holding roughly 62% of the pool with a connected inbox, and roughly 23% without one. If you sign up on your own, postings on systems such as iCIMS, Taleo, SuccessFactors or LinkedIn Easy Apply are not auto-applied.
- It does not replace referrals. In a Federal Reserve Bank of New York study of one large US employer's hiring from 2000 to 2011, referred applicants were more than twice as likely to be interviewed as other applicants. The data is old and from one firm, so read it as a direction rather than a rate. Matching helps with the cold applications; it is not a substitute for knowing someone.
- It does not judge whether a posting is genuine. It can drop roles that come down. It cannot see intent behind a role that stays up.
Where ApplyEze fits
Matching is the first step of a longer run. For each posting you queue, an AI model reads that posting and your real history and writes a resume for it, with a cover letter when the form asks. ApplyEze then completes the application in a visible browser on your own machine, with a dry-run mode that fills the form without clicking Submit. Duplicate detection and a posting lock are built to stop double applications. New accounts get 50 free credits, once per email address with no card, and a credit is used only when an application is confirmed as submitted. Beyond those, plans are listed at $19 a month for 700 applications and $99 a month for 3,000. The full pipeline, from board to submitted form, is in job search automation.