Screening
Don't Let a Match Percentage Decide Where You Apply
The match percentage next to a job posting is the job board's comparison of your profile text against the posting text. In most cases the employer never sees it, never configured it, and never uses it, and the method behind it isn't published. It rewards vocabulary overlap, so it peaks where a resume echoes the posting, a move that stopped separating anyone once every applicant could do it. Treat it as one weak input, never a gate, and check instead whether you can evidence the role's core work.
The takeA career change makes the score actively misleading rather than merely unhelpful, because the score measures word overlap and a career changer's whole pitch is that the words differ while the underlying capability transfers. The platforms showing this number are under no obligation to explain it, and a low score that keeps you scrolling and applying costs them nothing, whatever it costs your search.
Where Olive fits
Open a role and see what the work shows
No screen can tell which resume a model wrote, and Olive doesn't try: if an employer sends you an Olive assessment, it is a 40-to-60-minute assignment done openly with an AI assistant, and you receive the same six-finding report the employer does, free.
Rank your shortlistName Who Computes That Number, and Why
The number is the platform's own comparison between your stored profile and the posting's text, run by the platform, for the platform's purposes. It is not a copy of anything running inside the employer's hiring system, and it is not something most employers asked for, configured, or would recognize if shown.
That matters because a score with no documented method and no employer sign-off cannot be treated as a verdict. It behaves like a retention feature more than a hiring signal: something that keeps a user scrolling and applying inside the platform, which is a reasonable thing for a job board to want and a poor thing for an applicant to make decisions on.
No major job board publishes the weighting behind the number, which of your profile fields it reads, or how often it's recalculated. That silence is itself informative. A figure a company is confident in tends to come with a page explaining it. A figure with no such page can only be glanced at, which is a reason to weigh it lightly.
Why the Score Is Weakest for a Career Change
The score is weakest for a career change because word overlap is its whole mechanism, and a career change is the one case where the words differ while the work carries over. Any applicant can now mirror a posting's phrasing in seconds with a free tool, so the resemblance the score measures is cheap to produce and thin as evidence.
Employers themselves report the cost of the opposite failure directly. In a Harvard Business School and Accenture survey of 2,275 senior leaders in Germany, the UK and the US, fielded in early 2020, 88% of the employers running a filtering system said qualified high-skills candidates were vetted out for not matching the exact criteria set in the job description.1 That is employers describing a problem with their own process, not a candidate defect. A job board's percentage is a cruder relative of the same idea, computed outside the employer's system and without its criteria.
The gap is in the vocabulary, not the capability. The title and the surrounding words genuinely differ between an old field and a new one, even when the underlying capability transfers cleanly. A score built on word overlap has no way to see that transfer, so it reads as a mismatch precisely in the cases where a human reviewer, given the chance to look, might see the opposite.
This is not an argument that the vocabulary never matters. A resume that never once names the tools, systems, or methods the new field actually uses will read as unfamiliar to a person too, not only to a score. The distinction worth holding onto is that a person can weigh unfamiliar phrasing against evidence of real capability. A percentage cannot weigh anything.
Ask the Two Questions the Score Can't Answer
Two questions are worth more than the percentage when you're deciding where an hour goes, and you can check both of them yourself. Can you evidence the role's core responsibilities with something specific from your actual background, a project, a number, a decision you made and its result? And is there a real human path into the role, a referral, a named hiring manager, an application that reaches a person rather than disappearing into a queue?
A low score can be flagging a title mismatch rather than a capability mismatch, and that part is addressable: rewrite the summary around the work rather than the job titles, and let the evidence carry the argument the vocabulary can't. Naming the transferable task directly, instead of the old title that carried it, is the part of the gap you can do something about on the page.
A very high score is its own flag in the other direction, because it mostly registers that the document is echoing the posting closely, which is worth a second look at whether real evidence is on the page or the resume has quietly become a mirror. Neither number answers either question.
Don't Let the Number Be the Whole Decision
The software layer matters less than the number implies, and the score is approximating the wrong layer anyway. In a correspondence experiment that sent more than 83,000 fictitious applications to entry-level jobs at 108 large US employers, which third-party application system a firm used explained only 0.1% of the variation in racial contact gaps between firms, and the authors conclude that screening algorithms are unlikely to drive the differences between employers.2 The large differences sat between companies, not between vendors.
Inside a real hiring system the keyword tool is often something a person drives. Greenhouse documents its Talent Filtering as a search a recruiter runs over the full text of resumes and internal notes, where a term has to match exactly to appear in the results.3 That is one vendor's product and other systems handle matching differently, but none of it is what the number on the listing measures, and none of it is something a job board can see.
In a discrete-choice experiment with 543 US hiring managers and HR specialists, fielded in December 2025, moving a candidate's relevant experience from zero to two years raised the probability of selection by 21.4 percentage points.4 That study measured a stated choice between hypothetical candidate profiles, not a real hire, so treat the size of the effect as directional and not as a forecast of your own odds. What it does settle is the priority question. Relevant, evidenced experience has a measured effect on a hiring decision, and no comparable measurement has ever been published for a posting-overlap score.
Put the hour a low score would have cost you toward whether you meet enough of the posting to make the case, and read what employers themselves are told about checking whether their own resume screen rejects the wrong people. The percentage on the listing is an input the hiring side usually never supplied and never sees, which is a reason to weigh it last rather than first.
Common questions
Does the employer ever see the match percentage I'm shown?
Usually not. The score is typically computed and displayed by the job board itself, comparing your stored profile to the posting text, and most employers never configure, see, or use it as part of their own hiring process.
Is a 40% match score a reason not to apply?
Not by itself. The score tracks word overlap with the posting, which is weak evidence of fit, especially if your background uses different terms for genuinely equivalent work. Check whether you can evidence the core responsibilities before the number decides anything.
Why is my match score so low when I'm changing careers?
The score rewards vocabulary overlap, and a career change is defined by the vocabulary differing while the underlying capability transfers. That is exactly the case the score is worst at reading, not a sign the transfer itself is weak.
Should I rewrite my resume just to raise the match score?
Rewriting toward the posting's exact wording is quick, and it stops distinguishing you once every applicant can do it in seconds. Rewriting toward specific, checkable evidence of the actual work gives a human reader something to weigh, which is the better use of the same hour.
What should I actually check before deciding whether to apply?
Whether you can name something specific and true that evidences the role's core responsibilities, and whether there's a real path to a human reader rather than only a queue. Both are things you can check yourself before applying, and the percentage on the listing isn't.
References
- 1. Hidden Workers: Untapped Talent - How leaders can improve hiring practices to uncover missed talent pools, close skills gaps, and improve diversity hbs.edu Supports the claim that employers themselves report exact-match filtering losing qualified candidates.
- 2. Systemic Discrimination Among Large U.S. Employers (NBER Working Paper No. 29053, revised May 2022) nber.org Supports the claim that which ATS vendor a company uses explains very little of who gets contacted.
- 3. Talent Filtering (Greenhouse Support - Recruiting) support.greenhouse.io Supports the description of a real keyword tool as a recruiter-run search rather than an automatic score.
- 4. Examining Employers' Perceptions of Online Credentials: A Discrete Choice Experiment sr.ithaka.org Supports the claim that relevant experience has a large, measured effect on hiring managers' selection decisions.
4 sources, numbered by first appearance. How Olive sources claims
General guidance for hiring teams. What works at one company and one volume may not transfer to yours.
Olive assesses how a person works with AI. It does not detect AI-written documents, and it never produces a score, a ranking, or a match percentage for a person. Candidates read the same report the employer reads.