Which Requirements Are Real, and When to Apply Anyway

Where does the 100-percent rule actually come from?

Nowhere anyone can point to. The line gets repeated as a settled internal study, most often attributed to an internal Hewlett-Packard report, but no version of it that circulates online carries an author, a sample size, or a document anyone has opened. That doesn't make the observation false, only a folk statistic doing the work of a citation it doesn't have.

What is measurable, and comes from the employer side of the same process, tells a more useful story. In a survey of 2,275 senior leaders, 88 percent of employers said their own hiring system filters out qualified high-skills candidates who could do the job but don't match the exact criteria in the posting, rising to 94 percent for middle-skills roles ¹. That is employers describing their own filtering as unreliable, not a rejection rate and not proof about your resume specifically, but it is real, sourced, and it points the same direction as the folk rule without needing an invented number to make its case.

Read the responsibilities, not just the requirements block

A requirements list is usually written early, by someone assembling everything that might plausibly help, long before anyone knows who will apply. The responsibilities section, by contrast, is closer to what the hiring manager actually needs solved this quarter: the one to three things the role exists to get done. That section is where the real bar lives, and it's worth reading twice before you decide whether to apply.

A few ways to tell a real requirement from a wish-list item:

• A licence, clearance, or legal must-have is real. If the posting names a specific credential tied to the work itself, a gap there is a genuine reason to skip.
• A long list of tools and years-of-experience numbers is usually aspirational. Postings assembled from multiple contributors accumulate these without anyone checking whether the role needs all of them on day one.
• A responsibility repeated in different words across the posting is the real signal. If "owns the migration plan" and "leads the technical roadmap" both appear, that's the actual job.

Requirements have also gotten harder to weigh in the last two years specifically because AI-skill language has entered postings that never carried it before, sometimes as a genuine requirement and sometimes as boilerplate nobody thought hard about. In a June 2026 ZipRecruiter survey of over 1,000 hiring decision-makers, 74 percent called AI skills a strong advantage or an outright requirement, and half said they expect candidates to already be practical or advanced users ². That's employers describing their own expectation, not a measured outcome, and it means an AI-skills line in a posting deserves the same responsibilities-first read as everything else: is it tied to something the role does, or is it boilerplate. The employers writing these lines are being told the same thing from their side; how to write an AI-skills requirement that isn't legally vague is what a hiring team is coached toward, and a posting that follows that advice will name the task, not just the buzzword.

Don't add a proficiency you would fail the first follow-up on

Applying with a gap is different from closing that gap on paper before anyone asks about it, and the second move is the one that actually costs you. A resume line claiming daily AI use, or fluency with a tool you've opened twice, converts a promising screen into a short one the moment a specific question lands on it.

Employers are increasingly built to ask that question early and directly: verifying a claimed skill like "uses AI daily" is now something a short task drawn from the role can settle rather than taking a candidate's word for it, which is the employer-side version of exactly the risk this creates. A cleaner way to close a real gap without inventing one:

• Name what you've actually done, at the scale you've actually done it. A student project or a self-directed one counts if you can walk through it.
• Say what you're actively building toward, honestly, rather than claiming arrival. "Learning X on a current project" survives a follow-up; "proficient in X" does not if it isn't true yet.
• Spend the evidence on the two or three things the responsibilities section names, not on the whole checklist. A hiring manager who sees real depth on the actual work rarely holds a missing nice-to-have against you.

Experience aimed at the real work also outweighs a credential collected to paper over its absence. In a choice experiment where 543 US hiring managers and HR specialists picked between hypothetical candidate profiles, raising relevant work experience from zero to two years raised the probability of being chosen by 21.4 percentage points, more than a degree's format or the institution behind it moved it ³. That study tested degrees rather than AI certificates specifically, so the exact number doesn't transfer directly, but the shape of the finding does: real, relevant work moved the outcome more than the credential wrapped around it did. The unglamorous version of that finding is the same advice this article opened with: put your effort into evidence for what the role is actually for, not into rounding a checklist up to 100 percent.

Weigh a worked example against your own posting

A posting for a mid-level analyst role lists a master's degree, five years of experience, three named software tools and "AI-proficient." The responsibilities section names one thing three separate times in different words: owning a monthly reporting cycle end to end.

A candidate with two years of experience, no master's, and a working, self-taught habit of using an assistant to draft and check that kind of report has a real answer to the actual job, and a real gap on two line items that were never load-bearing. Apply, and lead with the reporting cycle, not with an apology for the degree. Two candidates who both meet every listed requirement but can't describe how they'd run that cycle are weaker applicants for this specific role than one who is short on the checklist but has clearly done the thing the posting exists to get done. The checklist was never the job description. The responsibilities section was.

See a sample report >