Why your resume's grade changed: inside the 2026 rubric

If you’re reading this because an email told you your saved resume’s grade changed, here’s the short version: your document didn’t move, the rubric did.

That’s disorienting when a grade feels like a fact. It isn’t. A grade is a reading against a rubric, and rubrics improve as evidence accumulates. RezScore has now graded more than 13 million resumes, and each round of grading tells us more about what actually separates a resume that gets a callback from one that doesn’t. The 2026 revision folds that learning back into the scoring model. The document you uploaded is exactly as it was. The bar it’s measured against isn’t.

What the rubric actually scores

Every resume is scored across five categories: parseability, structure, evidence, positioning, and credibility. Each runs 0 to 100, and together they combine into one letter grade on a 13-step scale from A+ down to F.

Parseability asks whether the format survives contact with the systems and people who screen it first: applicant tracking software, then a recruiter skimming for ten seconds. Structure asks whether the information is organized so a reader finds what they need without hunting. Positioning asks whether the resume tells a coherent story about what role you’re going for and why you fit it. Credibility asks whether the claims hold together, no contradictions, no gaps that raise questions.

Evidence asks something different: can you prove it. Numbers, scope, results you could defend in an interview if someone pushed back on them. Evidence carries the most weight of the five categories, and in the 2026 revision that weight went up.

Why evidence got heavier

This wasn’t a stylistic call. It’s what the data says separates resumes that work from resumes that don’t.

In the newest 70,000-resume study, 62.7 percent of A resumes contain at least one dollar or percent figure. Among F resumes, that number is 13.5 percent. The share climbs at every grade step between F and A, in a straight line. It is the cleanest signal in the dataset, and it is why evidence now pulls harder on the final grade than it used to.

The median resume in the study lands a C. Getting to an A puts you in the top 17 percent of resumes graded. What separates that top group isn’t better formatting or more polished language. It’s proof. A bullet that says “responsible for improving team performance” and a bullet that says “cut onboarding time from six weeks to eleven days for a twelve-person team” can describe the same work. Only one of them survives a rubric that’s been tuned on 70,000 resumes to reward what’s verifiable.

If your grade dropped and evidence is one of your weaker categories, that’s very likely the reason. The rubric didn’t get harsher across the board. It got sharper about the one thing that was already the biggest predictor of a strong resume, and it started counting that more.

What to do next

Start with your report card, not your gut. Sign in and pull up your saved resume’s category breakdown. Find your lowest score among the five.

If evidence is among the weak categories, start there before anything else. Take your top two or three bullets, the ones a recruiter would actually read, and rewrite them as proof instead of duties. Replace a description of what you were responsible for with a number: how much, how many, how fast, over what scope, compared to what baseline. You don’t need every bullet to carry a metric. You need the first few a reader sees to show, not tell.

Structure is worth one pass, and only one. Tighten section order, cut dead white space, make sure headings are consistent. That’s it. Fonts and margins are the weakest place to start, and if you find yourself adjusting kerning before you’ve touched a single bullet, you’re solving the wrong problem.

The grade that changed in your inbox is a more accurate reading of how your work reads on paper, measured against a rubric built on what 70,000 other resumes taught us about what gets results. It was never a verdict on the work itself. Sign in, look at the breakdown, and start with evidence if that’s where the score is lowest.