Current model: v3 · Last reviewed August 12, 2026
PrismCV scoring methodology
PrismCV turns the resume content its parser can extract into a diagnostic score and a prioritized set of checks. This page documents the current production model so you can interpret that output without treating it as a hiring forecast.
What the score is — and is not
The score is PrismCV's assessment of the text it parsed, the completeness and strength signals its rules detected, and—when supplied—the resume's relevance to a specific job description. It is not a score issued by an employer or ATS vendor. It does not predict whether you will receive an interview, reproduce every proprietary ATS configuration, or guarantee that a recruiter will see or advance an application.
Inputs and parsing
- PDF, DOCX, or TXT input, or pasted resume text. The public checker accepts files up to 5 MB; the account workspace accepts up to 10 MB.
- An optional job description for job-specific relevance checks.
- Structured fields extracted from the input, including contact information, experience, skills, summary, and education or certifications.
Extraction quality affects the analysis. If a field is visually present but is not parsed, the report may treat it as missing. That mismatch is useful diagnostic evidence, but it is not proof that every employer's parser will behave the same way.
How v3 is calculated
Within an analysis, four readiness gates must pass before a competitiveness score is shown: a name, a valid email address, at least one experience entry with a job title, and at least three skills. If an analyzed resume fails a gate, the report returns a zero and a direct fix for the missing input. The public route separately rejects an input that is too incomplete for meaningful analysis before this model runs.
- With a job description: keyword relevance contributes 55% and content strength contributes 45%.
- Without a job description: the score uses content strength and is capped at 70, because job-specific relevance cannot be evaluated.
- A completeness multiplier accounts for contact details, summary, experience, skills, and education or certifications. The final result is rounded and bounded to the applicable 0–100 or 0–70 range.
Keyword relevance includes job-description coverage, role-title alignment, hard skills, experience requirements when detectable, and skills demonstrated in context. Content strength includes measurable achievements, action verbs, specific language, skill context, summary quality, length, spelling signals, and related checks. The core public-checker score is rule-based; AI-assisted rewrites are suggestions you review, edit, accept, or reject.
Known limitations
- ATS vendors, employer configurations, parser versions, and recruiter workflows differ.
- Visual layout, tables, columns, icons, and unusual headings can extract differently by parser.
- Keyword analysis cannot determine whether a claim is true or whether experience is persuasive.
- Job descriptions may be incomplete, duplicated, or written differently from internal criteria.
- Scores can change when parsing, taxonomies, matching rules, or weights are revised.
Versioning and change control
Analysis responses include a scoring-version marker. A result that passes all four readiness gates currently returns version 3. A failed-gate result returns zero and retains the legacy version 2 marker in the current implementation. Historical scores from different versions should not be treated as directly comparable. This page dates the current method; selected product changes are recorded in the public changelog.