Parsing, Screening, Matching — Which One Do You Need?
Vendors use the three terms interchangeably; the three jobs are not interchangeable. A two-minute disambiguation that will save you from buying the wrong tool.
Resume parsing extracts structured data from a document. Candidate matching compares that data against a specific job and scores the fit. Resume screening is the decision layer that uses parsed data and match scores to decide who moves forward. Parse → match → screen — in that order.
The three jobs side by side
| Input | Output | Question it answers | |
|---|---|---|---|
| Parsing | Resume file | Structured JSON — 50+ fields | “What does this resume say?” |
| Matching | Resume + job description | Scores — 20 dimensions, 0–100, with reasons | “How well does this candidate fit this job?” |
| Screening | Scored candidate pool | A shortlist and a decision trail | “Who should a human talk to first?” |
Parsing: the data layer
Parsing knows nothing about your open role. It answers one question (what is in this document) and answers it identically whether the candidate applied for engineering or accounting. That job-independence is its strength: parse once, reuse the record for every search, match and report that follows. Details in what is resume parsing; implementation on the Parsing API page.
Matching: the judgment layer
Matching is always relative to one job. The same candidate scores 85 against one posting and 40 against another, and both numbers are correct. Modern matching is multi-dimensional — the SharpAPI match score returns 20 dimensions with written explanations rather than one opaque percentage, because a recruiter who cannot explain a ranking cannot defend it.
Screening: the decision layer
Screening is what your organization does with the scores: rank the pool, apply knockout criteria, pick who gets a phone call. It is the only layer with regulatory weight — bias-audit laws target screening decisions, not JSON extraction — which is why explainable inputs matter. The screening API pattern wires the other two layers into this one.
Which do you need?
- Building an ATS or job board? Start with parsing — it fills your candidate records and powers search.
- Recruiters drowning in applicants? Add matching — ranked shortlists with reasons.
- Automating the first-pass decision? That is screening — and it demands the explainability and PII handling described in the compliance guide.
Questions, answered
Which comes first — parsing, screening or matching?
Parsing always runs first: it creates the data everything else consumes. Matching scores parsed candidates against a specific job; screening is the decision layer that uses those scores to shortlist.
Do I need all three for an ATS?
A minimal ATS needs parsing alone — it fills the candidate record. Add match scoring when recruiters face more applicants than they can read; the screening decision then rides on ranked, explained scores.
Is resume screening the same as resume filtering?
Filtering is the crudest form of screening: knockout rules on keywords or fields. Modern screening ranks by multi-dimensional match scores instead of rejecting on missing keywords — fewer good candidates lost.