Why Candidate matching Matters
Matching can help a team revisit large databases and recognise adjacent experience that a title search would miss. It can also create false confidence: profiles are incomplete, job descriptions contain inflated requirements, models learn from past decisions, and a high score may conceal the exclusion of capable people who use different language or have non-traditional careers.
Terms Recruiters Commonly Compare
Candidate search
Search retrieves profiles that meet a query or filter. Matching estimates alignment between a candidate record and a role; both depend on data quality and need human interpretation.
Candidate screening
Screening is the decision process used to assess whether someone should progress. Matching may help order or surface profiles, but it does not replace validated screening evidence or candidate conversation.
Recruitment Example
A system ranks support engineers for a cloud operations role. The recruiter reviews the explanation and finds that certification names drive most of the score while incident ownership and diagnostic work are underweighted. The team revises the role criteria, tests known relevant and irrelevant profiles, and uses the result as a review queue rather than an automatic rejection list.
Questions to ask about a match
A useful result supports a review decision; it does not hide how the role and candidate were represented. Before acting, the recruiter should be able to answer the following questions.
- Which evidence increased or reduced the result?
- Was absent information treated as a negative?
- Which version of the role requirements was used?
- Can an authorised reviewer find and correct a false exclusion?
- How are candidate records, inferred data, and feedback retained or reused?
Implementation Playbook
- Start with an approved role analysis that separates essential evidence, learnable skills, preferences, and prohibited considerations.
- Test the method on realistic profiles, including career breaks, adjacent titles, international terminology, incomplete data, and equivalent experience.
- Make missing information visibly different from evidence that a candidate does not meet a requirement.
- Give reviewers enough explanation to challenge a ranking and a route to find candidates the model missed.
- Monitor selection patterns, error reports, overrides, and changes in the job or labour market throughout use.
Common Mistakes
- Training or configuring a matcher against historical hiring outcomes without checking embedded bias.
- Converting every line of an unrealistic job description into a mandatory filter.
- Presenting a numerical match percentage as an objective measure of future performance.
- Rejecting candidates automatically because their CV lacks a keyword or parseable format.
Metrics to Track
Questions Recruiters Ask
What does a candidate match score mean?
Only the provider’s documented method can answer that. It may measure keyword overlap, rule satisfaction, similarity, or a model output. It is not automatically a probability of success, performance, or acceptance.
Can candidate matching reject applicants automatically?
The organisation must assess applicable law, impact, validation, notice, and review rights before any consequential automation. Operationally, an unexplained score is a weak basis for rejection and should not replace accountable evidence review.
How should recruiters test a matching tool?
Use representative job and candidate cases with known reasons for relevance, investigate both inclusions and exclusions, test data and accessibility failures, compare outcomes across groups where lawful, and repeat testing when the system or context changes.
Sources and Review
ATZ CRM Recruitment Editorial Review · Reviewed 2026-08-05
Put Candidate matching Into Practice with ATZ CRM
Use ATZ CRM to convert glossary concepts into daily recruiter workflows with sourcing pipelines, automation, scorecards, and reporting built for staffing and recruitment teams.