Skip to content

refresh

Jobs
Blog
Pricing

Resume evidence research method

See the proposed inputs, sampling rules, exclusions, privacy safeguards, and reporting limits for resume evidence research method. No aggregate claim is published until the required evidence and review threshold is met.

Write evidence-backed resume bullets

Methodology explains a preparation signal; it does not turn the result into a hiring prediction.

Measure whether resume claims carry evidence, not whether they sound impressive

Use occupation tasks and competency units to organise the work. Count a claim only when the source record supports the work, ownership, context, and a result or verification point.

Which resume claims can a reader trace back to a specific piece of work evidence?

Verified public facts

31.0Database release2026-08-27

O*NET 31.0 organizes worker and job information across skills, knowledge, abilities, tasks, work activities, work context, occupations, and related domains.

How to use it
A defensible vocabulary for role requirements and work evidence categories.
What this data cannot establish
That every task or skill applies to a particular opening or candidate.
2026-08-27

Korea’s NCS public data describes work through the competency units required to perform it, providing a Korean reference alongside occupation data built for the U.S. labour market.

How to use it
Korean role and competency framing without translating U.S. occupation labels literally.
What this data cannot establish
A vacancy count, hiring trend, salary, or employer-specific requirement.

Synthetic evidence example

Synthetic example for interpretation practice—not an observed Refresh result.

Scenario input
Example only: 14 of 25 reviewed experience bullets name an artifact and a verification method.
Calculation and interpretation
The unit is a bullet, not a person, and the review covers one resume version.
Decision this supports
Use the gaps to revise unsupported bullets
Comparability limit
Do not infer that 56% of applicants lack evidence.
Research protocol and unpublished result fields

How the data is handled

  1. Trace each claim

    Link each analysed resume passage to the work record, file, commit, report, or other source that supports it.

  2. Separate missing proof from weak prose

    A sentence can be well written and unsupported, or rough and well supported. Report those conditions separately.

Definitions that travel with the result

Evidence-linked claim
A resume claim linked to a source record that supports the work, ownership, date or period, and stated outcome.
Observable outcome
A shipped change, decision, artifact, resolved issue, measured movement, or documented follow-up. It need not be a percentage.

Results that still require observed data

Evidence coverage

How many material resume claims link to a source record, and which evidence dimensions are present?

Awaiting observed data
Required result fields
  • Resume identifier

    The versioned resume used in the analysis.

    string · required
  • Evidence identifier

    The source work record supporting a resume or application claim.

    string · required
  • Language

    The language of the source document or text sample.

    string · required
  • Category

    A documented category with a stable definition.

    string · required
  • Count

    The numerator after deduplication and exclusions.

    number · required
  • Denominator

    The eligible records used to calculate a share or rate.

    number · required
  • Metric definition

    The exact inclusion, exclusion, and calculation rule.

    string · required
Publish only when
Publish only after material-claim selection, confidential-data handling, and evidence-category rules are documented.
Awaiting observed data
No evidence percentage is shown until resume passages can be linked to source records.

What this data cannot establish

  • A missing link may mean the source was not provided; it does not prove the work never happened.
  • Confidential work may require redacted or higher-level evidence rather than public disclosure.

How to use it

  • Fix unsupported or conflicting claims before optimising tone or keywords.

Official data sources

  • O*NET DatabaseU.S. Department of Labor, Employment and Training Administration2026-08-27
    Coverage
    Worker and job characteristics, including occupations, tasks, work activities, skills, knowledge, and work context.
    Boundary
    O*NET describes occupations in the U.S. economy. It does not reproduce one employer’s job description or prove that a candidate has a skill.
  • Korea National Competency Standards dataHuman Resources Development Service of Korea2026-08-27
    Coverage
    Korean competency standards and competency-unit information used to describe work by the capabilities required to perform it.
    Boundary
    NCS is a competency standard, not a live count of vacancies, resumes, interviews, or hiring outcomes.

Career evidence research methods

Review proposed samples, exclusions, privacy safeguards, and reporting thresholds before refresh.cv publishes aggregate career findings.

View page

Interview evidence research method

See the proposed inputs, sampling rules, exclusions, privacy safeguards, and reporting limits for interview evidence research method. No aggregate claim is published until the required evidence and review threshold is met.

View page

Frequently asked questions

Read the result with its method and limits.

Use the relevant methodology before deciding which finding to act on.

Write evidence-backed resume bullets