Open postings by role family

8 rows in release 2026-08-07.11.

• Figures describe an observed sample of open postings, not the whole job market.
• Permitted claim class for this release: SNAPSHOT_ONLY.

Columns

role_family

Role grouping assigned from the posting title by a fixed set of patterns. UNMATCHED means the title did not clearly belong to any family, not that the role is unusual. The classifier is deliberately conservative, so UNMATCHED is common.

open_postings

Postings in this family observed as open at the most recent complete measurement. A point-in-time count, not a flow: it does not say how many roles opened or closed.

total_open_postings

All open postings observed, whether or not they could be placed in a family. This is the denominator for pct_of_observed_postings.

classified_postings

Open postings the classifier could place in any family. The difference between this and total_open_postings is the part of the sample the taxonomy does not cover.

pct_of_observed_postings

Share of the whole observed population, not of the classified subset. Shares therefore do not sum to 100 — dividing by the classified subset would imply the taxonomy covered everything.

classification_coverage_pct

Share of observed postings the classifier could place at all. A family count means very little without this number next to it.

postings_with_salary

Postings in this family that disclosed pay in a readable form.

salary_disclosure_pct

Share of postings disclosing pay. Disclosure is not random — it correlates with jurisdiction and employer size — so it is a property of the sample, not of the market. Null where the family has too few postings for a percentage to mean anything.

postings_with_location

Postings whose location resolved to a canonical place. Unresolved locations are excluded from location metrics rather than filed under a guess.

postings_remote

Postings that stated they are remote.

postings_hybrid

Postings that stated they are hybrid.

postings_onsite

Postings that stated they are on-site.

postings_remote_unknown

Postings whose work arrangement was not stated. Not evidence of on-site work; most postings simply do not say.

supports_comparison

False when the row's base is too small to compare against another row or to read as a trend. The count itself is still published and still exact — a small family is a small family, not an unreliable measurement. What is withheld on a small base is the rate computed from it.

small_sample_caveat

Plain-language statement of what a small row may and may not be used for.

computed_at

When this row was calculated, which is the release build time rather than when the underlying postings were observed. For the period observed, see earliest_observation_at and latest_observation_at in the observation window dataset.