Research methodology

SchoolJobs Research reports advertised school and education vacancies observed on SchoolJobs at a single observation timestamp. It describes what we observed, not the whole of UK education recruitment.

Most vacancies we observe are imported from source feeds rather than posted directly on SchoolJobs. A rise in observed vacancies can reflect our coverage as much as recruitment activity.

This is a current snapshot, not a trend. A single observation cannot show change over time, so no change, growth or decline figure is reported.

What counts as a vacancy

Exactly the public live-job population (public_live_jobs: eligible now, approved advertiser, not demo or test, listing kind 'job') with a role title. The same population feeds /insights, /press and public job search, so they can never disagree.

Duplicate handling

The canonical Phase 2 job dedupe authority. Rows marked duplicate, or pointing at a canonical_job_id, are excluded; Research runs no dedupe algorithm of its own.

Direct and imported vacancies

Vacancies are classified as direct (posted on SchoolJobs) or imported (ingested from a source feed). Both are counted, and the split is recorded on every snapshot.

Canonical categories

Only canonical dimensions are used: canonical role slug, canonical subject slug, canonical phase slug, canonical contract slug, canonical working pattern, canonical location nation/region, and accepted ownership attribution. Free-text location and free-text employer strings are never dimensions.

Missing and unmapped data

Unmapped rows are counted as unmapped and reported as coverage. They are never redistributed, imputed or silently dropped from a denominator without disclosure. A missing contract type is shown as "Not stated".

Publication thresholds

A dimension is publishable at >= 95% mapping coverage, publishable with an explicit unknown bucket and limitation note at >= 60%, and insufficient below 60%. A dimension with zero mapped rows is not_measured.

Privacy suppression

Metrics derived from anything other than already-public vacancy or directory evidence are suppression-sensitive: a cell is suppressed below 10 observations or fewer than 3 contributing organisations, and complementary cells are suppressed where differencing would reveal the suppressed count.

Comparing two observations

A change of more than 10 percentage points in the direct/imported split, or more than 25% in observed volume, qualifies or invalidates any comparison between two snapshots. Growth in observed vacancies is never reported as growth in UK recruitment.

Dates

Snapshots are stamped with the observation timestamp of the calculation. Vacancy-level dates use source_created_at where known, first_imported_at only as 'first observed by SchoolJobs', and never relabel first_imported_at as an employer publication date.

Salary

Phase 17 is the sole salary authority. Research references Phase 17 aggregates and snapshots by id and version; it never parses or recalculates salary.

Geography and UK nations

Nation comes from the school's official record; region from the vacancy's canonical location, else the school's location, else the school's official region. Nations without source coverage (currently Wales and Scotland) are shown as not yet covered, never as a number. Nation-specific phase and qualification structures are respected rather than mapped onto English phases.

1. 2 Nov 2026 · research-2026-11-v2

   • One live count everywhere: Research, /insights and /press now count exactly the vacancies listed live on SchoolJobs (the public live-job count), which excludes demo and test adverts. Previously Research used its own rules and could differ by a few hundred.
   • Nation now comes from the school's official record (the workplace school, else the advertising school). Previously it came from the vacancy's town, and about a third of towns had no nation recorded.
   • Region falls back to the school's own location, then the school's official region, when the vacancy's location has no region.
   • Vacancies with no contract type are shown as "Not stated". They are never guessed or redistributed.
   • Wales and Scotland are shown as "Not yet covered" instead of a number, because our current sources only list vacancies at schools in England.
   • Agency adverts from a retired source were closed on 31 Oct 2026, so the population is not comparable with August 2026. This version is marked breaking: no comparison is made across it.

2. 8 Aug 2026 · research-2026-08-v1

   • First Research methodology.

Research publishes aggregates only. No candidate, application, individual or single organisation's behaviour is part of any snapshot. A cohort is withheld below 10 observations or fewer than 3 contributing organisations, and a neighbouring cohort is withheld too where a withheld figure could otherwise be recovered by subtraction.

Small cohorts are withheld rather than published, because a very small group can identify the organisation or people behind it.

• Application outcome research — Most vacancies apply off-site, organisation ATS workflows differ, and the internal application dataset is far too small. Public application outcome research is deferred and classified restricted.
• Time to fill — A vacancy leaving the site does not prove a hire, and closure reasons rarely record one. Time to fill is not measurable and is not published.
• Shares of UK education recruitment — we observe advertised vacancies on SchoolJobs, which is not the whole of UK education recruitment, so no such share is published.

A published report is fixed to the observation that produced it. Corrections are published as a new revision with a correction note rather than by editing a figure in place. If a figure looks wrong, report it from the Research hub.