With two years of data the Explorer can say whether a subject’s entries appeared in both. It uses three deliberately flat labels — and avoids three tempting ones.
Consistent — the institution recorded entries in this subject in both 2023/24 and 2024/25.
Newly observed — entries recorded in 2024/25, none in 2023/24.
Previously observed — entries recorded in 2023/24, none in 2024/25.
Because the data cannot support those words. Newly observed covers a genuinely new course, a course that existed but recruited nobody the previous year, a cohort that finished in a different year, and a subject previously entered through a partner institution. All four look identical here.
Previously observed is the same in reverse. A school that taught German in both years but entered no candidates in 2024/25 appears exactly like one that stopped teaching it.
The comparison is limited to institutions present in both years’ data. Around 70 of the institutions in this Explorer appear in 2024/25 only, and a similar number appear in 2023/24 only.
Most are not new or closed schools. They are re-established ones: academy conversion, merger or a change of phase issues a new URN, so the same school on the same site appears as one record closing and another opening. Including them would manufacture a large number of false “newly observed” and “previously observed” results. Institution profiles for those schools say plainly that no two-year comparison is available.
A subject with a median entry cohort of two needs only one year without takers to vanish from the record, and one keen group to return. A subject with a median cohort of twenty produces entries almost regardless. So the subjects that move most between years are, with few exceptions, the subjects with the smallest cohorts — which says as much about how entries are recorded as about how provision changes.
Two years cannot separate a trend from ordinary variation. These labels describe what happened between two specific years, and nothing more. With more years the Explorer could distinguish consistent, regular and intermittent provision; with two, it cannot.
Ten A-level subject labels were renamed between the two releases — “Art and Design (Fine Art)” became “Fine Art”, “Computer Studies / Computing” became “Computer Science”. These are mapped so a rename does not masquerade as a change in provision. A small number of labels that could not be matched unambiguously are excluded from the comparison rather than guessed at.
Last checked 17 August 2026. Observed provision is based on recorded exam entries. An institution with no recorded entries in a subject may still have offered it — see what subject provision means.