Cohorts and students

Platform data, not official statistics. The numbers describe only students and graduates registered on Gradly and do not replace the data of the ministry or the statistics authorities. Cohort methodology — version a1.

A cohort is a group of students or graduates of one university who study or studied in the same field, at the same level and with the same graduation year. Every figure for universities and government bodies is computed by cohort, so this page explains who belongs to a cohort and how people are counted.

Who counts as a student of a university

• A student of a university is a user whose profile has an education entry at that university. A branch without its own organization on the Platform counts towards its parent university; a branch with its own organization is counted separately and sees only itself.
• Entries with the status "studying" and "graduated" are counted. Entries with the status "left without graduating" are not part of the student and graduate counts.
• Suspended accounts, deleted accounts and accounts scheduled for deletion are not counted.

How people are counted

• In a university's cohorts the unit is the education entry, that is, the programme. A person on two programmes (for example, a bachelor's graduate now in a master's) belongs to both cohorts.
• National totals of people count each person once, even with several programmes.
• So the sum of a university's cohorts can be larger than the number of people at that university.

Confirmed and declared students

Affiliation with a university is either confirmed or declared. It is confirmed when one of these sources confirmed it, from the strongest to the weaker:

1. a match with the university's list;
2. a code sent to a university email address (a platform administrator first confirms the email domain; public email services are not accepted);
3. a one-time career centre code;
4. a request confirmed by university staff.

It is declared when the student chose the university themselves or came through the university's general invitation link. Changing the university, level or field in the profile removes the confirmation.

Every breakdown comes in two variants: "everyone who named the university" and "confirmed only". The second is more accurate, the first more complete.

Breakdowns

Figures are available only for predefined breakdowns:

• the university as a whole;
• university × field;
• university × faculty;
• university × level × graduation year;
• university × field × graduation year;
• country × level × graduation year;
• university region × level (for government bodies).

Arbitrary combinations of filters are not built: otherwise a hidden small number could be worked out by comparing two overlapping requests (see the suppression methodology).

Weekly snapshots

Every Monday at 00:30 Tashkent time the Platform records the state of each education entry for the past week: university, faculty, field, level, status, graduation year and month, study form and funding, how the affiliation was confirmed, and also profile and CV completion, visibility to employers, consent to the university's access, the readiness score and the number of applications, interviews, offers and hires.

The cohort keys of past weeks are frozen: if a student changes their field later, their past weeks stay in the old cohort. Rebuilding past weeks from archived data (analytics:backfill) is approximate, because it takes the keys as they are at the time of the rebuild.

Cohort figures

• on the Platform; completed the first profile setup; have a CV; visible to employers; consented to the university's access (a count only, no names); confirmed;
• career readiness: the median, quartiles, a distribution over 10 intervals, the share at "ready" or above, and the change over the week. The readiness score is computed by version (currently r1), and trends are shown separately for each version, so that a change to the score's method does not break the series;
• applications, interviews, offers and hires in a month;
• skills by group (about 25 groups): the share of students with the group's skills and the distribution by level; individual skills only when at least 10 people list them;
• verified skills (from the week the Privacy Policy of 2026-12-01 is published; earlier weeks are not recomputed): the share of students whose skill in the group is verified by a Platform test. Only live test verifications count — the "course completed" mark does not — only for skills the student lists in the profile, only from item banks that passed expert review, and only badges the student has not hidden. The number of students with a verification and the number listing the skill without one add up to the number of students with the skill; the share is shown only when the number with a verification passes the publication rules itself;
• coverage — see the coverage methodology.

Updates and versions

Statistics are recomputed every night at 01:30 Tashkent time. Each number is shown with its update time, methodology version and coverage. A published computation never changes; a new one replaces it as a whole. If the counting rules change, the methodology gets a new version and older computations keep the previous label.