Skip to content
Gradly

Methodology

Salary benchmark

Methodology s1Draft for review by the ministry and native speakers

Platform data, not official statistics. The salary benchmark follows methodology s1. It shows the salaries employers state in vacancies on Gradly, not what people actually earn.

Which vacancies count

  • Vacancies of verified companies first published in the last four closed quarters. The benchmark is recomputed once a quarter, after the quarter closes.
  • Full-time vacancies with a stated salary only. "Negotiable", part-time pay and internship stipends are left out.
  • For a range, its midpoint is used; when only one bound is given, that bound.
  • Salaries in US dollars are converted to sums at the Central Bank of Uzbekistan rate of the publication day; if no rate was published that day, the latest published rate is used.
  • Repeats of one vacancy (the same company, the same title after normalization, the same region) count as one vacancy, as in labour market methodology m1.
  • Once recruitment agencies launch, a vacancy an agency posts for a client counts as the client's vacancy: repeats are found by client, and the dominance rule below must hold both by clients and by the agencies that posted the vacancies.

A vacancy's profession

  • The employer chooses the profession when publishing a vacancy.
  • If it does not, the Platform suggests one from the vacancy title: first by an exact match with the profession's known titles in three languages, then by the words of the title, then by the vacancy's skills, checked against its category. Below a confidence threshold the profession stays "not determined" — the classifier does not guess.
  • The classifier never changes an employer's choice.
  • Vacancies without a profession count in their category but in no profession.

What is shown

  • The median — from 10 vacancies in a cell.
  • The 25th and 75th percentiles (the typical spread) — from 20 vacancies.
  • The dominance rule: a cell has at least 3 companies, and none of them accounts for more than half of the vacancies. Otherwise the value is withheld.
  • Amounts are rounded to 100,000 sums.
  • The sample size is shown as a band — "10+", "50+", "200+" — not as an exact number.
  • A withheld value is shown as "…".

Breakdowns

Country; category; category and region; profession (always within its category); profession and region; profession and required experience.

When there is little data

The benchmark follows a chain and always says where the number comes from:

  1. the profession in the chosen region;
  2. the profession in the whole country;
  3. the profession's category in the region;
  4. the category in the whole country;
  5. "Not enough data".

A value is never put together from two cells or estimated: only what is published in full is shown.

How companies and people are protected

The cells are published by the same machinery as the statistics for universities and government bodies: values are checked and rounded before they are written, small samples are never stored, and the public pages read only the cells allowed for everyone. The benchmark contains no data about people.

Limitations

  • Not every company states a salary, and a stated amount is the employer's offer, not actual earnings.
  • Not every employer, sphere and region is represented on the Platform.
  • The profession classifier can be wrong about rare titles; that is why uncertain vacancies stay without a profession.
  • The professions' ISCO-08 codes are drafts and are not used in the computation.