FOI is the backstop for everything else in this register. When data is not published, a formal request is the remaining route, and how well that route works determines whether any of the rest is theoretical.

This entry is included as one with a good answer. Agencies and ministers provide the OAIC with quarterly and annual statistical returns on FOI activity, review numbers and estimated costs. The returns are published as CSV from 2011-12 onwards and in a summary spreadsheet, alongside a dashboard that lets a user filter by portfolio or individual agency and by variables including decision outcome, response time and the exemptions applied.

That is a well-run statistical release: timely, disaggregated, machine-readable, and covering the awkward variables rather than only the flattering ones. It is the reason the decline in full grants and the rise in refusals can be established at all, by anyone, without an FOI request about FOI.

This entry previously recorded a verdict of partial, with blockers aggregated and stale, on the basis that agency-level totals were published annually and the breakdown by exemption and outcome was much thinner. Checked on 2026-09-01, and it was wrong on both counts. Reporting is quarterly as well as annual, the data is machine-readable and goes back to 2011-12, and the exemption breakdown the entry said was missing is one of the dashboard's filters. Data for the first three quarters of 2025-26 was already available at the time of checking.

The correction is recorded in full rather than quietly applied because the mistake is instructive. This was a seed entry written from documentation about the OAIC rather than from the OAIC's data page, and it accused a regulator of thin reporting while that regulator was publishing one of the better statistical series in the Commonwealth. A false gap is worse than no entry, and a false gap aimed at the body responsible for the access regime is worse than most.

The substantive point the entry was reaching for survives, and belongs somewhere else. Good statistics on FOI outcomes do not fix slow FOI decisions, and the performance problem -- rising refusals, lengthening review times -- is real and well evidenced precisely because this dataset exists. Measuring a problem accurately is not the same as solving it, and this register should be careful not to confuse the two.

Related: Which Senate orders for documents has the government refused, and on what public interest immunity ground?, Commonwealth FOI.