Methodology

How a grant rate is calculated

Student visa grant rates come from the monthly student visa program tables published by the Department of Home Affairs and released through data.gov.au. StudyVisaHub does not adjust the counts. It sums them over a period and divides.

grant rate = visas granted in the period ÷ visa decisions in the period
refusal rate = (decisions − grants) ÷ decisions

Population

• Subclass 500 student visa applications, counted at the point they are decided, not at the point they are lodged.
• Split by country of citizenship, by education sector, and by whether the application was lodged inside or outside Australia.
• Periods are the monthly release codes the Department publishes for the financial year. The site shows a rolling window of the three most recently released months, and compares it against the three months before that.

Exclusions

• Applications still on hand, and applications withdrawn before a decision, are not decisions and are not counted on either side of the division.
• Three source countries are excluded from every published rate: Iraq, Lebanon and the Syrian Arab Republic. Case numbers are small and heavily affected by conflict, and a rate built on them would describe the conflict rather than the visa programme.

How change is expressed

A movement between two periods is always stated in percentage points, never as a percentage change. A rate that moves from 80 per cent to 76 per cent has fallen 4 points. Calling that a 5 per cent fall would be a different and much less useful statement.

Source: Department of Home Affairs, student visa statistics, released through data.gov.au.

Why the monthly series starts on 14 November 2025

The monthly grant rate series held here begins with November 2025. It does not run back further, and the site publishes no year on year comparison of grant rates.

The reason is a change in the rules, not a gap in the data. On 14 November 2025 a ministerial direction on how student visa applications are prioritised for processing took effect (Ministerial Direction 115). It changed the order in which applications are worked through, which changes what a monthly decision count is made of. Averaging across that date, or comparing a month after it to the same month a year before, would put two different processing regimes on the same line and call the difference a trend.

So the series stops at that boundary instead. Every query that builds a rate is bounded on the server, not just in the page, so a figure from before that date cannot reach a chart by accident.

Source: Study Australia, Ministerial Direction 115 (commenced 14 November 2025).

The sample floor, and what happens below it

A rate built from a handful of decisions is noise wearing the clothes of a statistic. Three thresholds are applied, and they are deliberately different because they protect different figures.

• Fewer than 30 decisions: no rate is shown for a country and course type combination. The page prints “Sample too small to show” with the actual count, so you can see how small it was. The suppressed rate is never rounded, averaged into a parent figure, or replaced with a default.
• Fewer than 20 decisions: the country row is not returned by the query at all for the rolling three month view.
• Fewer than 100 decisions: a country is left out of the ranked country tables. Ordering countries against each other demands a larger sample than showing one country on its own.

The floor applies to country and course type breakdowns only. It is never applied to a sector total or a national total, because those are built from tens of thousands of decisions and suppressing them would hide the very figure the small samples belong to.

Matching a course to a national qualification

To show which occupations a course leads to, a course on the Commonwealth Register of Institutions and Courses for Overseas Students (CRICOS) has to be identified as a particular national qualification. That match is a StudyVisaHub calculation, and it is the one derived object the professional view is built on, so its method and its failure rate are published here in full.

The method

• The course name on the register and the national qualification title are both normalised: case folded, punctuation and extra spacing removed.
• A match is recorded only when the two normalised strings are identical. Every stored match carries the method that produced it.
• Approximate or fuzzy matching is banned outright. It was tested, and 14 of the 15 matches it suggested were wrong. A wrong course to qualification match is invisible downstream: it produces a confident, sourced, entirely incorrect occupation.

Measured coverage

9,716 courses are matched. Measured against the 10,582 active vocational-level courses in the test population, that is a 91.8 per cent match rate.

Known misses

• The 91.8 per cent figure covers certificate-level courses plus whatever diploma courses happened to match. Diploma and advanced diploma coverage was never measured separately, and 5,985 courses sit in that unmeasured group. Do not read 91.8 per cent as the coverage of diplomas.
• A provider that gives its course a marketing name rather than the national qualification title will not match. That is a naming difference, not a judgement about the course.
• Where no match is held, the panel says the record is not held. It never says there is no pathway.

Moving between the two occupation code sets

Australia is mid-way through replacing one occupation classification with another. Skilled migration lists still use the Australian and New Zealand Standard Classification of Occupations (ANZSCO). Training and labour market data increasingly use its replacement, the Occupation Standard Classification for Australia. The two do not line up one to one.

One rule governs every join on this site: occupations are matched by code, through the correspondence file the Australian Bureau of Statistics publishes for that purpose, and never by occupation title.

The reason is easiest to see in a single case. The title “Chef” exists in both classifications, but the code that skilled migration treats as a chef is the one the training data labels “Senior Chef”, and the code labelled plain “Chef” in the newer classification has no ancestor in the older one at all. Joining those two lists on the word “Chef” produces a match that looks right, reads right, is sourced, and is wrong. Nothing downstream would catch it.

Source: Australian Bureau of Statistics, occupation classification correspondence (snapshot held 23 July 2026).

How course fee bands are worked out

Course listings show a fee band from 1 to 5. It is a StudyVisaHub calculation, not a published figure. The only input is the total course cost the provider has published on the course register, in Australian dollars, and the bands are fixed cut-offs:

band 1: under $10,000 · band 2: $10,000 to $19,999 · band 3: $20,000 to $29,999 · band 4: $30,000 to $39,999 · band 5: $40,000 and above

Where the provider has published no total cost, no band is shown. The band is never estimated from a similar course.

How often each source is refreshed

These are the intended intervals. Every page also prints the date the record it shows was last checked. Where that date is older than the interval below, that source is behind, and the gap is left visible rather than filled with an estimate.

Licence and attribution

The core datasets are open government data. They may be reused commercially, on the condition that the publisher is named and the licence is identified by name and version. StudyVisaHub is not the creator of any of them.

• Student visa program data and the course register are published under Creative Commons Attribution 2.5 Australia. Copyright rests with the Commonwealth of Australia, through the Department of Home Affairs and the Department of Education respectively.
• Training occupation pathways and Australian Bureau of Statistics material are published under Creative Commons Attribution 4.0, copyright Commonwealth of Australia. The pathways dataset carries a carve-out for third party material inside it.
• State and territory nomination lists are published by each state and territory under its own terms. Those terms are not documented consistently across the eight jurisdictions, so each row here carries a link to the state page it came from, and the date it was read.
• Scholarship entries reproduce facts and link out. Provider wording is not republished, because provider pages carry their own terms.

If an attribution on this site is wrong or incomplete, that is a defect and it will be corrected. Tell us and it will be fixed on the same footing as a wrong number.

What is not held and not published

• No prediction. Nothing here forecasts an outcome, a processing time, a list change or an individual result. Rates describe decisions already made.
• No advice. The site describes what was published. It does not tell anyone what to do about their own case.
• No provider ranking or quality score. Providers are listed alphabetically. Provider records are never joined to grant or refusal rates, and never coloured or ordered by processing priority.
• No country risk rating. A three-level country classification was published here until 30 August 2026 and was withdrawn. No authority publishes such a classification, so it was a StudyVisaHub inference presented beside official rates. Labelling an inference as an estimate does not make it publishable.
• No Victorian occupation list. Victoria does not publish one. That is recorded as not held rather than left blank.
• No paid placement. No provider, agent or course can pay to appear, to rank higher, or to be removed.

Corrections

Every correction made to a published figure or statement is listed on the corrections register, with what was published, what it says now, and the date it changed. A closed entry is never edited. A correction to a correction is added as a new entry.

If a figure here disagrees with the source it cites, the source wins. Send the page and the original publication to contact@studyvisahub.com.au. Reports are acknowledged within two business days, and a verified correction is published within five.

Some explanatory writing on this site is drafted with the help of software and reviewed before publication. Figures and calculations are not: every one traces to a government publication linked on the page it appears on. For a figure that carries real weight in a decision, check it against that publication.

Last reviewed: 2 September 2026. Record counts stated on this page are as at that date.