Methodology · rules v2.1 · 18 Sep 2026
How this site works
This page is written from the same file the site’s code runs on, so what you read here is exactly what the site does. If a rule changes, the version number changes and the change is logged in public.
1. Where the figures come from
Figures come from official publishers: the Australian Bureau of Statistics, the Reserve Bank, the Treasury and Budget papers, the Australian Office of Financial Management, and the agencies that run each program. International comparisons come from the International Monetary Fund and the World Bank. Most series are fetched automatically from the publisher’s own data service. Where a figure exists only in a PDF (parts of the Budget, some agency reports), a person reads it from the document, records where it came from, and sets a date to check it again. Each measure’s page says which of the two applies.
One figure is not official: market rents for new leases, which no government agency publishes. The household comparison uses Cotality’s national median rent for this and says so, alongside the ABS rent index for existing leases.
Every figure links to its source. Every page shows the date of the latest data point and the date it was last checked, which are different things.
2. What “checked” means
“Checked” is used on this site in two different senses, and it is worth being exact about both.
Automated checks, every day. A program re-fetches every automated series and tests it: is the source still reachable, is the newest figure inside a plausible range, does it agree with a second official source where one exists (for example the ABS and the Reserve Bank for inflation, unemployment and wages, and the AOFM’s register for government debt), have earlier figures been revised, and do Budget categories add up to the published totals. The latest run made 480 checks: 472 passed and 6 raised a warning (usually a source that was slow to respond). A failed critical check keeps the last verified figure on the page, marks it “under review” and alerts the publisher, rather than publishing a doubtful number. The full log is on the data page.
Human checks, when sources change. Figures typed in from documents, every commitment, every verdict rule, and every entry on the controversies page are checked by a person against the original source. Each shows the date it was checked and when it is next due. People do not review every automated daily update: the tests above do that job, and anything that fails them is held back for a person to look at.
3. Commitments get verdicts. Nothing else does.
A verdict is given only when three things are true: the government itself made the commitment, the commitment contains a number or a date, and official statistics can test it. This is deliberate. It is easy to make any government look good or bad by choosing the yardstick after the fact. Measuring a government against its own words is the one yardstick nobody can call unfair.
| Verdict | Meaning |
|---|---|
| Met | The commitment has been delivered. |
| On track | The latest official figures are consistent with the commitment being delivered. |
| In progress | The deadline is still ahead and the figures are moving the right way; it’s too early for a verdict. |
| At risk | The figures are behind where they need to be, but the commitment could still be delivered. |
| Off track | On the latest official figures, the commitment will not be delivered without a clear change. |
| Not met | The deadline has passed and the commitment was not delivered. |
The rule for each commitment
| Commitment | Rule |
|---|---|
| Keep consumer price inflation between 2% and 3%. | Within target if the latest annual inflation rate is between 2% and 3%. This is the Reserve Bank’s job, so it is shown but does not count in the government’s tally. |
| Cut household power bills by $275 a year by 2025. | Not met if the regulated default power price in the benchmark regions is not $275 a year lower than when the promise was made, now that the 2025 deadline has passed. |
| Cut greenhouse emissions to 43% below 2005 levels by 2030. | Judged on the government’s own latest emissions projections: on track if they show the 2030 target being met, off track if they show it being missed. |
| Build 1.2 million new well-located homes in the five years from 1 July 2024. | On track if homes completed since July 2024 are at or above the 60,000-a-quarter pace needed; at risk if up to 10% behind; off track if more than 10% behind. |
| 9 in 10 GP visits bulk billed by 2030. | Met at 90% or more. Before the 2030 deadline it is “in progress” while the rate is rising, and “at risk” if it is flat or falling. |
| Reduce gross debt as a share of the economy over time. | Measured the government’s own way, as a share of GDP. On track if the latest actual is lower than at 30 June 2022 and the Budget forecasts it lower still in four years; at risk if only one of those holds; off track if neither. |
| Slow annual growth in NDIS costs to no more than 8% by 1 July 2026. | On track if the latest reported annual growth in scheme costs is 8% or less; off track if it is above 8%. |
4. The record since the government took office
This site exists to hold the government of the day to account, so the main comparison on every measure is simple: where was it when the government took office, and where is it now? The starting point is the last figure before the government’s first full period: May 2022 for monthly figures, the June quarter 2022 for quarterly figures, and the 2021-22 financial year for annual ones. Every chart shades the government’s time in office and fades what came before.
A since-office comparison is a fact, not a verdict. Two things are always shown beside it so it can’t mislead: how much control the federal government has over that measure, and a note wherever the starting point was unusual (for example, unemployment near a 50-year low, or interest rates at a pandemic emergency setting).
Better or worse colouring is used only where nearly everyone agrees which direction is good: real wages: private vs public sector, electricity prices, wholesale gas prices, mortgage costs, unemployment, productivity, business collapses, consumer confidence, gdp per capita vs total gdp, household income per person, budget deficits, interest on government debt. For mortgage interest and income tax, fuel prices, interest rate rises, home prices, new residents per new home, rents vs wages, home prices vs earnings, public vs private sector jobs, living standards by decade, size of government in the economy, tax take, government spending (share of gdp), size of the public service, net overseas migration, defence spending, who pays the income tax, money sent overseas by individuals, spending on consultants, which way is “good” is a matter of political opinion, so the site says only higher, lower or little changed. The front-page count of measures that are better or worse since mid-2022 includes only the first group.
Over the past year is shown second. It compares the latest figure with the same period a year earlier (the previous year for annual figures), which avoids seasonal swings. A change smaller than 0.2 percentage points (for rates) or 1% (for levels and indexes) counts as little changed. Dollar figures that can go negative, such as the budget balance, are compared in dollars. One series, housing completions, isn’t seasonally adjusted, so it is left out of since-office comparisons and judged on its commitment instead.
5. Whose doing is it?
Every measure carries one of three labels, with a sentence explaining why.
- Direct federal control. The federal government decides this itself, mainly through the Budget or legislation.
- Shared control. The federal government is one of several hands on the wheel, alongside the states, regulators or markets.
- Indirect influence. Decided mostly by others (the Reserve Bank, world markets, businesses and households). Federal policy nudges it.
6. Revisions and corrections
Statistical agencies routinely revise past figures. When they do, the site shows the revised series, because that is the official record. If this site makes a mistake of its own, in a figure, a calculation or a description, it is fixed and the fix is listed on the corrections page with the date and what changed.
7. What this site doesn’t do
- It doesn’t forecast. Budget forecasts are shown as the government’s own forecasts, shaded differently from actual results.
- It doesn’t take a side on contested questions such as how large government or migration should be.
- It doesn’t publish figures from campaign material, media releases alone, or think tanks as if they were official statistics.
8. Reuse
ABS and RBA material is used under Creative Commons Attribution 4.0. This site’s own tables and calculations may be reused under the same licence: credit “Gov Score” and link to the page. Every measure page has a ready-made citation and a CSV download, and the data page has everything in one file.