Campaigns and party research

Every serious campaign runs on the same artefact: a seat table. Margins, swings, demographics, a few economic indicators, assembled by one person from tally room downloads and census tables over several weeks, and quietly out of date from the moment a boundary moves.

Three federal redistributions are underway — South Australia, Tasmania and the Australian Capital Territory — and NSW state boundaries are redrawn before 2027. Somebody in your office is about to rebuild that table by hand again.

The work is the join, not the data

None of the inputs are secret. The Australian Electoral Commission publishes results down to the polling place, the Australian Bureau of Statistics publishes demographics down to the mesh block, and both publish their boundaries. What nobody publishes is the join, because the two sets of boundaries move on different cycles and neither carries the other's keys.

Getting votes onto the statistical geography is the whole trick. Once they are there, a seat profile is a query: age structure, income, industry of employment, housing tenure, language, socioeconomic index, population change — and, because they sit on the same spine, flood and bushfire exposure, canopy cover, projected heat.

Redistribution prompt

›

For the proposed South Australian boundaries, give me notional margins from the 2025 result, the suburbs that moved between divisions, and how each moved area voted.

The brief that comes back names the divisions under the new boundaries, the notional two-party-preferred figure, the movement against the current boundaries, and the areas that changed hands with their profile. It also states how the apportionment was done and where it is weakest, because a notional margin someone cannot check is a liability in a preselection argument.

Seat selection for community campaigns

Community-independent campaigns and the organisations that fund them have a specific problem: choosing which seats to contest, with a defensible reason. That is a screening question shaped exactly like the site screening problem — a set of criteria applied across every division at once.

Seat selection prompt

›

Rank all 150 divisions by combined climate exposure, tertiary-educated share, median age and the 2025 margin, and show me the top twenty with the working.

Climate exposure by electorate is a cut almost nobody publishes, and it exists here only because the hazard layers and the demographic spine were already joined for other work. Dwellings inside a flood planning area per division, bushfire prone land share, projected days over 35 degrees by 2050, coastline change on the seats that have coastline.

Timing and terms

Federal results from 2004 forward, boundary sets including proposals still under consultation, and NSW state and local government results are the first build. Electoral Commission material carries its own licence rather than a Creative Commons one; derived work and redistribution are permitted with attribution, and that notice stays attached to the datasets and to anything produced from them.

If your question is about a particular state or a particular contest, tell us which — the build order follows the campaigns that need it.