Caveats & Limits

An honest list of what this data cannot tell you.

What we cannot show

  • Private communications. No emails, no meeting notes, no informal deals. Only public records.
  • True project cost. TEI includes contingency and assumptions. The real cost isn't known until a project is complete and audited — often years late.
  • Intent. We can show a donation happened before a contract was awarded. We cannot show why.
  • Suppressed data. If something was removed from Budget Paper 4 quietly, we may not have caught it.
  • Non-disclosed donations. AEC thresholds mean smaller donations don't appear.

Known limitations

Project data

  • BP4 changes category labels between years. We normalise where we can.
  • "Delivery entity" sometimes rolls up multiple agencies — we report the top-line.
  • Cancelled or indefinitely paused projects are flagged but their "final cost" is uncertain.

Donation data

  • AEC reporting lags by up to 18 months.
  • Party naming is inconsistent across years ("ALP", "Australian Labor Party", "ALP Victorian Branch") — we roll up.
  • In-kind donations are reported as cash-equivalent where disclosed.

Council data

  • VLGGC categories differ slightly year-to-year.
  • Councils sometimes restate prior years' figures — we carry the latest.
  • Small rural councils have more missing fields than metro councils.

Network / Mates Detector

  • Name-matching is probabilistic. A director called "John Smith" in one dataset and "J. Smith" in another may or may not be the same person.
  • Candidate matches we can't verify are excluded from public-facing pages.
  • Absence of a link on the graph does not mean no relationship exists. It means no public record of one exists in our data.

What this site is not

  • Not a court of law. We don't adjudicate.
  • Not an investigation. We're a transparency tool. If what we show motivates an investigation, good.
  • Not a campaign. We don't endorse parties, candidates, or outcomes.
  • Not complete. If a dataset becomes public we'll add it. The absence of something isn't proof of innocence, just of our coverage.

How to use this responsibly

  • Always click through to the source before quoting a number.
  • Read the Methodology before drawing conclusions from a pattern.
  • Use Statement Tracker to find public responses from named entities.
  • If the number on this site disagrees with the source, the source wins — and please tell us so we can fix it.

Last reviewed: 11 Apr 2026