What was checked, what is missing, and what was corrected

A script reads the text of every archived PDF and looks for each workbook figure in the release it was transcribed from (a presence check, not a position check). Small counts under 100 are skipped because they appear everywhere.

The exceptions

• Prospect and Walkerville, FY 2015/16. The workbook has $14,393,352.18 for the group; the CBS release prints $14,393,272.17. The workbook is $80.01 too high, which is also why its FY 2015/16 LGA total is $80.11 above the printed total.
• Light and Mallala machines, FY 2013/14. CBS printed 109 machines for the group. The workbook split them as 54.5 and 55, which add to 109.5, so the rebuilt group shows 110.
• Hotel entitlements, April 2025. The workbook has 11,484; the CBS release prints 11,474 for April (11,482 is May’s figure).
• One image-only PDF. 2022-23 Q3.pdf (Gaming Machine Licence Statistics) is a scanned print with no text layer, so its three months could not be checked automatically.

These are left as transcribed so the site matches the original workbook; the sizes involved are immaterial to every chart.

Two independent CBS series measure the same NGR: the monthly statewide release and the annual release by council area. Adding up every council row should give the statewide figure. Statewide months are printed to the nearest $10,000, so twelve of them can drift by up to $60,000 from the exact total.

The largest gap is $15,769 in FY 2016/17, about 0.002% of the year’s NGR. Because the two series agree, the FY 2014/15 LGA total can stand in for the missing statewide year on the annual chart, clearly marked.

CBS combines councils with few venues into groups so that no single venue’s revenue can be worked out. The releases up to FY 2021/22 say a council with fewer than five venues is grouped with another; from FY 2022/23 the threshold is fewer than three. The workbook stored each group by repeating it on every member’s row after dividing it equally. In FY 2013/14, for example, CBS published Barunga West, Copper Coast as one row with NGR of $10,277,669.86; the workbook shows $5,138,834.93 against each council. That is why 463 workbook rows have fractional machine or venue counts such as 85.5 or 187.33.

This site groups the workbook rows back together (rows of one year with identical NGR and NGR per venue belong to one published row), maps each group as a single shape, and never presents a member’s equal share as its own figure. The rebuilt rows match the number of rows CBS printed each year, and every rebuilt NGR figure is found in its PDF except the one listed above.

Council names drift across the LGA releases: typos, punctuation, council-type suffixes, renamed councils and the workbook’s own splits. Each of the 94 names in the workbook is mapped by hand to an ABS Local Government Area (2024 edition) in scripts/lga_crosswalk.csv. The 29 names that differ from the ABS area are listed here.

• Statewide: FY 2014/15 is missing

  No statewide release for that year is in the archive, and the CBS website blocks automated downloads, so the monthly figures were not filled in. The annual NGR comes from the LGA release instead and is marked as such.

• Licences: Jul 2017, Aug 2017, Sep 2017

  The 2017-18 Quarter 1 release is not in the archive.

• Manufacturers: Oct 2023, Nov 2023, Dec 2023

  The archived “2023-24 Q2” file repeats Quarter 1, so the second quarter of FY 2023/24 is missing.

• Councils: no machine counts for FY 2019/20

  That year’s LGA release has no machine column. The workbook stores 0 for every area; the site treats the counts as not published.

• Licences: casino entitlements under Special Circumstances, 2014

  From January to December 2014 the Special Circumstances entitlements jump by about 1,000 (to 2,185, then 2,212), and fall back to 1,192 in January 2015 when the casino is first listed with 1,020 of its own. The figures are shown as published.

• COVID-19, FY 2019/20

  Venues closed in late March 2020. The statewide release reports zero machines for March to June 2020 and almost no NGR for April to June. FY 2019/20 is shaded on every chart.

• Council venue counts, FY 2021/22

  The FY 2021/22 LGA release totals 603 venues, against 484 the year before and 488 the year after. The figure is shown as published.

• Revised LGA release, FY 2022/23

  CBS reissued the FY 2022/23 LGA release in January 2024 to correct its source data. The archive holds the reissued version.

• Grouping rule changed in FY 2022/23

  The FY 2013/14 to 2021/22 LGA releases group a council with fewer than five venues; the FY 2022/23 release and later ones group only those with fewer than three. That is why the number of combined groups falls from 16 to 12 that year, and why Campbelltown and Kangaroo Island (three venues each) are published on their own from then, as Grant was in FY 2022/23.

• Changing council groups

  The groups CBS combines also change from year to year under the same rule (Grant and Mount Gambier were separate in FY 2022/23 and combined again the year after, for example). A council’s history on the Councils page notes every change of group.

The original Power BI report puts a financial-year axis on tables that hold one row per month and uses the default Sum aggregation. That is right for flows such as NGR and tax, which accumulate over the year, but wrong for stocks (machines, venues, entitlements, licences) and for shares: twelve snapshots added together are about twelve times the real level. This site averages snapshots (or takes June) instead, as the workbook’s own INFO pivots do.