Data released this morning generally describes a month or a quarter that finished some time ago. That is not a reporting delay: an official statistic carries two dates, the period it measures and the day it was published, and the distance between them is the time it takes to collect and process the returns the estimate is built from.
Publication itself happens at one scheduled instant. The Australian Bureau of Statistics releases at 11.30 am Canberra time on the day named in its calendar and treats everything as confidential and under embargo until then, with government, commercial and public users all receiving it simultaneously. That is why a figure arrives fully formed rather than leaking out, and why release dates are knowable months in advance.
An early estimate is then expected to change. The Bureau describes revision as reflecting the trade-off between timeliness, accuracy and reliability: a first estimate is built on preliminary returns, more complete data arrives afterwards, and the figure is restated. It characterises revisions as an inevitable consequence of the process rather than as errors, so a number that moves is the system behaving as designed rather than a correction of something that was wrong.
Seasonally adjusted series move for a second and separate reason. The factors that strip out predictable calendar patterns are re-estimated as each new period arrives, which shifts past points a little as well as current ones. A genuine change in a seasonal pattern generally shows up in those factors after about three years, and they become reasonably stable after roughly six.