When the platform sets up a contract it asks you for a reference date: a day of the month, defaulting to the last day. It is one field, and it decides two different things.
"Reference date" is the acts' word. Your contract and your head contractor may call the same day the claim date, the cut-off date, the valuation date or simply the 25th, and a claim period is as often called a valuation period or just the month. The names are in the glossary; the mechanics are the same whichever one you use.
What it decides
It ends the claim period. A claim covers the work you completed up to and including the reference date, and runs from the day after the previous period's reference date. On a contract with a 25th reference date, your March claim covers 26 February to 25 March, not the calendar month.
It is the earliest day you can serve. Every act ties the right to make a progress claim to that date. Before it, there is no claim to make; on and after it, there is.
Those are the same date doing two jobs, and conflating them is the usual source of confusion. The reference date is not the day you get paid, and it is not the day the claim is due: it is the cut-off for what the claim is allowed to cover, and the starting gun for serving it.
Claim periods
A claim period is the stretch of work one claim covers. Periods run back to back with no gaps and no overlaps, and each is identified by the month its end date falls in, its reference month. That is how the platform can tell you "March is unclaimed" on a contract whose March period actually started in February.
The first period is special: it starts on the contract start date rather than on a previous reference date, because there is no work before the contract began. So a contract starting 10 February with a 25th reference date has a first period of 10 to 25 February, then full periods after that.
Where the reference date is the last day of the month (the default, and the statutory default nearly everywhere) every period after the first is exactly a calendar month.
Only one claim per period
You get one payment claim per period. In NSW, South Australia, the ACT and Tasmania this is explicit in the legislation: one claim per named month. Serving a second one in the same month risks the second claim being invalid.
This matters most for retention releases, which people naturally want to serve as their own document the moment a milestone is hit. If that month already carries a claim, bundle the release into it instead. See retention.
Claims are cumulative
Progress claims are claimed to date, not per period. Each claim states the total value of work completed on each line since the contract began; what you are paid this period is that total less what has already been certified.
The practical consequence is that a line reduced or rejected in one period is not lost. It is carried in the next claim's to-date figure, and re-presented. That is why the platform shows re-claimable amounts when you start a claim after recording a payment schedule that cut something.
Claiming backwards, not forwards
The platform values work already carried out up to the reference date. It does not support claiming for work you are about to do.
That is a deliberate constraint. NSW, Victoria, Tasmania and the ACT permit forward claiming; Queensland and Western Australia do not. Rather than make the behaviour differ by jurisdiction, the platform adopts the model that is valid everywhere.
The last day you can serve
The reference date is the earliest day. There is also a latest, and it is not the end of the month. It is a longstop measured from when the work was carried out.
Most acts express it as some number of months after the work the claim relates to was last carried out: 6 months in WA, Queensland and South Australia; 12 months in NSW, the ACT and Tasmania. Victoria measures its 6 months from practical completion instead (s 14C(b)). Everywhere else the platform uses your contract's practical completion date as the proxy for "when the work was last carried out", since exact last-work dates per line are not something anyone records.
Two things extend that window:
A claiming period in your contract. Most acts take the later of the statutory longstop and any period your contract sets, so a contract term can only ever extend the window. If your contract has one, record it on the contract and the platform will use it.
Termination. In NSW, the ACT and WA, terminating the contract opens a fresh claim window on the termination date. Record the termination date and the platform accounts for it.
New Zealand has no longstop at all. The Construction Contracts Act 2002 imposes no "within N months" cap on late service, so a New Zealand claim is never out of time on that basis.
The figures for your jurisdiction, each with its section reference, are in the timing summary.
The reference date is fixed at contract creation
You cannot change it later. Changing it mid-contract would retroactively re-bound periods you have already claimed: a March claim covering 26 February to 25 March would suddenly be a claim for a different span of work than the document you served says it was.
If a contract's reference date was recorded wrongly and no claims exist yet, the fix is to correct the contract before the first claim. After that, it is fixed.
Which day of the month counts
The date of service is the calendar date in the contract's own jurisdiction. A claim served at 8am in Perth on 30 June is served on 30 June, even though it is already 10am on 30 June in Sydney and still 29 June in UTC.
The platform resolves this using the jurisdiction's time zone, including daylight saving. You will not see it (that is rather the point) but it is why serving late in the evening on the last day of a window is safe rather than a coin flip.
Victoria adds one rule the others do not have: a claim served early is treated as served on the first day it could validly have been served. Serving ahead of the reference date there does not start the respondent's clock early.