BuildsConnect

Home/Project timing

Not “this project exists”. When it will actually need you.

Every NSW project moves lodgement → determination → construction certificate → occupation, and we hold 345,678 observed transitions to measure it against. This is the one product here that does predict, so every number on it carries the band it came from and the count of comparable projects behind it.

The honest headline

We sell a window that is right about half the time, and we tell you which half.

The default window is the p25–p75 band. Measured on held-out data it contained the real date 52.9–57.2% of the time against a nominal 50, slightly better than it claims, which is the safe direction to be wrong in. The wider p10–p90 band is published alongside for anyone who needs the high-confidence answer.

82–85%of actual dates fell inside the wide band (nominal 80)
180dtypical narrow window, metro high-value
344dwide window for the same segment
4market segments measured separately

Trained on projects lodged 2021 to 2025 and tested on the six months after, forwards in time, the way it is used. The midpoint’s median error is 52–62% depending on the segment, and we publish that too: it describes a point estimate nobody is quoted, and it is not what the product promises.

The field you actually work from

Your patch, ranked by what is likely to move this quarter.

A band tells you how long this kind of project usually takes. It cannot tell you who to call this month. So each row also carries the probability that this project reaches its construction certificate in the next 90 days, given how long it has already been waiting, which updates every day and sharpens as a project ages.

  • The list is ranked, never filtered. Nothing is hidden for scoring low, and the number sits on every row so you decide where your list stops being worth the drive.
  • Where a segment’s history is too short to support a window, we publish the widest one it does support and say so, rather than extrapolating past the last project we watched finish.
  • Projects whose consent is approaching lapse, or that sit outside what our observations can speak to, are excluded from scoring rather than given a low score.

What this will never tell you

Whether an application will be approved.

We hold no probability of approval, we publish none, and there is no field on this product that could be read as one. That would be planning advice, and the honest answer for any individual application is that we do not know. This product covers what a consent requires and when, never whether it will be granted.

This product is not on sale yet. It is measured, calibrated and running against live data, and it stays internal until the accuracy holds where we would need it to. There is nothing to buy on this page.

How we measure things See coverage