Cancellation signal model

Every contract in the book scored weekly against the financing conditions, contact history and stage cost that preceded your last two years of cancellations. Find out in week three instead of week fourteen.

$1.50M

Reference-scale value

Data sets in the model

8 weeks

Build and deploy

Derived

Evidence grade

Seventy-four a year is a fact of this book. The model does not argue with it.

A credit file does not stop a buyer losing a job, and a risk score does not save a marriage. Any model that tells you it prevented a cancellation is selling you a coincidence with a chart on it.

So this one claims something narrower and defensible: how many weeks earlier you knew, and what those weeks were worth in cost that had not yet been committed.

The decision behind a cancellation takes weeks and leaves a trail — a financing condition that slipped its date, a buyer who stopped answering, a selections appointment booked twice and never kept.

All three live in systems you already run. None of them is read as a signal, because nobody has ever joined them to the outcome that followed.

Cause 01

Carrying and re-selling the released home

Forty-one cancellations after start, standing an extra 3.4 months plus a concession on re-sale. Forty per cent is time no warning recovers.

Cause 02

Committed incentive and selections

Seventy-four contracts carrying $4,200 each of cost committed before the cancellation landed.

Cause 03

Structural work the next buyer did not want

Forty-one started homes carrying $9,700 each of non-standard work.

Cause 04

Starts against a book that overstates itself

Twenty-six starts a year released against contracts that cancelled. Credited to trade capacity forecasting and not claimed here.

Want more detail?

The rest of this model — the data sets it runs on, the cause split, the working behind the reference figure and a sample output — goes out by email. Two fields, nothing else.