The first time I tried to map out a hidden truth in housing, I wasn’t using ChatGPT or machine learning—I was scanning water bills.

Back then, we built a report called Speculative Vacancies. It used abnormally low water consumption as a proxy to reveal something the property industry didn’t want to admit: that thousands of homes in speculative markets weren’t lived in at all. They were parked. Financialised. Treated more like assets than shelter.

It wasn’t elegant—but it worked. Our early model helped show how private capital was eating up the housing supply while everyday people were being priced out.

Later, with Staged Releases, we mapped how land supply was being drip-fed into the market—not to match demand, but to protect prices. Again, a few key data points told a big story: approvals, completions, developer land banks, median price fluctuations, and release schedules. Maybe six to eight solid metrics.

Today, developers would be lucky to get away with only eight.

Because now, the system watches everything.