I built the tool I wished existed.

The gap between policy and practice

I’ve sat on every side of the planning table. Designing schemes as an architect. Assessing them at a London council. Acquiring sites and managing planning as a developer.

At every stage, I saw the same problem: published policy tells you what a borough wants, but it doesn’t tell you what it actually approves. The gap between the two is where schemes fail, money is wasted, and homes don’t get built.

Small site developers — the people responsible for most of London’s new housing — are making six-figure acquisition decisions based on policy documents, market chatter, and instinct. They deserve better data.

What Perfect Scale does

We analyse every determined small-site application in a borough over 3–5 years. Every application is coded by type, area, density, and outcome. Every officer report is read. Every finding is evidence-graded.

The result is a planning intelligence report that tells you, for any given site type and location, what density has actually been approved, what the refusal risks are, and where the viability cliffs sit.

No hand-waving. No unqualified claims. Four evidence tiers, from Robust to Anecdotal, so you know exactly how much weight to put on each finding.

Why evidence grading matters

The property industry is full of claims without evidence. “This area approves density.” “Pre-app is a waste of time.” “Conservation areas always get refused.”

Some of these turn out to be true. Some are completely wrong. The only way to tell is to look at the data. Every Perfect Scale finding comes with a sample size, a statistical test where appropriate, and an evidence tier. If the data is thin, we say so.

This matters because developers are making decisions worth hundreds of thousands of pounds. They need to know the difference between a finding backed by 200 applications and one based on a handful.

Methodology

Each report is built from a complete dataset of determined applications, sourced from the GLA Planning Datahub. We use a combination of automated data processing and AI-assisted document analysis, with manual verification at every stage.

Our density model is backtested against every approved application before publication. For our first borough, the model predicted within ±1 unit of the actual approved scheme 73% of the time, and within ±2 units 90% of the time, with near-zero prediction bias.

We grade accuracy by evidence tier: Robust cells are 37% more accurate than Anecdotal, giving you a clear signal of where to trust the numbers most.