New: the hosted MCP server, sign-in without a key, and a Claude plugin · Methodology 1.12.6 is live.Changelog
Chalk cliffs and a long-exposure sea at dusk, a seaside town lit in the distance

The intelligence layer for place.

Sourced, dated facts for every neighbourhood in Great Britain, explained scores, change monitoring and plain-English questions. One API, one methodology, with the version on every response.

Why we exist

Every property and location product rebuilds the same pipeline.

A product team that wants neighbourhood context in its listings, its dataset or its customer portfolios starts the same way every time: a dozen publishers, each on its own calendar, its own geography and its own licence. ONS, police.uk, HM Land Registry, Ofsted, three deprivation indices, the flood agencies of three nations, NaPTAN, OpenStreetMap. Loading them is a month. Keeping them joined, dated and correct is a job that never ends.

We built that pipeline once, for everyone. Every neighbourhood in Great Britain on one permanent place ID, every value with its source, the period it describes and a confidence, every score explained, every change published with its cause. One API, one methodology, with the version on every response.

It exists so the team can spend its time on the product, not the plumbing: the listing, the search, the dataset, the assistant. Sourced by construction, explained on every response.

What we believe

Six principles, applied in code and stamped on every response. These are not marketing words. Each one is enforced in code, documented in the methodology, and visible on every response.

Deterministic at the source

The engine sets the number, not the AI. Every score, signal and percentile is computed by a versioned pipeline from official data, and every response names the version that produced it. AI sits on top as a planner and a query plane, never as the source of truth.

Methodology is public

Every dimension, every weight, every aggregation step is documented on the methodology page. The version is stamped on every response. If a number lands in your report, the trail is citable.

The AI writes the query, never the number

A question in plain English becomes a typed plan before anything runs. The plan is what executes, and your product can keep it and run it again later without an AI call, against the data as it stands that day. The model is auditable because it is not the executor.

Sample-size honest

When the underlying data is thin, confidence drops and the response says why. Price moves on two transactions never trigger a webhook. We would rather say less than say something wrong.

Versioned and published

Every response carries the methodology version that produced it. Every release is published with the figures it moved, and Monitor labels a change caused by a release as a method change, not a move in the area.

Country-scoped percentiles

England compared against England, Scotland against Scotland, Wales against Wales. Never a cross-border rank. Three official deprivation methodologies, three percentile spaces, by design.

1.12.6

Engine version on every response

4

Products on one engine

3

Nations, each on its own index

52,025

Places with a permanent ID

How it got built

Engine first, surfaces second, AI last.

OneGoodArea is built lean by a small team. The first phase was the engine: the ONS postcode spine, the three national deprivation methodologies, the price and crime time-series, the confidence rubric, the version stamp. The four products you see today, Signals, Scores, Monitor and Intelligence, all sit on that same core. Same numbers, four shapes.

AI joined the stack only after the engine was deterministic and the methodology was published. The planner emits a typed plan first; the query runs second; the model never touches the numbers. That ordering is the architecture, not a slogan. It is why a question asked through the AI becomes a plan your product can keep and run again, without an AI call, against the data as it stands that day.

Read the methodology

Talk to us

Procurement question? Methodology question? Building on top of us?

Whichever it is, the fastest channel is the one you already use.

Email

operation@onegoodarea.co.uk. We read everything that lands here.

X

@onegoodarea. Engineering and product updates as we ship.

LinkedIn

company/onegoodarea. Longer-form notes and hiring when we open roles.

See it on an address you know.

Book a pilot call, or open the playground and run a postcode yourself.