Retail co-location ranking system

@@ -11,7 +11,7 @@ status: active
audience: customer-woodfine
bcsc_class: current-fact
language_protocol: PROSE-TOPIC
last_edited: 2026-08-25
last_edited: 2026-08-26
editor: pointsav-engineering
short_description: "The deterministic mechanics behind cluster ranking on the co-location platform — country-relative percentile axes, the overlap test between neighbouring clusters, and the tiebreak order applied within a tier."
paired_with: site-selection/co-location-ranking-system.es.md
@@ -19,25 +19,21 @@ paired_with: site-selection/co-location-ranking-system.es.md

The Woodfine [[co-location-methodology|co-location methodology]] assigns each cluster's tier through binary predicate gates, not a composite score — no cluster earns a tier by accumulating points toward a threshold. This article covers the mechanics behind those gates: how a cluster's catchment position is measured against its national peers, how competing clusters are compared for overlap, and how clusters are ordered once ranked. The gate definitions themselves — which combination of tests each tier requires — are set out in the [[catchment-ranking-methodology-v3|V3 catchment ranking methodology]]; the tier labels are described in [[co-location-tier-nomenclature|tier nomenclature]].

## Country-relative percentile axes
## Country-relative ranking

A cluster's tier depends on where it stands against every other cluster in its own country, not against a fixed global threshold. Each cluster is ranked against its national peers on eight measures: primary and secondary catchment population, and primary and secondary spend across grocery, hardware, and wholesale categories. A cluster's percentile on each axis is its rank divided by the country's total cluster count — a cluster in the top 10% by primary population carries a percentile of 0.10 on that axis.
A cluster's tier depends on where it stands against every other cluster in its own country, not against a fixed global threshold. Each cluster is ranked against its national peers on eight measures: primary and secondary catchment population, and primary and secondary spend across grocery, hardware, and wholesale categories. Ranking within-country, rather than against one global bar, keeps a smaller market's structure intact: a nationally significant cluster in a small country is read against its own national field, not swamped by the sheer scale of a larger one. The percentile calculation itself is planned for publication at gis.woodfinegroup.com.

Two catchment zones feed the population and spend axes: a primary zone within 35 km of the cluster and a secondary zone between 35 km and 150 km, per the [[od-catchment-methodology|O-D catchment methodology]]. Spend estimates draw on national household-spending surveys applied to the same population grid. Ranking within-country, rather than against one global bar, keeps a smaller market's structure intact against a larger one.
Two demand zones feed the population and spend axes: a primary zone within 35 km of the cluster and a secondary zone between 35 km and 150 km, per the [[od-catchment-methodology|distance-band methodology]]. Spend estimates draw on national household-spending surveys applied to the same population grid.

These thresholds are intentionally coarse. The system is built to separate nationally significant clusters from local nodes, not to finely rank clusters against one another within a tier.

## The overlap test

A cluster is only credited for its tier when it is not dominated by a stronger neighbour nearby. Overlap between two clusters is measured as the intersection-over-union (IoU) of two equal-radius disks, each 3.0 km, centred on the cluster anchors — the shared area between the disks, relative to their combined area. As two cluster centroids move apart, IoU falls toward zero and the clusters are treated as spatially independent. A cluster whose disk substantially overlaps a stronger cluster's disk is held below the tier its composition and catchment would otherwise earn. Regional carries the strictest overlap limit of any tier, per the [[catchment-ranking-methodology-v3|gate definitions]].
A cluster is only credited for its tier when it is not dominated by a stronger neighbour nearby. Overlap between neighbouring clusters is measured by a published geometric test over a fixed radius around each cluster: the closer two clusters sit, the higher the measured overlap, and clusters far enough apart are treated as spatially independent. A cluster that substantially overlaps a stronger cluster is held below the tier its composition and catchment would otherwise earn. Regional carries the strictest overlap limit of any tier, per the [[catchment-ranking-methodology-v3|gate definitions]]. The overlap measure and its radius are planned for publication at gis.woodfinegroup.com.

## Ordering within a tier

Clusters that share a tier and a country are ordered by three criteria, applied in sequence: store count within 3.0 km, then primary catchment population, both highest first. The cluster identifier breaks any remaining tie, guaranteeing a deterministic order.

## Provenance

- **Verification:** Ranking mechanics confirmed against the GIS platform's V3 scoring methodology.
Clusters that share a tier and a country are ordered by a fixed sequence of criteria, ending in a tiebreak that cannot produce a tie. The ordering is therefore fully deterministic: the same data returns the same order on every run, so a ranking cited in one review can be reproduced in the next. The ordering criteria and tiebreak sequence are planned for publication at gis.woodfinegroup.com.

## Data Sources