Diff: reference/colocation-tier-nomenclature
From 1c02ec1 to 1c02ec1
+3 / −6 lines
| Before | After |
|---|---|
| --- | --- |
| schema: foundry-doc-v1 | schema: foundry-doc-v1 |
| title: "Co-location Tier Nomenclature" | title: "Co-location Tier Nomenclature" |
| slug: colocation-tier-nomenclature | slug: colocation-tier-nomenclature |
| short_description: "Definition and criteria for the T1 Regional, T2 District, and T3 Local classification tiers applied to every co-location cluster, including the two-pass DBSCAN algorithm and the Change B span gate." | short_description: "Definition and criteria for the T1 Regional, T2 District, and T3 Local classification tiers applied to every co-location cluster, including the two-pass DBSCAN algorithm and the Change B span gate." |
| category: reference | category: reference |
| type: topic | type: topic |
| content_type: topic | content_type: topic |
| quality: complete | quality: complete |
| status: active | status: active |
| audience: vendor-public | audience: vendor-public |
| bcsc_class: public-disclosure-safe | bcsc_class: public-disclosure-safe |
| language_protocol: PROSE-TOPIC | language_protocol: PROSE-TOPIC |
| last_edited: | last_edited: 2026-06-16 |
| editor: | editor: editorial |
| paired_with: colocation-tier-nomenclature.es.md | paired_with: colocation-tier-nomenclature.es.md |
| cites: [] | cites: [] |
| --- | --- |
| The co-location platform | The co-location platform assigns every identified retail cluster to one of three tiers — T1 Regional, T2 District, and T3 Local — based on anchor composition and geometric span. T1 clusters anchor on a warehouse club or a hypermarket-plus-hardware combination and span no more than 2.5 km; T2 clusters carry a hypermarket anchor but fall short of T1 on composition or span; T3 clusters contain significant co-location in the absence of hypermarket formats. As of Phase 23+Change B, the dataset carries 1,746 T1, 2,726 T2, and 2,021 T3 clusters across 18 countries. |
| three tiers: T1 Regional, T2 District, and T3 Local. This article defines what | |
| each tier means, explains how the classification is determined, and documents | |
| the evolution of the naming convention. | |
| ## The Three Tiers | ## The Three Tiers |
| **T1 Regional** denotes the highest-order co-location clusters: those anchored | **T1 Regional** denotes the highest-order co-location clusters: those anchored |
| by the largest retail formats in the market. A cluster reaches T1 when it | by the largest retail formats in the market. A cluster reaches T1 when it |
| contains at least one warehouse club — a membership-based bulk retailer such | contains at least one warehouse club — a membership-based bulk retailer such |
| as Costco, Sam's Club, or PriceSmart — or when it contains a full-format | as Costco, Sam's Club, or PriceSmart — or when it contains a full-format |
| hypermarket accompanied by a hardware anchor. The hypermarket-plus-hardware | hypermarket accompanied by a hardware anchor. The hypermarket-plus-hardware |
| combination is required for T1 designation because a hypermarket without a | combination is required for T1 designation because a hypermarket without a |
| hardware complement indicates a smaller retail catchment than the combined | hardware complement indicates a smaller retail catchment than the combined |
| offer. As of Phase 23+Change B, the platform identifies 1,746 T1 clusters | offer. As of Phase 23+Change B, the platform identifies 1,746 T1 clusters |
| across 18 countries. | across 18 countries. |
| **T2 District** denotes clusters that meet the hypermarket anchor threshold but | **T2 District** denotes clusters that meet the hypermarket anchor threshold but |
| do not qualify for T1. This encompasses two distinct situations. A cluster that | do not qualify for T1. This encompasses two distinct situations. A cluster that |
| contains a hypermarket and hardware anchor but whose member locations span more | contains a hypermarket and hardware anchor but whose member locations span more |
| than 2.5 km from the geometric centre to the most distant member is classified | than 2.5 km from the geometric centre to the most distant member is classified |
| as T2 rather than T1. The geometric span gate — introduced in Change B — reflects | as T2 rather than T1. The geometric span gate — introduced in Change B — reflects |
| the finding that very widely dispersed clusters, even when they contain equivalent | the finding that very widely dispersed clusters, even when they contain equivalent |
| anchor composition to T1 clusters, represent different consumer catchment | anchor composition to T1 clusters, represent different consumer catchment |
| dynamics. Tighter clusters indicate denser co-location effects. A cluster that | dynamics. Tighter clusters indicate denser co-location effects. A cluster that |
| contains a hypermarket but lacks a hardware anchor is also classified as T2, | contains a hypermarket but lacks a hardware anchor is also classified as T2, |
| regardless of its geometric span. The platform currently identifies 2,726 T2 | regardless of its geometric span. The platform currently identifies 2,726 T2 |
| clusters. | clusters. |
| **T3 Local** denotes clusters that contain significant retail co-location but | **T3 Local** denotes clusters that contain significant retail co-location but |
| lack anchor formats sufficient for T1 or T2 designation. T3 clusters typically | lack anchor formats sufficient for T1 or T2 designation. T3 clusters typically |
| contain a mix of mid-format grocery, electronics, and lifestyle retailers without | contain a mix of mid-format grocery, electronics, and lifestyle retailers without |
| a full hypermarket presence. They represent locally important retail nodes rather | a full hypermarket presence. They represent locally important retail nodes rather |
| than the regional and district drawing-power markets of T1 and T2. The platform | than the regional and district drawing-power markets of T1 and T2. The platform |
| currently identifies 2,021 T3 clusters. | currently identifies 2,021 T3 clusters. |
| ## The Two-Pass DBSCAN Algorithm | ## The Two-Pass DBSCAN Algorithm |
| Tier classification is not assigned independently of cluster discovery: the | Tier classification is not assigned independently of cluster discovery: the |
| tier emerges from the geometry and composition of clusters identified by the | tier emerges from the geometry and composition of clusters identified by the |
| two-pass DBSCAN algorithm. | two-pass DBSCAN algorithm. |
| The first pass uses anchor stores as seeds. DBSCAN (Density-Based Spatial | The first pass uses anchor stores as seeds. DBSCAN (Density-Based Spatial |
| Clustering of Applications with Noise) groups points that are within a defined | Clustering of Applications with Noise) groups points that are within a defined |
| radius of each other into clusters. In the first pass, only the anchor-category | radius of each other into clusters. In the first pass, only the anchor-category |
| stores — warehouse clubs and full hypermarkets — are used as seed points. This | stores — warehouse clubs and full hypermarkets — are used as seed points. This |
| pass identifies the cores around which high-order retail clusters form. | pass identifies the cores around which high-order retail clusters form. |
| The second pass fills each identified core cluster with additional retail | The second pass fills each identified core cluster with additional retail |
| locations within the cluster's geometric reach. Hardware anchors, lifestyle | locations within the cluster's geometric reach. Hardware anchors, lifestyle |
| retailers, electronics stores, and other category members are added when they | retailers, electronics stores, and other category members are added when they |
| fall within the radius of an existing seed cluster. The composition of the filled | fall within the radius of an existing seed cluster. The composition of the filled |
| cluster — which anchor categories are present — determines the tier. | cluster — which anchor categories are present — determines the tier. |
| The geometric span measurement used in the Change B gate is taken after the | The geometric span measurement used in the Change B gate is taken after the |
| second pass, when the cluster's full member set is known. Span is measured as | second pass, when the cluster's full member set is known. Span is measured as |
| the maximum distance between any two member locations within the cluster. | the maximum distance between any two member locations within the cluster. |
| ## The Span Gate: Change B | ## The Span Gate: Change B |
| The 2.5 km span gate was introduced in Change B (2026-05-28). | The 2.5 km span gate was introduced in Change B (2026-05-28). |
| Prior to Change B, tier was determined solely by anchor composition: if a cluster | Prior to Change B, tier was determined solely by anchor composition: if a cluster |
| contained the right anchor mix, it was T1, regardless of how spread out its | contained the right anchor mix, it was T1, regardless of how spread out its |
| members were. The gate corrected for a finding that a subset of T2-composition | members were. The gate corrected for a finding that a subset of T2-composition |
| clusters had span values well above 2.5 km and were representing geographically | clusters had span values well above 2.5 km and were representing geographically |
| diffuse retail distributions rather than the tightly co-located patterns that | diffuse retail distributions rather than the tightly co-located patterns that |
| define the market analysis use case. | define the market analysis use case. |
| The Change B span gate demoted approximately 667 clusters from T2 to T3, shifting | The Change B span gate demoted approximately 667 clusters from T2 to T3, shifting |
| the T2 count from approximately 3,393 (Phase 22) to 2,726 (Phase 23+Change B) and | the T2 count from approximately 3,393 (Phase 22) to 2,726 (Phase 23+Change B) and |
| increasing T3 from approximately 1,354 to 2,021. | increasing T3 from approximately 1,354 to 2,021. |
| ## Naming History | ## Naming History |
| The T1/T2/T3 nomenclature was adopted to provide a stable, non-hierarchical | The T1/T2/T3 nomenclature was adopted to provide a stable, non-hierarchical |
| labelling system that conveys market function without implying a simple rank | labelling system that conveys market function without implying a simple rank |
| ordering. Earlier internal development used placeholder labels — Alpha, Beta, | ordering. Earlier internal development used placeholder labels — Alpha, Beta, |
| and Gamma — which were retired before any public output was produced. The T1/T2/T3 | and Gamma — which were retired before any public output was produced. The T1/T2/T3 |
| labels have been canonical since Phase 18 and appear in all published data files, | labels have been canonical since Phase 18 and appear in all published data files, |
| wiki articles, and academic manuscripts. | wiki articles, and academic manuscripts. |
| The descriptors — Regional, District, Local — are intended to convey catchment | The descriptors — Regional, District, Local — are intended to convey catchment |
| geography, not administrative hierarchy. A T1 Regional cluster typically serves | geography, not administrative hierarchy. A T1 Regional cluster typically serves |
| a regional catchment extending 20–50 km; a T2 District cluster a more localised | a regional catchment extending 20–50 km; a T2 District cluster a more localised |
| catchment; a T3 Local cluster a neighbourhood or small-town catchment. These are | catchment; a T3 Local cluster a neighbourhood or small-town catchment. These are |
| empirical generalisations, not definitional constraints. | empirical generalisations, not definitional constraints. |
| ## Counts by Geography (Phase 23+Change B) | ## Counts by Geography (Phase 23+Change B) |
| The platform covers 6,493 clusters across 18 countries. T1 clusters are most | The platform covers 6,493 clusters across 18 countries. T1 clusters are most |
| common in the United States (889), Germany (227), and France (247). T2 clusters | common in the United States (889), Germany (227), and France (247). T2 clusters |
| are most common in the United States (1,779), Germany (338), and Great Britain | are most common in the United States (1,779), Germany (338), and Great Britain |
| (400). The full per-country breakdown appears in the [co-location summary](https://gis.woodfinegroup.com/research-summary.html). | (400). The full per-country breakdown appears in the [co-location summary](https://gis.woodfinegroup.com/research-summary.html). |
| ## References | ## References |
| - [DBSCAN](https://en.wikipedia.org/wiki/DBSCAN) — Wikipedia, accessed 2026-06-14 | - [DBSCAN](https://en.wikipedia.org/wiki/DBSCAN) — Wikipedia, accessed 2026-06-14 |
| --- | --- |
| *Data provenance:* clusters-meta.json Phase 23+Change B (2026-05-29T05:00Z rebuild); | *Data provenance:* clusters-meta.json Phase 23+Change B (2026-05-29T05:00Z rebuild); |
| platform taxonomy definitions. | platform taxonomy definitions. |