Woodfine Management Corp.
Retail Co-location Engine — Coverage Portfolio
PointSav GIS Platform · v0.4 · 29 May 2026
Page 1 of 2
§1  ·  Geographic Coverage Portfolio  ·  6,493 clusters  ·  18 countries  ·  24,403 anchor members
North America  ·  3,765 clusters
United StatesUS
3,104
10,011 anchor members
T1 889T2 1,545T3 670
CanadaCA
375
1,168 anchor members
T1 64T2 253T3 58
MexicoMX
286
1,028 anchor members
T1 68T2 33T3 185
United Kingdom  ·  457 clusters
Great BritainGB
457
3,543 anchor members
T1 22T2 191T3 244
Continental Europe  ·  2,111 clusters
GermanyDE
722
T1 227T2 282T3 213
FranceFR
624
T1 247T2 138T3 239
SpainES
218
T1 62T2 40T3 116
ItalyIT
177
T1 43T2 25T3 109
PolandPL
164
T1 53T2 74T3 37
NetherlandsNL
76
T1 19T2 16T3 41
AustriaAT
64
T1 21T2 7T3 36
PortugalPT
41
T1 8T2 18T3 15
GreeceGR
22
T1 5T2 4T3 13
IcelandIS
3
T1 0T2 0T3 3
Nordic  ·  160 clusters
SwedenSE
28
T1 11T2 5T3 12
DenmarkDK
67
T1 4T2 47T3 16
FinlandFI
55
T1 2T2 42T3 11
NorwayNO
10
T1 1T2 6T3 3
Source: PointSav GIS Platform · gis.woodfinegroup.com · OpenStreetMap (ODbL), Kontur Population (CC BY 4.0), US Census LODES 2021, GISCO LAU2 2021, StatCan 2021 T1 Regional  ■  T2 District  ■  T3 Local
Woodfine Management Corp.
Retail Co-location Engine — Methodology & Research Validation
PointSav GIS Platform · v0.4 · 29 May 2026
Page 2 of 2
§2  ·  Platform Scale
6,493
Co-location Clusters
18 countries · 24,403 members
1,746
T1 Regional
Hyper + Hardware + Price Club
2,726
T2 District
Hypermarket + Hardware
2,021
T3 Local
Single-format anchor
§3  ·  T2 Anchor Pattern

T2 District Composition (2,726 clusters)

Hypermarket + Hardware
2,590  ·  95.0%
Hyper + Hardware + Sport
136  ·  5.0%
95% of T2 clusters converge on a single anchor pair — Hypermarket + Hardware. This uniformity emerged from the geometry, not from a rule. It is a supply-side signal available before any demand panel is commissioned.
§4  ·  Build Pipeline — Four Stages From Open Data to Ranked Cluster Index
1
POI Ingest
Anchor retailers ingested from OpenStreetMap and Overture Maps using chain-specific Wikidata Q-codes and name-query fallbacks. Each chain validated against bounding-polygon and country-ISO filters to eliminate ghost records and cross-border contamination. 24,403 anchor POIs · 130 chains
2
DBSCAN Clustering
Haversine DBSCAN (ε = 300 m, min-samples = 2) groups anchor POIs into co-location clusters. The 300 m radius is calibrated to the median parking-lot diameter in the NA retail park typology — tight enough to exclude strip-mall adjacency, generous enough to capture split-lot campus layouts. 6,493 clusters across 18 countries
3
Tier Assignment
Each cluster is assigned a tier by anchor composition — the specific combination of format categories present. T1 requires Hypermarket + Hardware + Price Club; T2 requires Hypermarket + Hardware; T3 requires any single anchor category. Tier is a function of the supply-side mix, not of size or catchment. T1 1,746 · T2 2,726 · T3 2,021
4
Rank & Enrich
Within each tier, clusters are ranked by geometric span (convex-hull diameter) — smaller span = tighter co-location = higher rank. AEC data layers (ASHRAE climate zones, FEMA/EU flood, seismic SDC, solar irradiance, Köppen eco-regions) are joined at H3 resolution 7 (1.22 km²/cell). 8 AEC layers · 16 countries · open-licence

Open-data advantage: The cluster index is derived entirely from open-licence inputs — OpenStreetMap, Kontur Population (CC BY 4.0), US Census LODES, GISCO LAU2, StatCan. No proprietary mobility panel is required. The pipeline produces a supply-side location signal available before any demand data is commissioned, across 18 countries simultaneously.

§5  ·  Research Validation

Peer-Review Pipeline

  • Regression-validated taxonomy: R² = 0.503, N = 6,481 clusters, country fixed effects, ISO-clustered standard errors. T1 clusters are statistically 63% larger by span than T3 (β = 0.489, p < 0.001).
  • Seven-test falsification programme: span gate, catchment-size test, composition stability test, permutation test, cross-country replication, AEC co-variance check, and hold-out tier recovery. Full protocol in Working Paper v0.4.2.
  • 128-cell open-data audit (16 countries × 8 AEC layer types): the United States achieves regulatory-grade open coverage on all four Tier 1 AEC layers — climate zone, flood hazard, seismic design category, and solar resource.
  • Multi-vertical substrate: the same cluster index serves retail site selection, AEC site analysis, commercial real estate due diligence, and civic infrastructure planning — a single proprietary dataset with four addressable markets.
  • Two manuscripts in preparation for intended submission to peer-reviewed journals — Economic Geography and Construction Technology.
Woodfine, J. M., Woodfine, P. M., & Woodfine, M. (2026). Retail Anchor Co-location Composition as a Spatial Leading Indicator of Commercial Activity. Working Paper v0.4.2. Woodfine Management Corp., New York, New York.
Woodfine, J. M., Woodfine, P. M., & Woodfine, M. (2026). Open-Source Building-Systems Data Layers for Urban-Scale Site Analysis. Working Paper v0.2. Woodfine Management Corp., New York, New York.
Source: PointSav GIS Platform · gis.woodfinegroup.com · Data: OpenStreetMap (ODbL), Kontur Population (CC BY 4.0), US Census LODES 2021, GISCO LAU2 2021, StatCan 2021 All figures preliminary · Working papers in preparation for intended submission · CC BY 4.0