Find the pattern in a chain’s existing stores, then test a new address against it. A worked example is loaded for you, so there is something to read before you add anything. Every figure links to the dataset it came from. Configuring is free; building a pattern and scoring a live address use your account’s credits.
14 of 25 measures are consistent enough to score against.
See the pattern behind Dollar General, North Dakota.
It ships with the page, already built from 74 real store locations, so reading it costs nothing and needs no account. The prepared candidates on the next step are free to score as well. .
74 stores in 42 counties all in North Dakota Ward County 7, Williams County 5, Stark County 4, Grand Forks County 4, Richland County 4, McLean County 3, Mercer County 2, Cass County 2, Pembina County 2, Rolette County 2, Dickey County 2, Ransom County 2, Morton County 2, Walsh County 2, Traill County 2, Mountrail County 2, McKenzie County 2, Golden Valley County 1, Bowman County 1, Towner County 1, Foster County 1, Oliver County 1, Griggs County 1, Divide County 1, Ramsey County 1, LaMoure County 1, Grant County 1, Sargent County 1, Wells County 1, Stutsman County 1, Dunn County 1, Nelson County 1, Cavalier County 1, Burleigh County 1, Emmons County 1, Renville County 1, Hettinger County 1, Eddy County 1, Pierce County 1, Kidder County 1, Barnes County 1, McHenry County 1
7 of 74 sites sit in Ward County. County-grain traits are summarised one value per county, so that concentration does not dominate them.
Passing traffic. Closer to a main road means more of it.
nearest_major_road_distance_m
How many ways there are to reach the site on foot or by car.
roads_within_500m_count
Homes within a short trip. The everyday customer base.
housing_units_within_1km
How tightly those homes are packed in.
housing_units_density_per_km2
How much other trade is already drawing people to the area.
poi_count_1km
Distance to a grocery anchor, which pulls repeat trips.
nearest_grocery_store_distance_m
Floor area of the building on the site.
primary_building_footprint_sqm
People living in the census tract the site sits in.
tract_population
People moving in from elsewhere in the country.
county_net_domestic_migration · 42 counties
How much new housing the county approves each year.
county_building_permits_total_annual · 42 counties
How flat the ground is, which drives site preparation cost.
slope_degrees
FEMA flood zone, which affects insurance and permits.
within_floodplain_polygon
Federal Opportunity Zone, which can change the financing.
in_opportunity_zone
Unweighted, so this describes where these stores are, not where they do best. Add performance figures on the previous step to change that.
The stores in this pattern vary so much on these that no range would rule anything out. Using them would invent a preference the data does not show.
Every store returns one value, so it cannot explain why these sites were picked. Requirements in this group still block a candidate, they just do not pad the score.
Too few stores share one value to hold a candidate to it.
Describes the region these stores sit in, not a siting choice. Scoring it would mark down any candidate in a different part of the country.
The model asks for these but no such field exists, so they can never fill in. They are left out of the request rather than sent and silently returned empty.