Method & sources
Every number, cited. Every limit, disclosed.
Most location reports ask you to trust the vendor. Ours asks you to check our work. This page lists every source we use, its vintage, its geography level, and — just as deliberately — the things we refuse to do.
Public data only
Every input is public, free, and licensed for commercial use. A counterparty can dispute an opinion; a named Census figure is available for both sides to check.
Cited line by line
Each section of a report names its source and vintage. You can hand the document to your attorney, lender, landlord, or client, and every figure survives the follow-up question.
Failure is disclosed
Reports are generated fail-soft: a source outage never silently degrades the analysis. The Data Quality section tells you what responded and what did not.
Source register
What we build on
| Source | Role in the report | Vintage | Geography | License |
|---|---|---|---|---|
| US Census Bureau — ACS 5-year | US demographics: population, households, income, age, rent | 2023 5-year estimates | Census tract, aggregated over the trade area | Public domain |
| US Census Bureau — TIGERweb | US geographic boundaries and geography lookup | Current TIGER release | Tract, county, state | Public domain |
| Statistics Canada — 2021 Census Profile | Canadian demographics at neighbourhood scale | 2021 Census (SDMX API); 2026-census readiness monitored monthly | Dissemination area — 57,936 boundaries held locally, with city-level fallback | Open Government Licence (Canada) |
| Overture Maps Foundation — Places | Points of interest, chains, competition, co-tenancy | Monthly releases, pinned per report (~50 million places) | Coordinates, US + Canada | CDLA-Permissive-2.0 |
| US Census Bureau — LEHD LODES | Daytime population: jobs located inside the trade area | Latest LODES workplace-area release | Census block, aggregated (US only) | Public domain |
| FBI — Crime Data Explorer | US reported-crime index, benchmarked to the national average | Latest full reporting year | Reporting agency (nearest agency to the site) | Public domain |
| Statistics Canada — Crime Severity Index | Canadian reported-crime index | Latest annual release | Municipality | Open Government Licence (Canada) |
| NCES — EDGE public school locations | US public-school proximity | 2023–24 school year | Point locations | Public domain |
| Valhalla routing engine | Drive-time trade areas: 5/10/15-minute isochrones | Computed per report on current road network | Road-network polygon (auto, pedestrian, bicycle) | Open source |
| City of Toronto — traffic count program | Calibrates our modeled street volumes against 238 measured intersections | Latest published counts | Intersection points (calibration city) | Open Government Licence (Toronto) |
| US DOT/BTS — National Transportation Noise Map | US ambient noise exposure (all transportation modes) | 2020 model (30-metre grid), tract aggregation per Seto & Huang 2023 | Census tract, population-weighted over the trade area | Public domain |
| NBER Working Paper 34298 (Moretti & Wheeler 2025) | Noise capitalization methodology: ~0.95% of home value per dB above the quiet threshold | September 2025 | Applied to local exposure and median home value | Cited methodology |
| FHFA — annual house price index (developmental) | US home-price appreciation trend (1/3/5-year) | Annual, through latest release | Census tract (absent where repeat-sales are thin — disclosed) | Public domain |
| Statistics Canada — New Housing Price Index (18-10-0205) | Canadian price trend — new construction only, and labeled as such | Monthly | Metro area | Open Government Licence (Canada) |
| US Economic Census 2022 (ecnbasic) | Retail sales actuals for the supply side of the Retail Gap — economic-place grain where published, establishment-allocated county fallback, demand-only rows where disclosure rules suppress | 2022 (final; next census ~2029) | Economic place / county | Public domain |
| BLS Consumer Expenditure Survey — Table 1203 | Household spending by income bracket, folded through the local ACS income distribution for the demand side of the Retail Gap (per consumer unit) | 2024 (annual) | National profile × local income mix | Public domain |
| Statistics Canada — Survey of Household Spending (11-10-0222) | Canadian demand side of the Retail Gap: provincial spending pattern × local household composition — always labeled as coarser than US county grain | 2023 (biennial) | Province, indexed to local median income (clamped 0.6–1.6) | Open Government Licence (Canada) |
| Statistics Canada — Retail trade sales (20-10-0056) | Canadian supply side of the Retail Gap: trailing-12-month unadjusted sales, allocated by household share | Monthly | Province (9 CMAs available) | Open Government Licence (Canada) |
| ACS B03002 + B02015 / StatCan visible minority & immigration | Origins & communities: ethnicity distribution vs benchmark, detailed Asian origins (e.g. Asian Indian, Chinese, Filipino), immigrant share — the community-fit layer for culturally specific concepts | ACS 2023 5-yr / Census 2021 (25% sample) | US tract / CA CSD & DA | Public domain / Open Government Licence (Canada) |
| Census LEHD LODES — origin-destination | Where a market's workers live: top home tracts, inflow, out-of-state share (main + aux files — the aux file alone is ~14% of origins) | 2023 (LODES8) | Census block, aggregated to tract | Public domain |
The Overture Maps release used is pinned in each report and stated in its citations, so two reports on the same site are comparable month over month.
Deliberate refusals
What we will not do
A method is defined as much by its refusals as its sources.
We do not project revenue.
We report measurable inputs — demographics, daytime population, competition, traffic, and noise — then show break-even conditions from the buyer's assumptions. We stop where measurement stops.
We do not sell foot-traffic ‘measurements’ we cannot verify.
Phone-ping foot-traffic panels are estimates built on undisclosed samples with undisclosed bias. Rather than resell someone's black box, we use observable proxies — the composition and churn of places themselves — and we label them as proxies.
We do not use black-box scores.
A score is printed beside the factors that moved it, the source vintages, and a confidence rating. The public method explains what each factor measures without publishing the protected scoring recipe.
We do not quote a number without its vintage.
A median income means nothing without knowing it is an ACS 2023 5-year estimate. Every figure in every report carries its source and vintage. When a source fails, the report says so instead of silently substituting.
Known limits
Stated plainly
If a limitation matters to your decision, you should hear it from us first.
- Street traffic volumes are modeled, and we publish the fit: calibrated against 238 measured Toronto intersections (pedestrian r² 0.39), with wider uncertainty bands outside the calibration region. Ranges, not guarantees.
- Canadian noise figures are a modeled proxy (no public noise map exists in Canada). We calibrated it against the real US noise map at 31 metro sample points: after offset correction the typical error is ±5.3 dB, and cross-neighbourhood ranking power is limited — treat proxy decibels as coarse bands, not measurements. The accuracy file ships in the repo alongside the model. The US map itself is a 2020 model with ±1.5 dB validated accuracy, not a microphone.
- House-price trend is tract-level in the US but absent where repeat-sales are thin (e.g., condo-dominated cores); in Canada it reflects new construction only. Both cases are labeled, never interpolated.
- Daytime population is US-only for now. No Canadian source currently meets our free-and-open licensing bar; when one does, it ships. We will not quietly fill the gap with a modeled guess.
- POI counts are reported as a lower bound when a query hits the source's result cap — and flagged as such in the Data Quality section.
- US crime indices are agency-level, not block-level. We index the nearest reporting agency against the national average and say exactly that.
- Retail Gap dollars are modeled estimates rounded to $10k (US) / $50k (CA) — comparing demand modeled from resident spending against supply measured (US) or allocated (CA) from sales data. Capture rates above 100% mean the market imports spending from the surrounding region — common in county seats — not an error. Where Census disclosure rules suppress a category's sales, we print demand and the live venue count and say the supply is unpublishable; we never fabricate it.
- CEX 'Entertainment' and 'Personal care' spending include services beyond store retail, so demand for those gap categories runs high relative to their retail-only sales — disclosed here rather than trimmed to look precise.
- Canadian demographics use the 2021 Census — the most recent available. We monitor Statistics Canada monthly and will move to the 2026 Census when its profile tables publish.
If this is the standard you want applied to your location, the report costs $149.
Get the cited report — $149