Research methodology & provenance

1. Visible Sources under each chart or table: organization, URL, and as-of date. Every source is tracked in our own source list.
2. Page stamp: Last verified and datePublished (ISO date strings, never invented).
3. Prefer primary government and Hub-owned sources. Unverified press and third-party directories must not enter KPIs or map fills.
4. Unit of analysis is labeled: a license is not an open storefront, which is not an active listing, which is not a menu SKU.
5. Revision note when Hub or the agency later corrects a figure.

Registry: 58 sourced rows (51 metric-eligible). How to cite

Sources: Hub extract from MDH/DOR/OCM, Extracted tables index (98 CSVs), as of 2026-09-22 (Hub capture 2026-09-22). Extracted tables index covers products built from PDF sources · PRV-052 [hub_owned]

• Primary government: preferred for all metrics (42 registry rows).
• Hub-owned: Hub directory, openings, and menu samples when methodology is documented (9).
• Secondary press: leads only, labeled unverified, never used in KPIs (2).
• Unverified or method-only: excluded from metrics (5).

Critical Hub mismatches stay flagged as needing verification (directory counts, year-to-date sales, and an excise rate shown as 10% vs 15%). Both claims are preserved rather than silently reconciled.

Hub surfaces sometimes disagree. Research Desk products prefer the columns below, not stale marketing copy.

• directory / license counts: avoid “Older desert blog “96 licensed” style copy”; prefer Homepage/directory capture ~234 active listings (as of capture) or live DB count — never mix with OCM licenses issued.
• sales YTD: avoid “Homepage teaser ~$36.9M”; prefer Monitor archive Jan–Jul 2026 combined ≈ $163.6M (7 months); July monthly ≈ $27.5M.
• excise tax: avoid “10% excise examples on old pages”; prefer 15% Minnesota cannabis excise on Research Desk / tax tools.

• Sales: /market-data/sales + Monitor archive (not homepage teasers).
• Openings: Opening Tracker and confirmed-open directory status, not “licenses issued.”
• Directory size: the live count, not hardcoded blog figures.
• Excise tax: 15% state cannabis excise. Stale 10% examples are not republished on research pages.

• Homepage claims (context only): ~234 dispensaries, 124 cities, 28,559 reviews.
• Opening tracker snapshot: 6 recently opened + 6 coming soon.
• Sales: July 2026 monthly ≈ $27.5M; Jan–Jul YTD ≈ $163.6M (7 mo). Aug/Sep Monitor months are still missing, not invented.
• Prices: 19 menu samples logged. Dollar rows are pending and not invented.

• DOR cannabis aid map: /research/local-government-cannabis-aid matches DOR county names to each county's official boundary for the map.
• Open storefront points: Hub confirmed-open locations only, on /research/open-storefront-pipeline. Licenses are never plotted as open.
• Zoning / Social Equity embeds: We show the official Minneapolis and St. Paul zoning and equity maps directly from each city, with attribution. A simpler fallback view is used if the interactive map is not available, on /research/zoning-friction.

• Patient-level medical registry data
• Invented Market Monitor months (Aug/Sep 2026 gaps stay gaps)
• Invented DOR aid dollars or MDH percentages
• Invented menu prices ahead of observation CSV rows
• KPI metrics sourced only from secondary press or unverified directories
• ArcGIS layers claimed as Hub-owned without city or OCM attribution
• POS “market share” claims without retailer contracts
• Paid data access: only research APIs are offered today, pending legal review