Methodology

How the tables on this site are built, so you can judge them.

Pipeline

1. A script downloads only whitelisted columns from each portal's public SODA API for January 2026 to September 2026 (plus records up to the retrieval date for the "newest records" tables).
2. Records are normalised to a common schema; dates use the date named on each dataset page (the issue date).
3. Free-text descriptions that contain an email address or phone number are withheld entirely (blank cell); exact duplicate rows are removed; text cells that start with =, + or @ get a leading apostrophe in the CSV so spreadsheets do not run them as formulas.
4. Each month page is generated from the same rows that will go into the paid CSV once it launches, so the record count on the page equals the number of rows in the file.

Freshness

This is a dated snapshot, not a live feed. Each page states when the data was retrieved. Cities add and amend records after the fact, so counts for recent months can rise and individual records can change.

Limits

• Dataset definitions differ by city (for example, whether trade permits are separate rows). Each dataset page says what a row is.
• Reported values are whatever the applicant or the city entered and include typos.
• Missing columns in a city's file stay blank; we do not guess.