San Francisco permit timing visualization. Every figure comes from the city's public building-permit records for permits filed on or after January 1, 2021 and issued through July 2, 2026, counted as distinct permits after duplicate permit rows are collapsed. Time to first permit issuance is measured in whole completed 24-hour periods from filing. Key medians: accessory dwelling units 339 days and single-family homes 577 days. A separate matched comparison of newly built ADUs, every permit watched for exactly two years, found 60.1 percent of the 303 filed under the 120-day law had a recorded issuance within two years, versus 44.0 percent of the 334 filed under the 60-day law; the earlier group crossed half at day 560, the later group never reached half. Those figures use their own matched yardstick and are not comparable to the issued-only medians. The filing timeline shows one real ADU that finished right at the median, with direct identifiers omitted. Its four main milestones are Filed on Day 0, Reviews fan out on Day 38, Permit Center on Day 337, and Issued on Day 339. The source has 12 initial-routing rows and 11 labeled rows are shown. One row from a station with no plain-language name, spanning days 38–318 and one incomplete Building hold are omitted. All 12 source rows have blank review results, so this permit cannot illustrate comments, revisions, or review cycles. Nine recorded routing-step finish events appear on one spine across Building, Planning, Fire, Public Works, Water and Sewer, and Mechanical. A finish date means the routing step finished; it does not necessarily mean the station approved the plans. Seven source hold windows consolidate into three display unions. The opening union shares the solid filed-to-routing segment; the two later unions use amber stripes. The unions do not show that the whole permit was uniformly on hold or establish who caused a hold. Beside the spine, a thin vertical line for each fanned-out review station runs from its arrival day to its finish day, with amber segments marking that station's recorded holds. Among the 532 ADUs that eventually issued, 3, or 0.6 percent, had issued by filing day 60; 134, or 25.2 percent, by filing day 204; and 267, or 50.2 percent, by filing day 339. These figures are stated in this description rather than plotted as timeline markers, and they are not shares of every filed ADU or legal-compliance rates. For qualifying ADUs, California law generally requires the permitting agency to approve or deny the application within 60 days after receiving a completed application. DataSF does not report when an application became complete, so these records cannot measure compliance with that clock. A quarter of issued ADUs were issued in about 200 days and a quarter took over 590. The station chart spans each permit and station from earliest recorded arrival to latest recorded finish, including gaps and waiting; it is not continuous active staff time. Station medians range from roughly 163 to 263 days. Across 350 recorded planning-zoning hold windows, the median span was 96 days. The recent-issued analysis covers 325 ADUs issued January 1, 2024 through 2026-07-02, each verified against its project description. 299 of 325, or 92.0 percent, had at least one recorded comment date. 244 of 325, or 75.1 percent, had recorded comments at the same station on more than one date; that is 244 of 299, or 81.6 percent, among commented permits. Among those 299 permits, the median was 7 recorded station-comment dates across 4 stations. When the most recorded comment dates at one station was one, 55 permits had a median 29-day first-to-last comment window; two dates, 111 permits and 105 days; three dates, 67 permits and 215 days; four or more dates, 66 permits and 279 days. The chart excludes 26 permits with no recorded comment event. A recorded comment date is one unique initial-routing Issued Comments event per permit, station, and finish calendar date on or before issuance, after exact duplicate rows are removed. The frozen routing source contains 15966 raw rows, 14765 exact-unique rows, and 1201 duplicate extras. All 325 permits have at least one populated review result somewhere, but individual routing rows can be incomplete. The relationship between repeat comments and longer windows is descriptive, not causal, and incomplete routing records make recorded rates lower bounds. These counts include only permits that were issued; they do not estimate the share of all filed permits that finish. The project destination is /sandbox.
On desktop, tab to the story text and use arrow keys, Page Up, or Page Down to move through the narrative.