Chattanooga holds pothole requests to a 5-working-day target, and the pothole_data_feed dataset records for each request whether it was finished by its due date. Across 16,385 closed requests from 2017 through 2024, the share finished late rose from 10.4% in 2017 and 10.2% in 2018 to 24.6% in 2019 and 35.8% in 2020, then fell to 23.0% in 2021, 5.0% in 2022, 3.3% in 2023 and 1.9% in 2024. Median repair time moved the same way: 1 working day in 2017, 3 in 2019, 4 in 2020, and 2 in 2024. The worst stretch was July through September 2020, when 65.9% of requests ran late and the median request took 8 working days. The long tail stretched too: the slowest 5% of 2020 requests took 14 working days or more, against 8 in 2019 and 9 in 2021.

The swing is not a change in what was being counted. The 5-working-day target is recorded identically on all 16,385 requests, so the deadline itself never moved. Holding request priority constant at EMERGENCY, the dominant code, late share still runs 10.4% in 2018, 25.1% in 2019, 36.8% in 2020 and 1.9% in 2024. Holding the responsible department constant at PW - Streets, which handles 92 to 97% of requests every year, the same shape appears. The share of requests marked URGENT, which are slower than EMERGENCY ones, actually fell from 48.3% in 2019 to 22.4% in 2020, which would have made 2020 look faster rather than slower. All nine council districts were late between 29.7% and 41.3% of the time in 2020, so the slowdown was citywide rather than concentrated in one area. Bucketing by the date a request came in, the date it was completed, or the date it was due all give the same yearly figures, and restricting every year to January-September requests so each year gets an equal observation window also holds: 36.9% late in 2020 against 1.9% in 2024.

Two limits. The last completion recorded in this dataset is 2024-10-04, so these figures say nothing about 2025 or 2026. And they cover only the 16,385 requests logged under the service-request system that begins in late 2016; the 13,159 older pothole records, which run from 2003 to 2016, carry no due date or completion-time field and cannot be compared. Whether the 2019-2021 slowdown reflects crew capacity, weather, pavement condition or changes in how work was logged is not something this dataset can settle.