W3C TAG design-review closure latency

Snapshot: 2026-10-19T22:13:04Z. 1,197 issues; 1,149 closed; 48 open. Excludes 43 pull requests.

All-time mean: 210.8 days; median: 117.4 days; 90th percentile: 511.2 days.

Important scope effect: the bot includes charter reviews and questions. Excluding those labeled issues, the 2025 and 2026 YTD closure-year medians are both about 89 days and means are about 211 and 214 days. Thus the apparent 2025-to-2026 improvement in the all-issues average does not persist under this scope restriction. This is a label-based sensitivity analysis, not a hand-classification of every issue.

Long-term improvement is not monotonic. Among non-charter, non-question filing cohorts, the censoring-adjusted median is about 76 days for 2025 and 112 days for 2026 YTD. The latter is preliminary: 35 of 66 requests remain open, and the estimate assumes filing-date-related censoring is independent of eventual duration within the cohort. The 2026 closed-only filing-cohort mean of 54 days is not a valid estimate of the eventual mean for all 2026 requests.

Method and interpretation

Latency is elapsed calendar days between created_at and the current closed_at timestamp. It includes external waiting and any closed intervals before reopening; it is not staff effort or a measure of cumulative time in the open state. No individual closure-actor attribution is made.
Baseline matches the bot metrics page: all issues in w3ctag/design-reviews, regardless of labels. Closure does not necessarily mean a substantive review or approval: declined, withdrawn, timed-out, charter and question issues are included.
Closure-year results answer how long the issues closed that year took. Filing-year results answer how long a cohort of requests took. Recent closed-only filing cohorts favor quick completions; Kaplan-Meier accounts for administrative right censoring of open issues. It assumes censoring is independent of eventual duration within each cohort. Survival medians are estimates without confidence intervals, not forecasts.
For a filing cohort, fixed-horizon completion percentages are withheld if any still-open issue has less follow-up than that horizon. Percentiles of completed issues use linear interpolation. Kaplan-Meier uses the first time estimated survival reaches 0.5, which can differ slightly from an interpolated sample median even without censoring.
Annual numbers are issue-weighted, not averages of monthly averages. Sensitivity subsets use labels as observed at snapshot time; label coverage and review mix change over time. These are descriptive trends, not causal claims about TAG productivity.

Sources: review repository; bot metrics source; bot issue query. Raw API metadata and reproducible Python analysis are saved alongside this report.

By closure year

Closed in	Closed issues	Mean days	Median days	P90 days	>365 days
2013	1	73.0	73.0	73.0	0
2014	15	300.2	323.4	455.6	6
2015	30	175.8	155.1	450.6	4
2016	48	199.7	121.2	418.7	8
2017	64	185.5	101.3	479.0	11
2018	75	224.7	147.6	455.7	13
2019	117	298.5	235.5	610.2	37
2020	115	220.4	110.9	502.6	19
2021	148	220.1	153.4	505.6	30
2022	96	190.1	124.3	416.5	14
2023	93	164.8	83.1	320.8	9
2024	111	228.0	125.9	497.1	22
2025	150	190.8	79.2	518.9	21
2026 YTD	86	159.1	62.3	403.9	9

By filing year

Filed in	Filed	Still open	Closed-only mean	Closed-only median	KM median	Closed <=180 days %	Closed <=365 days %
2013	13	0	414.0	445.9	445.9	7.7	38.5
2014	18	0	321.8	235.4	223.5	33.3	72.2
2015	53	0	343.4	188.8	188.8	45.3	71.7
2016	48	0	273.5	163.7	161.0	52.1	72.9
2017	70	0	212.5	117.4	116.3	65.7	81.4
2018	110	0	326.0	247.0	245.6	36.4	64.5
2019	120	0	281.6	190.1	186.7	48.3	79.2
2020	127	0	233.4	133.0	133.0	59.1	80.3
2021	105	0	156.8	107.7	107.7	66.7	90.5
2022	88	0	193.1	118.2	117.4	69.3	83.0
2023	117	0	206.2	132.0	132.0	63.2	78.6
2024	94	0	136.4	91.3	90.1	72.3	92.6
2025	136	7	95.2	55.6	62.1	77.2	Not estimable
2026 YTD	98	41	49.7	39.6	75.8	Not estimable	Not estimable

Smoothed closure trend

Sensitivity to issue selection

Closed in	Subset	Closed issues	Mean days	Median days
2018	all_issues	75	224.7	147.6
2018	exclude_charters_and_questions	75	224.7	147.6
2018	substantive_resolution_labels_only	2	155.2	155.2
2019	all_issues	117	298.5	235.5
2019	exclude_charters_and_questions	117	298.5	235.5
2019	substantive_resolution_labels_only	29	192.4	147.5
2020	all_issues	115	220.4	110.9
2020	exclude_charters_and_questions	115	220.4	110.9
2020	substantive_resolution_labels_only	59	167.8	83.2
2021	all_issues	148	220.1	153.4
2021	exclude_charters_and_questions	148	220.1	153.4
2021	substantive_resolution_labels_only	105	179.8	125.3
2022	all_issues	96	190.1	124.3
2022	exclude_charters_and_questions	96	190.1	124.3
2022	substantive_resolution_labels_only	76	175.4	111.9
2023	all_issues	93	164.8	83.1
2023	exclude_charters_and_questions	93	164.8	83.1
2023	substantive_resolution_labels_only	69	140.8	72.7
2024	all_issues	111	228.0	125.9
2024	exclude_charters_and_questions	111	228.0	125.9
2024	substantive_resolution_labels_only	82	177.2	104.4
2025	all_issues	150	190.8	79.2
2025	exclude_charters_and_questions	133	211.0	88.7
2025	substantive_resolution_labels_only	104	152.3	77.1
2026 YTD	all_issues	86	159.1	62.3
2026 YTD	exclude_charters_and_questions	58	213.8	88.8
2026 YTD	substantive_resolution_labels_only	53	110.0	66.2

Filing cohorts excluding charter and question issues

Filed in	Filed	Still open	Closed-only mean	Closed-only median	KM median
2013	13	0	414.0	445.9	445.9
2014	18	0	321.8	235.4	223.5
2015	53	0	343.4	188.8	188.8
2016	48	0	273.5	163.7	161.0
2017	70	0	212.5	117.4	116.3
2018	110	0	326.0	247.0	245.6
2019	120	0	281.6	190.1	186.7
2020	127	0	233.4	133.0	133.0
2021	105	0	156.8	107.7	107.7
2022	88	0	193.1	118.2	117.4
2023	117	0	206.2	132.0	132.0
2024	94	0	136.4	91.3	90.1
2025	117	7	105.6	70.5	76.2
2026 YTD	66	35	53.6	42.5	111.6

Current backlog

48 open issues; mean age 163.0 days; median age 133.5 days. 19 older than 180 days; 3 older than 365 days.