Review Turnaround

status="ready" · 6 reviewers, 96 first reviews, 4 pull requests still waiting. Rows are ordered by the longest wait, so the bottleneck is the top row: Priya Raman maintains the payments package alone and everything that touches it queues behind one calendar. The accent, the "slowest" mark and the dashed team-median rule all point at the same row. Pick any reviewer to retarget the big number to their median, and pick again to go back to the team.

Review turnaround · Platform

Pull requests marked ready for review in the last 30 days · main

9.1h

median time to first review

pooled across 96 reviews from 6 reviewers · p90 34.9h · merges in 19.8h · 4 still waiting

Priya Raman is slowest — 4.5× the team median

team median 9.1h

Review turnaround · Platform: every reviewer with their review count, their share of the window, the median and p90 time to first review, how that compares with the team median, the median time to merge, and how many pull requests are still waiting for a first look.
Reviewer	Reviews	Share	Median to first review	p90 to first review	Against the team median	Median to merge	Waiting
Priya Raman	12	12%	41 hours	2.7 days	4.5 times	2.3 days	3
Ade Okonkwo	17	18%	11.4 hours	16.3 hours	1.2 times	25.9 hours	0
Tom Whitfield	10	10%	9.3 hours	16.2 hours	1.0 times	24.9 hours	0
Marcus Bell	22	23%	8.3 hours	17.2 hours	0.9 times	19.7 hours	0
Sofia Ruiz	16	17%	5.9 hours	11.9 hours	0.6 times	12.3 hours	0
Lena Fischer	19	20%	4.7 hours	12.2 hours	0.5 times	19.4 hours	1

time to first reviewthen review to mergemedian to p90team median

Medians and the p90 tail use linear interpolation between closest ranks (type 7). The team median pools every first review in the window rather than averaging the per-reviewer medians, so a heavy reviewer weighs more than a light one. Shares are whole percentages apportioned by largest remainder, so the column adds to 100. Pull requests still waiting for a first review are counted beside the name and left out of every percentile — an unfinished wait can only get longer.

Pick a reviewer to lead the card with their median, and pick them again to go back to the team. The accent marks whichever row the card is reading.

status="empty" · a valid payload carrying no reviewers at all, told apart from "loading" and from "failed to fetch".

Review turnaround · Platform

Pull requests marked ready for review in the last 30 days · main

Nothing has been reviewed yet

Turnaround needs at least one pull request that has had its first review inside this window.

status="loading" · a headline placeholder over four stepped lanes, aria-hidden with an sr-only role=status beside it. No timers are involved, and the pulse stops under prefers-reduced-motion.

Review turnaround · Platform

Pull requests marked ready for review in the last 30 days · main

status="error" · the fetch failed. "Try again" appears only when onRetry is passed, and pressing it switches this card to ready.

Review turnaround · Platform

Pull requests marked ready for review in the last 30 days · main

Couldn’t load this chart

The data source didn’t respond.

Right-censoring: Dana Whitlock joined the rota this week, has finished no first review and already has 4 pull requests assigned. Her row has no median to draw — an unfinished wait is not a fast one — so she sorts last, prints an em dash and takes 0% of a share column that still adds to 100. The other three counts are 14 / 11 / 9: rounding those shares independently gives 41 + 32 + 26 = 99, so the leftover point goes to the largest remainder and the column reads 41 + 32 + 27 + 0.

Review turnaround · new rota

Rota changed on Monday · 34 first reviews, 6 still waiting

4.3h

median time to first review

pooled across 34 reviews from 4 reviewers · p90 24.6h · merges in 16.5h · 6 still waiting

Hugo Almeida is slowest — 2.7× the team median

team median 4.3h

Review turnaround · new rota: every reviewer with their review count, their share of the window, the median and p90 time to first review, how that compares with the team median, the median time to merge, and how many pull requests are still waiting for a first look.
Reviewer	Reviews	Share	Median to first review	p90 to first review	Against the team median	Median to merge	Waiting
Hugo Almeida	11	32%	11.8 hours	29.9 hours	2.7 times	20.4 hours	1
Irene Kovač	14	41%	2.8 hours	16 hours	0.7 times	15.7 hours	1
Wen Zhao	9	27%	1.3 hours	6 hours	0.3 times	13.5 hours	0
Dana Whitlock	0	0%	no reviews yet	no reviews yet	not comparable	none merged	4

time to first reviewthen review to mergemedian to p90team median

Medians and the p90 tail use linear interpolation between closest ranks (type 7). The team median pools every first review in the window rather than averaging the per-reviewer medians, so a heavy reviewer weighs more than a light one. Shares are whole percentages apportioned by largest remainder, so the column adds to 100. Pull requests still waiting for a first review are counted beside the name and left out of every percentile — an unfinished wait can only get longer.

Pick a reviewer to lead the card with their median, and pick them again to go back to the team. The accent marks whichever row the card is reading.

Every duration is zero — bots approving their own dependency bumps the moment CI goes green. The headline reads "0m", the team median is 0 so no ratio can be divided by it, the axis has no width to scale against, and nobody can sit above the median, so no row is named as a bottleneck.

Review turnaround · automation

Generated dependency bumps, auto-approved on green

0m

median time to first review

pooled across 35 reviews from 3 reviewers · p90 0m · merges in 0m

Nobody sits above the team median — the wait is spread evenly

team median 0m

Review turnaround · automation: every reviewer with their review count, their share of the window, the median and p90 time to first review, how that compares with the team median, the median time to merge, and how many pull requests are still waiting for a first look.
Reviewer	Reviews	Share	Median to first review	p90 to first review	Against the team median	Median to merge
renovate[bot]	18	52%	0 minutes	0 minutes	not comparable	0 minutes
dependabot[bot]	11	31%	0 minutes	0 minutes	not comparable	0 minutes
release-please[bot]	6	17%	0 minutes	0 minutes	not comparable	0 minutes

time to first reviewthen review to mergemedian to p90team median

Medians and the p90 tail use linear interpolation between closest ranks (type 7). The team median pools every first review in the window rather than averaging the per-reviewer medians, so a heavy reviewer weighs more than a light one. Shares are whole percentages apportioned by largest remainder, so the column adds to 100.

Pick a reviewer to lead the card with their median, and pick them again to go back to the team.

Four reviewers, twenty reviews, every wait exactly 8 hours. Every median equals the team median, so nobody is strictly above it: no row is crowned, no row takes the accent, and the card says the wait is spread evenly rather than picking a slowest at random. p90 equals p50, so no row grows a tail either.

Review turnaround · even rota

Pairing rota · every wait 8 hours to the hour

8h

median time to first review

pooled across 20 reviews from 4 reviewers · p90 8h · merges in 14h

Nobody sits above the team median — the wait is spread evenly

team median 8h

Review turnaround · even rota: every reviewer with their review count, their share of the window, the median and p90 time to first review, how that compares with the team median, the median time to merge, and how many pull requests are still waiting for a first look.
Reviewer	Reviews	Share	Median to first review	p90 to first review	Against the team median	Median to merge
Ana Bergström	5	25%	8 hours	8 hours	1.0 times	14 hours
Bilal Haddad	5	25%	8 hours	8 hours	1.0 times	14 hours
Chen Yifan	5	25%	8 hours	8 hours	1.0 times	14 hours
Dmitri Sokolov	5	25%	8 hours	8 hours	1.0 times	14 hours

time to first reviewthen review to mergemedian to p90team median

Medians and the p90 tail use linear interpolation between closest ranks (type 7). The team median pools every first review in the window rather than averaging the per-reviewer medians, so a heavy reviewer weighs more than a light one. Shares are whole percentages apportioned by largest remainder, so the column adds to 100.

Pick a reviewer to lead the card with their median, and pick them again to go back to the team.

The smallest sample there is: one reviewer, one review. The median and the p90 are the same single observation, so there is no tail to draw; the share reads 100% because there is nothing else to divide with; and one review is under the three-review floor, so nobody is named as a bottleneck.

Review turnaround · one review

First week of a new repository

5.5h

median time to first review

pooled across 1 review from 1 reviewer · p90 5.5h · merges in 19h

Nobody has the 3 reviews needed before a median is worth quoting

team median 5.5h

Review turnaround · one review: every reviewer with their review count, their share of the window, the median and p90 time to first review, how that compares with the team median, the median time to merge, and how many pull requests are still waiting for a first look.
Reviewer	Reviews	Share	Median to first review	p90 to first review	Against the team median	Median to merge	Waiting
Nadia Farouk	1	100%	5.5 hours	5.5 hours	1.0 times	19 hours	0

time to first reviewthen review to mergemedian to p90team median

Medians and the p90 tail use linear interpolation between closest ranks (type 7). The team median pools every first review in the window rather than averaging the per-reviewer medians, so a heavy reviewer weighs more than a light one. Shares are whole percentages apportioned by largest remainder, so the column adds to 100.

Pick a reviewer to lead the card with their median, and pick them again to go back to the team.

Six first reviews across four people. Kit Alvarez is the slowest by a wide margin, but two samples cannot carry a median: minSamples (3 by default) keeps the slowest mark off every row, and each row says out loud that it is under the floor. A dashboard should not start an argument it cannot support.

Review turnaround · thin week

Holiday week · 6 first reviews across 4 people

16.8h

median time to first review

pooled across 6 reviews from 4 reviewers · p90 41.1h · merges in 23h

Nobody has the 3 reviews needed before a median is worth quoting

team median 16.8h

Review turnaround · thin week: every reviewer with their review count, their share of the window, the median and p90 time to first review, how that compares with the team median, the median time to merge, and how many pull requests are still waiting for a first look.
Reviewer	Reviews	Share	Median to first review	p90 to first review	Against the team median	Median to merge
Kit Alvarez	2	33%	41.1 hours	43.8 hours	2.5 times	2.2 days
Obi Nwosu	1	17%	26.3 hours	26.3 hours	1.6 times	33.8 hours
Moe Tanaka	2	33%	5.7 hours	6.9 hours	0.3 times	10.1 hours
Rae Lindqvist	1	17%	3 hours	3 hours	0.2 times	12 hours

time to first reviewthen review to mergemedian to p90team median

Medians and the p90 tail use linear interpolation between closest ranks (type 7). The team median pools every first review in the window rather than averaging the per-reviewer medians, so a heavy reviewer weighs more than a light one. Shares are whole percentages apportioned by largest remainder, so the column adds to 100.

Pick a reviewer to lead the card with their median, and pick them again to go back to the team.

Contract refusal: one row reports merging after 7 hours but being first reviewed after 9. Both clocks start when the pull request is marked ready for review, so that ordering cannot happen — the usual cause is a merge time exported from the first review instead. Better to refuse and name the row than to draw a review segment longer than the pull request's whole life.

Review turnaround · mismatched clocks

Merge times exported from a second query

This feed can’t be plotted

Quinn Baptiste's review 2 merges after 7 h but is first reviewed after 9 h. Both clocks start when the pull request is ready for review, so a merge cannot land before the review that let it in — the two columns are almost certainly measured from different origins.

sort="volume" with showMerge={false}: the same board ordered by who reviews the most, with the merge segment dropped for a denser tile. Priya no longer sits at the top, yet the accent, the slowest mark and the headline still point at her — the order is a lever, the finding is not.

Review turnaround · Platform

Pull requests marked ready for review in the last 30 days · main

9.1h

median time to first review

pooled across 96 reviews from 6 reviewers · p90 34.9h · merges in 19.8h · 4 still waiting

Priya Raman is slowest — 4.5× the team median

team median 9.1h

Review turnaround · Platform: every reviewer with their review count, their share of the window, the median and p90 time to first review, how that compares with the team median, the median time to merge, and how many pull requests are still waiting for a first look.
Reviewer	Reviews	Share	Median to first review	p90 to first review	Against the team median	Median to merge	Waiting
Marcus Bell	22	23%	8.3 hours	17.2 hours	0.9 times	19.7 hours	0
Lena Fischer	19	20%	4.7 hours	12.2 hours	0.5 times	19.4 hours	1
Ade Okonkwo	17	18%	11.4 hours	16.3 hours	1.2 times	25.9 hours	0
Sofia Ruiz	16	17%	5.9 hours	11.9 hours	0.6 times	12.3 hours	0
Priya Raman	12	12%	41 hours	2.7 days	4.5 times	2.3 days	3
Tom Whitfield	10	10%	9.3 hours	16.2 hours	1.0 times	24.9 hours	0

time to first reviewmedian to p90team median

Medians and the p90 tail use linear interpolation between closest ranks (type 7). The team median pools every first review in the window rather than averaging the per-reviewer medians, so a heavy reviewer weighs more than a light one. Shares are whole percentages apportioned by largest remainder, so the column adds to 100. Pull requests still waiting for a first review are counted beside the name and left out of every percentile — an unfinished wait can only get longer.

Pick a reviewer to lead the card with their median, and pick them again to go back to the team. The accent marks whichever row the card is reading.