Statistics / Closings
One in thirteen was dead before broadcast
22 rescued bars — about 1 in 13 of the 284 — had already closed by the time their episode aired. This page is about endings: when they happen, the few causes on record, and what a sale does and doesn't mean.
Dead before broadcast
Episodes are filmed months before they air, and some bars don't outlive the gap. The five longest pre-air spans:
Closed October 2012 · episode aired June 1, 2014
Closed March 2018 · episode aired March 31, 2019
Closed September 2016 · episode aired July 2, 2017
Closed September 2024 · episode aired May 4, 2025
Closed October 2018 · episode aired May 12, 2019
20 of the 22 are month-certain; 2 are same-month calls (the closure month is the air month, so the order can't be proven from month-level dates). The pattern is roughly proportional across eras — a constant of TV production lag, not a late-show decay.
Closures by calendar year
Dated closures per year. The peak years are 2019 (17) and 2016 (16).
2020 is the quietest full year in the record — 3 closures, 2.8 per 100 exposed bars. If the pandemic killed rescued bars, the record doesn't show it: COVID-19 is the documented cause for just 4 of the 153 closures. 2025–26 counts are shown greyed: closures surface in the record with a lag, so both years are certainly incomplete.
Why bars close — the documented minority
For 43 of the 153 bars that closed, the reporting supports a specific cause. The other 110 closures have no documented cause and are left unclassified rather than guessed — so read this as a case series, not a sample: causes get documented when there is a story.
No per-cause lifespan ranking is published: with this few dated cases per cause, the intervals span years and nothing separates them.
The sale, decomposed
Being sold isn't a death sentence. Bars sold as going concerns to new owners are still open at the same rate as bars that were never sold (53% vs 54%) — the "sold" bars that drag the average down are the ones where the sale was the closure, or only the building was sold. Split properly, the 284 bars divide as:
Closed by construction — the event type asserts the closure, so this is bookkeeping, not a finding.
Post-closure real-estate sales: the business had already closed; only the building changed hands.
Put up for sale; an anecdote at this count, not a group.
Carry the sold flag from an older, untyped record; the sale type was never re-audited, so these are excluded from the headlines above.
How long the closed bars lasted
Distribution of final lifespans for the 120 bars that closed after their episode aired and have a recorded date — the population behind the 1.86-year median:
The longest closed runs:
Closures show no seasonal pattern (chi-square p = 0.86 over the 122 month-precise closures falling outside their air month). Methods: lifespans run from air date to closure date; the 11 closures without a recorded date are counted in totals but in no timing figure. Back to the overview →