How long does an ANA award seat last? We re-measured, by source.

The corrected finding: genuine ANA business seats

Genuine ANA business seats, confirmable (not waitlisted) with a Star Alliance partner column present, closed with a median lifetime of about 261 hours (11 days) in our sample. That sample is small, 7 closed episodes, so read it as directional, not precise.

Looking at the full 136 days of scanner data rather than only closed episodes, 73% of these seats were still open at a re-look within 24 hours, 50% at 1 to 4 days, and 34% after 4 days. The longest confirmed life was 31 days. Onset timing is poorly resolved: most seats were first seen more than 7 days after our previous look at that flight, so we cannot say how long they sat before we noticed them. As with the original study, all durations are lower bounds at scan resolution.

The 46 episodes without a partner column present (a state we cannot confirm is bookable) show a much shorter median, 22.5 hours, on a sample of 4 closed episodes. We are not confident that distinction is real rather than noise at this sample size, and we are not building product claims on it yet.

United's portal: an upper bound, not a finding

The United-portal rows above (ANA-operated seats bookable with United MileagePlus miles, a different redemption from the Virgin Atlantic one Layflat watches) closed with a median of 3.9 hours and an 82% single-sighting rate. Across the full 136-day dataset, 273 flights disappeared from United's display and reappeared at least once, and 44 of those flapped four or more times.

We do not trust those numbers as a measurement of United's portal, and you shouldn't either. We found that our United worker was writing a "seat gone" absence for every calendar day of the month with no sighting in that run, including days United's rolling 30-day calendar never rendered and days the worker deliberately skipped as already verified. Every one of those became a recorded disappearance. That inflates every instability figure above, single-sighting rate, median and p90 lifetime, reappearance count, flap count, by an unknown amount. We confirmed the defect against saved page HTML from seven runs; it was not a block or a captcha. The write path is fixed as of 2026-09-02 (an absence now requires a rendered price tile as evidence), but every historical United figure predates the fix and is contaminated to a degree we don't yet know.

Read the numbers above as upper bounds on United's instability, not an established property of its portal. We will re-measure once enough post-fix data accumulates and update this page then. What doesn't change: this churn, real or not, was never a seat vanishing, it's a different redemption from the one Layflat watches, and it has nothing to do with the genuine ANA business finding above. ANA's absence path does not share this defect, and the paired-observation disagreement between United and ANA compares sightings, not absences, so neither is affected.

Waitlisted inventory

Waitlisted seats behave differently from bookable seats, as the original report said, and the re-run confirms the original figures were roughly right here: median time-to-resolution of 284.7 hours, close to the original 96-hour figure once measured against the full episode set rather than only closed waitlist episodes. Waitlisting is mostly a dead end: only 1.7% of waitlisted flights ever became confirmable, a median of 64 days later.

The by-city-pair table is gone

The original report broke closed episodes down by city pair. We removed that table rather than correct it: it was built on the same pooled, source-blind episode set as the 79% headline, so every row mixed United-portal display churn with genuine ANA business seats in city-pair-specific proportions we have not re-derived. Producing an honest per-city-pair table means re-running the source split above by city pair, and at 7 genuine closed episodes total, most individual city pairs would not have enough data to say anything precise yet. We will publish a corrected version once we have more genuine ANA business episodes to split.

Methodology

Layflat's fleet scans ANA business class award availability across 9 North American hubs to Tokyo every few minutes, around the clock. Every scan records either a sighting (a seat is bookable on a given flight/date) or an absence (a previously-seen seat is no longer there). We reconstruct episodes, continuous stretches from first sighting to confirmed absence, from that sighting/absence stream.

What this means for you

We no longer claim a 4-in-5 chance of a seat vanishing before a second scan. That claim was measuring United's display, not ANA award space. The honest picture is closer to the opposite: genuine ANA business seats often last days once they open, and most are still there a day later. United's own display of ANA-operated seats looked unstable on an hours timescale in our data, but a scraper defect inflated that number by an unknown amount (see above), so we are not calling it a display quirk or anything else until we re-measure. Either way, it was never the seat.

None of that makes watching less valuable, it changes why. ANA releases exactly one saver business seat per flight. Whoever calls Virgin Atlantic first books it, and it is gone for everyone else, regardless of whether the seat itself would have lasted hours or days. The value of a watch is not racing a fast-vanishing seat, it is a corroborated, fresh alert you can trust: when Layflat tells you a seat is there, that claim is backed by this same measured data, not a guess, and not a stale scrape from a source we now know churns for reasons that have nothing to do with the seat.