How we measure this

Every SAS-operated nonstop long-haul route from Copenhagen, Stockholm and Oslo, in both directions, on every departure date out to roughly a year ahead, in Business, Premium Economy and Economy. That is 17 destinations.

Nonstop flights only. A connecting itinerary re-sells the same physical seats as the nonstop legs it is built from, so counting both would double-count the supply. One-ways in both directions are kept separate, which is why we can say anything about outbound versus return.

SAS metal only. SkyTeam partner award availability is a different market with different pricing and different behaviour. None of it is in these numbers.

Continuously, with each flight and date re-checked roughly once a day and busier routes more often. This is the single most important limitation on the page: a seat that appears and sells out within a few hours can pass unseen entirely. Every figure we publish for how long seats stay bookable is therefore an upper bound. The real world is, if anything, slightly faster.

A stretch of continuous availability on one flight, one date, one cabin. Two adjustments make that honest, and both changed the answers materially.

Inventory blinks. Availability frequently vanishes and returns within hours. Treating each blink as a fresh release would inflate the count and make seats look shorter-lived than they are, so a re-appearance within 24 hours on the same seat continues the same stretch instead of starting a new one.

Finding a seat is not the same as watching it open. When we begin following a route, everything already on sale becomes visible at once — but those seats may have been sitting there for months. Counting them as releases once produced a convincing pattern showing SAS released seats on Mondays. It was an artefact of our own first look, and it disappeared the moment we excluded it. So a release counts only if we were already watching that exact flight and date before the seats appeared.

Both bases are published, never blended. History (“has this destination ever had space”) includes everything we have seen. Timing and lifetime figures use only the openings we witnessed.

Many seats were still bookable when a measurement window closed, so their true lifetime is unknown. Simply ignoring them would bias the answer downwards, because the longest-lived seats are exactly the ones most likely to still be open. We use a survival estimate that accounts for them instead. It matters: the naive approach understated how long Business seats last by roughly 40%.

The window is short. Christmas inventory, the summer peak and the far end of the booking calendar are not yet properly measurable. Release timing beyond about six months out is under-counted rather than absent — proving such a release requires having watched that flight and date beforehand, and the observation window bounds how far back we can reach.

We see availability, not bookings. When seats vanish we cannot tell whether someone booked them or SAS withdrew them.

We cannot explain SAS’s decisions. We report what the booking engine showed and when. Why a seat was released when it was is outside what this data can answer, and we do not speculate.

Coverage has gaps. Rate limits and occasional outages mean not every flight is checked on every cycle. Where a number rests on few observations we say so on the page rather than presenting a thin figure as a firm one.

If you think a number here is wrong, please tell us — we would rather fix it than defend it. Write to philip@roamsnap.com and, where possible, include the route, date and cabin so we can check it against the log.

RoamSnap is independent and has no relationship with SAS. Award availability changes without warning; always confirm on sas.se before planning around anything you read here.