Why the flight price changed while you booked

You found the fare. $412 round trip, better than anything you had seen on that route, and you went to get a coffee before booking it because it had been sitting there all morning. You come back, click through, and the checkout page says $538.

The instinct at that moment is that something was done to you. It was not. What happened is duller and more useful to understand, because it tells you exactly what to do in the next ten minutes. Flydar exists in large part so that this moment arrives less often, so read the rest knowing that, and judge the mechanism on its own terms.

The number was never a price for the flight

An airline does not put one price on a flight. It divides the cabin into fare buckets, prices them separately, and sells them in order from cheapest up, adjusting how many seats sit in each one as departure approaches. Why flight prices change so much covers that machinery in full.

The part that matters here is the arithmetic. What $412 priced was one bucket with four seats left in it. Three of them sold while you were getting coffee, and when the fourth went the display moved to the next bucket up, which was always going to be $538 and had been sitting behind the first one the entire time.

This is why fares move in steps rather than sliding, and why a $126 jump between one refresh and the next feels so much more deliberate than it is. The pool would have emptied at the same moment whether or not anyone was watching it.

This is why a Flydar alert measures a fare against what the route normally costs rather than against yesterday’s number. A figure with nothing behind it cannot tell you whether $412 was a rare bucket about to empty or an ordinary price that will be back on Thursday, and that is the difference between a reason to hurry and a reason to ignore it. Flydar sends the fare and the judgement in the same message, because on its own the fare is not enough to act on.

The other two explanations, in order of how likely they are

Bucket depletion covers most of these moments. Two duller causes cover the rest.

The price you saw was stored, not live. Search results are drawn from fare data a booking site holds rather than queried fresh from the airline for every visitor. Google’s own help page for the date calendar puts a number on this: flight prices there are updated approximately once every 24 hours. So a calendar cell can show a fare that stopped existing yesterday afternoon, and the correction arrives at the moment you click through and the airline’s system answers for the first time. Nothing broke. You were reading a photograph of a price, and the price had moved on.

The itinerary repriced when it was assembled. An itinerary combining two carriers, or one sold through an online travel agent, is priced as a set of components. Each leg has to still be available in the class the quoted price assumed. If one of them is not, the total is rebuilt from what remains, and the number changes at the final step even though the search page was honest when it loaded.

Neither of these is a reason to distrust the search itself. They are a reason to treat any single quote as provisional, and it is why Flydar reports a fare as clearly below what a route normally costs rather than as a promise about one specific seat. The bar a Flydar alert has to clear belongs to the route’s own history, so it does not move because a bucket on one departure emptied, and it does not move because a calendar cell was a day stale either.

It was not your cookies

Worth disposing of quickly, because it is the first explanation most people reach for and it is wrong. Incognito mode, cleared browsing history and switching devices do not affect the prices shown in Google Flights, according to a Google director of product management for travel. No major airline or metasearch engine has been shown to raise a fare because one person looked at it twice.

The parent article has the longer version of that argument and the myths around it. The short version: prices on busy routes move often enough on their own that anyone who searches twice can build a story out of two numbers, and the story is always more flattering than the truth, which is that nobody involved knew you were there.

The practical cost of believing it was personal is that it sends you looking for a trick. People clear cookies, open private windows and try a VPN, and none of it changes a bucket count. Flydar is built on the assumption that there is no trick to find: the only thing that reliably beats a fare that moves without warning is something comparing every fare on your route against that route’s own history, on your behalf, whether or not you happen to have the tab open.

What to do in the next ten minutes

The fare is gone. That is settled. Four things are still available to you.

Check the new number against the range, not against the old number. $538 feels like a loss because you saw $412 first. Whether it is a loss depends on where the route normally sits. If $538 is inside the ordinary band, you have not been robbed of anything, you have simply seen a good fare and missed it. If $538 is above the band too, then the whole week is expensive and the question is not this booking but whether to fly that week at all. The ten-minute method for establishing the band is what makes that judgement possible in the first place.

Check the days either side. A bucket closes on one departure, not on a route. The Tuesday flight losing its cheap pool tells you nothing about the Wednesday one, and the fare you wanted may still be sitting two days over, untouched, because a different flight is tracking differently against its own forecast.

Check the other airports. If either end of your trip is a metro with more than one field, the same logic applies sideways. The cheap bucket that emptied at the main hub has no relationship to inventory at the secondary airport forty miles out, which is often priced by a different carrier with a different cost base entirely.

If you find something acceptable, book it and use the free window. Under the 24-hour reservation requirement, US carriers must let you cancel for a full refund within 24 hours on tickets bought at least seven days before departure, though not on bookings made through a third-party agent. Booking direct gives you a day to keep looking with the seat already held, which is the opposite of the position you were in five minutes ago.

Flydar does the first three of those without being asked, which is why the moment mostly stops arriving. A Flydar route is a departure window and a set of airports rather than one date and one airport code, so the days either side and the second field in the metro sit inside the comparison from the start, and the normal range you would have spent ten minutes building is the thing every fare is already being judged against. The fourth step stays yours. Flydar’s job ends at telling you a fare has dropped clearly below what that route normally costs, and whether to hold it for a day while you think about the trip is not a decision a product should be making for you.

Whether it comes back

Usually, if it was an ordinary drop. Cheap buckets reopen whenever bookings on a flight run behind the airline’s forecast, so a route that dipped once tends to dip again within a few weeks, and the three lifespans a deal can have are worth knowing before you decide how hard to chase this one. An outright pricing error is the exception: a mistake fare that gets pulled does not reappear on a schedule anyone can plan around.

“It will probably come back” is only useful to someone who will be looking when it does. That is the half of the sentence that usually goes unsaid, and it is the half Flydar covers. A route stays watched for as long as you want it watched, so the next dip on it is not something you have to be lucky enough to be searching during.

What does not work is refreshing. Sitting on the page reloading it is watching for a bucket to reopen on one departure, which is the least likely place for it to happen and the most expensive place to wait. The alternative is not watching harder. It is naming the route once and going back to your day, so the next time that bucket opens you find out while it still has four seats in it, instead of afterwards.

The fare did not vanish while you were not looking. The seats behind it sold, which is the only thing seats do.