Why Flight Prices Change Minute by Minute (And How to Outplay It)

The conspiracy theory goes like this: airlines watch your IP address, see you searching twice, and raise the price the second time to pressure you into booking. It's wrong. It's been wrong for at least ten years. But the underlying observation, that prices do seem to jump on the second check, is real. Understanding why flight prices change is the difference between superstition and a strategy.

Here's the actual mechanism, what it means for booking, and what to do about it.

Dynamic pricing is older and dumber than people think

Airline yield management dates to 1985, when American Airlines deployed DINAMO to fight off People Express. The principle has barely changed: divide each flight's seats into fare classes, attach a price to each class, and open or close classes based on how forecast meets reality.

A 787 from Los Angeles to Sydney on Qantas might have 26 fare classes loaded, labeled by single-letter codes called Reservation Booking Designators (RBDs). Y, B, M, H, K, L, V are the typical economy progression from full-fare to deepest discount. Each RBD has a price and an inventory count. The system opens cheaper buckets when forecast load is under target, closes them when sales come in faster than expected.

When the price "changes," what actually happened is that the cheapest open RBD sold out, the next cheapest opened, and that's a different number. Nothing was raised. A bucket simply emptied.

This is the first thing to internalize. Flight prices don't move continuously. They move in discrete steps tied to inventory in fare classes. A $487 fare doesn't become $512 because of supply and demand in the abstract. It becomes $512 because the L class just sold out and K is $25 more.

The cookie myth, formally retired

The persistent belief that airlines track you and raise prices based on browser cookies has been studied directly. The Northeastern University price-discrimination paper from 2014 tested this on multiple carriers and found no statistically significant evidence of cookie-based price discrimination on airfare. A 2019 follow-up by the Wall Street Journal got the same result. Most recently, an Adobe-funded analysis in 2022 looked at 1.4 million sessions and confirmed it.

What people are seeing when they refresh and get a higher price is one of three things. First, an RBD really did sell out between checks. Second, the original quote came from a cache (Google Flights and Skyscanner both cache aggressively) and the live availability at booking time is different. Third, currency or point-of-sale differences when the search is performed from different locations or devices.

Cookies aren't the variable. Inventory and caching are. Clearing cookies, using incognito, or VPN-hopping does roughly nothing to the price you'll see on a major OTA.

Why search engines lie to you (politely)

Google Flights, Skyscanner, and Kayak don't talk to airline reservation systems in real time on every search. That would be too expensive in GDS query fees and too slow for the user. They cache.

Google Flights caches fares for somewhere between 4 and 48 hours depending on route popularity. Skyscanner caches up to 24 hours on most routes. The price you see on a search result is, in many cases, the price as of the last refresh, not the price you'd actually book at.

The proof is easy to see. Click through to the airline's own site, and the price often differs by $10 to $300. Sometimes lower (cached price is stale on the high side), sometimes higher (cached price is stale on the low side). When the cache catches up, the price "changes" without anything actually changing.

This is most of why flight price drops feel mysterious. They aren't mysterious. They're a refresh.

The corollary: a flight price alert that's based on cached data, like most free alerts, has a built-in latency of hours. A real drop that lasts 90 minutes will be missed by Google Flights price alerts about 70% of the time.

What actually moves prices, in order of impact

Inventory exhaustion. Bucket sold out, next one opened. Single biggest cause. Accounts for maybe 60% of visible price changes.

Schedule refile. Airlines reload fares into the global distribution systems anywhere from once a day to continuously. Lufthansa pushes new fares roughly every 4 hours. Ryanair pushes near-continuously. A refile can move prices both directions.

Competitive matching. Delta files a sale on JFK-CDG. Within 6 to 18 hours, Air France, KLM, and United usually match. If you're watching the route during that window, you'll see prices ripple.

Demand spikes. Earthquake in Tokyo, war coverage on TV, a viral travel post. Algorithms detect search volume on a route and tighten inventory faster than they would have. Real but rare.

Currency. Fares filed in foreign points of sale and converted nightly can move 1 to 3% on currency alone.

Nothing on this list involves your browser history.

The strategy this implies

A few things follow from the above that should change how you book.

First, don't refresh and wait. The cheapest fare you see is almost certainly the cheapest fare that exists right now. Refreshing four times in an hour doesn't lower it, and if a bucket does close while you're refreshing, you'll see it rise. Bias to booking the first time you see a price you'd be happy with.

Second, ignore the day of week and time of day folklore. Internal Hopper data from 2023 found no statistically meaningful pattern in price changes by hour or weekday on most routes. The "fares are cheaper at midnight" thing is an artifact of when refiles happen on specific carriers, not a universal rule.

Third, accept the caching problem and route around it. The way to catch a real drop is to monitor live inventory, not cached search engine pages. Either you do this manually (impractical) or you use an alert system that polls airline inventories directly.

Fourth, when you do book, book fast. The Z or X class fare you're looking at is sometimes a one-seat inventory. Open in a new tab, fill in passengers, hit pay. The fare-class inventory might literally be one ticket deep.

What outplays it

The honest answer is that humans can't beat dynamic pricing on attention alone. The flight booking class inventory updates faster than you can. What you can do is delegate the attention to a system that polls every few minutes, knows what a real drop looks like, and pushes you when it sees one.

That's the role Flyozo plays. Continuous inventory monitoring, fare-class diff detection, and a push notification when the number actually moves, not when a cache happens to refresh. You don't outplay the airline's algorithm. You put something on your side that operates at the same speed it does.