The decision is not which fare is cheapest, but which cost you are willing to carry: three extra hours on the ground, a connection at an airport you do not know, or a bag that has to be checked. I book flights the way I book trains, with two options open and the calendar in front of me, and I usually take the second-cheapest one on purpose. Work through these sections in order and the headline price starts to mean something.
Decide what you are optimising
Money, time and the size of the bag pull against each other on every route, and naming which one wins stops the search from going in circles. On a two-week trip I give time priority on the outbound leg and money on the return, because arriving tired costs the first day while a cheap night flight home costs nothing but sleep. Write the priority down before the first search rather than after the fifth.
- Cheapest total: accept a connection, one bag, and the airport forty minutes further out.
- Fastest door to door: pay the difference for a nonstop and a terminal you can cross quickly.
- Lightest logistics: hand baggage only, one airline, no terminal change at 05:00.
When to search, and for how long
Fares move in a pattern rather than at random: domestic routes settle four to six weeks out, long-haul ones tend to sit lowest eight to ten weeks before departure, and the last ten days are rarely kind to anyone. I search once on a Tuesday morning, note two candidates and close the tab, because repeating the same query all day only trains the site to show the fare I already saw. The pattern is a guide, not a promise, and a route with two competing low-cost carriers breaks it completely.
One worked example from my own notebook: Lisbon to Toronto in November, 612 dollars direct on a Wednesday, 548 with one stop on a Tuesday, and the same routing back at 790 by the following week. I took the 548 and lost four hours in a hub I already knew, which is a trade I will make and not one I would repeat with a morning connection.
Where I actually search
Start with the operator itself, because the fare rules, the baggage allowance and the change policy live there, and the aggregators compress them into a footnote. Then compare two independent search engines for identical dates, and check the low-cost carriers the big engines leave out of their results. If a rail alternative exists under six hours, I price that as well, since the train arrives in the centre and needs no two-hour security buffer.
- The airline site: fare rules, seat map and the real change fee in writing.
- Two search engines for the same dates, to catch a fare only one of them indexes.
- The low-cost carrier nobody aggregates: often cheapest on short European hops.
- Rail operators: any leg under six hours where the train lands you in the centre.
What I give up on purpose
The cheapest fare on a page usually hides a 04:40 alarm, an unprotected self-transfer, or a bag rule that costs more at the gate than the saving was worth. A twenty-dollar difference disappears the moment you add a checked bag or one night near the airport. The three things I routinely trade are the middle seat of a night flight, the second airport of a city, and the fare that cannot be changed at all.
- Give up the lowest fare, keep a ticket that can be moved for a fee under 60 dollars.
- Give up ten minutes, choose the nonstop when the connection window is under 90 minutes.
- Give up the aisle, take the window on anything over four hours for sleep and light.
Baggage and seats, priced honestly
Before paying, I add the bag, the seat and the airport transfer to the fare and compare that number with the direct option, because the headline price is a starting point rather than a price. A 10-kilo checked bag on a low-cost carrier can add 45 dollars each way, which is often the exact gap to the full-service flight with one bag included. Seat selection earns its money on flights past six hours and is hard to justify on anything shorter than three.
The transfer is the line almost everyone forgets: a landing at 22:40 with no night bus leaves a taxi fare of 40 dollars or a room near the strip for 90. I price that leg before I pay for the ticket, not after the booking confirmation arrives.
Check the calendar, then confirm
The last step is a paper one: put both candidate fares next to the real dates, count the hours door to door including the airport transfer, and confirm whichever survives that count. I do it with both tabs open and a pen, and the cheaper fare loses about a third of the time, usually on the transfer rather than in the air. Once the booking is in, I set a price-drop reminder for 24 hours, because several airlines still hand back the difference as credit.