Why we only search direct flights (and when that's wrong)
Every flight search product faces the same temptation: pad the results. Put a €74 itinerary with a six-hour layover next to the €89 nonstop and a slice of users will click the cheaper number, the results page looks more "complete", and everyone books something. We decided Flight Combinator would only ever search nonstop flights, and since that's an unusual choice, I want to defend it properly — including the cases where it's wrong.
Connections are usually a false economy under four hours
That's the strong opinion, so here's the arithmetic behind it. Take Madrid to Casablanca: about two hours nonstop. The connecting alternatives route you through Lisbon or Paris and turn a two-hour hop into six to nine hours door to door. When we've compared same-day fares, the layover typically saves €10–30 on routes like this. That's selling six hours of your life at €3–5 an hour — and that's the good outcome, where nothing goes wrong.
Factor in the bad outcomes and it gets worse. Every connection is a lottery ticket on a missed transfer, a mishandled bag (checked luggage fails at multiples of the nonstop rate once it has to change planes), an airport meal, sometimes an airport hotel. Short-haul connections concentrate all of the risk of long-haul flying onto trips that didn't need any of it. On sub-4-hour routes, my view is simple: the connection discount is almost never worth what it actually costs.
What refusing connections buys the search
There's an engineering honesty to it too. Because we only consider nonstops, we can afford to sweep wide instead of deep: every airport in your origin city, every airport at the destination, every day in your date range, outbound and return combined freely across different airlines. That last part matters more than people expect — pairing a Ryanair outbound with an easyJet return is frequently the cheapest shape of a trip like London to Marrakesh, and it's a combination single-airline round-trip search can't even represent. (Why round trips deserve suspicion in general is its own article.)
When direct-only is wrong
Three honest cases.
Long haul. Over six or eight hours, the connection discount stops being €20 and starts being €200+, hub schedules are built for it, and one extra hour in Istanbul or Doha against a three-figure saving is a fine trade. Our model isn't built for those trips and I won't pretend otherwise.
Routes with no direct service at all. If no airline flies your pair nonstop, we will honestly show you nothing — and nothing is not a flight. Sometimes the answer is a conventional one-stop ticket. Sometimes it's smarter: Barcelona to Madeira has essentially no direct service (the one direct fare we've caught recently was €730), and the winning move is a self-built split through Porto for under €100 — the Madeira trick, which is really just a connection you control instead of one an airline sells you.
When schedule beats price. If the only nonstop leaves at 6am from the wrong airport and a one-stop leaves at a civilised hour from the close one, reasonable people take the layover. Price is not the only axis; it's just the one we can search hardest.
So that's the deal we offer: within its lane — short-haul, point-to-point, price-driven — direct-only search is sharper precisely because of what it refuses to do. Outside that lane, use a different tool and no hard feelings. Most tools won't tell you where their lane ends; this one just did.