What it looks like
The example plan: six cities, three orders kept
from the example plan- Order chosen
- Dallas → London → Athens → Budapest → Rome → Paris → Madrid → Dallas
- Flights, per person
- $1,210
- Runner-up order
- $1,238, $28 more
- Third order kept
- $1,237, a dollar under the runner-up and ranked below it
- Nights it picked
- 3 in Rome, 2 in each of the other five, inside a 2 to 3 range
The three are ranked on what they cost door to door rather than on the fare, which is why the cheaper of the two runners-up is the one printed last: the hours it adds are worth more than the dollar it saves. The gap at the top is what you are being told, and it grows with the number of cities and the slack in your window.
When it does not work
- Two cities and fixed dates leaves almost nothing to optimise. The win grows with the number of cities and the slack in your window.
- Locking the order, or pinning a start and end city, means you are asking for a specific trip rather than the cheapest one. That is legitimate; the plan just costs more.
- Nights are allocated from departure dates at day granularity, so the night spent on an overnight transatlantic leg is charged to the city you have not landed in yet. The optimizer solves in those terms, because that is what your own minimums are expressed in. But the calendar file, the printed sheet and the itinerary card all count from the flights instead, and say so where the two differ.
- More than seven cities is not supported: the number of orders grows faster than any budget for pricing them.
Risk
No risk
One booking, normal rules. Nothing here can be taken away from you.