I'm considering the first IFT "Not Successful", in that it didn't insert Starship into a near-orbital trajectory. Even though personally I think it was a successful data gathering flight. The guiding principle for success is something like "Did this flight prove to customers their payload will go where they want it to go?".

Essentially, will SpaceX make 500 launches that a customer would consider proves satisfactory reliability and safety of the system by Jan 1st 2030.

Clarified criteria:

A "successful launch" means any launch in which:

• A Starship–Superheavy full stack lifts off from the surface of Earth under the thrust of its engines, and

• Starship reaches an altitude of at least 100km, and

• Starship either:

  • Successfully deploys or delivers a payload to its intended trajectory or destination, or

  • If a test flight with no payload, reaches its target trajectory or destination - disregarding later parts of the flight pertaining only to vehicle recovery or disposal.

The last point is context-dependent:

• If a test flight is attempting to land Starship on e.g the moon or on Mars, then Starship must land in one piece at the target destination to count as a success.

• If a test flight is testing Earth orbital capabilities, then Starship must reach its target orbital or near-orbital trajectory, but the launch will still count as a success if it burns up on re-entry.

• If a test flight is of point-to-point suborbital hops, then Starship must land in one piece at the target destination for it to count as a success. Point-to-point hops can be identified by their highly elliptical trajectories, with target perigee well inside the Earth.

In all cases it doesn't matter whether the booster survives.