Well, the launch itself didn't fail. But the craft did go slightly off course, which meant it would not be able to rendevouz with the International Space Station. However, the big news is definitely a safe landing.
From my vague recollections of the event, the craft didn't go off course. The Atlas rocket performed flawlessly. There was a software bug in the starliner that resulted in the "mission elapsed time" being 11 hours(!) off, which (1) prevented it from performing its insertion burn and (2) caused the RCS thrusters to expend a large amount of fuel before Boeing was able to stop it. Boeing was able to stabilize its orbit (and in the process damaged the RCS thrusters), and successfully return the spacecraft to the US. It, prima-facie, failed a key contractual milestone of autonomous docking with the ISS. Based on conversations / interviews afterwards it appears that NASA may alter the contractual requirements, but this is speculation (NASA and Boeing are probably not going to broadcast all the dirty laundry on the public news). Personally, it doesn't look good for Boeing QE since they've had two significant QE failures in two tests. A parachute was lost during the pad abort test because of a missing pin, and a significant software bug prevented rendezvous and autonomous docking. I think if the starliner gets rubber-stamped right now, it might be yet another case of NASA "normalization of deviance" that resulted in two loss of craft events for the space shuttle.