As others pointed out : Fail forward = success but is only required when the goal of the endeavor is the singular and obligatory path forward.
To address this further: as I understand things, Jonathan Tween in 13th Age is correct about the origins of "fail forward" as a self-consciously identified technique (from the 13th Age rulebook, p 42):
A simple but powerful improvement you can make to your game is to redefine failure as “things go wrong” instead of “the PC isn’t good enough.” Ron Edwards, Luke Crane, and other indie RPG designers have championed this idea, and they’re exactly right. You can call it “fail forward” or “no whiffing.”
Tweet has something similar in his commentaries in the 20th Anniversary edition of Over the Edge.
Fail forward or "no whiffing" as Luke Crane and Ron Edwards advocate it is
not, in general, "success but . . .". Because in the games that use it,
success means not just
success at task but
success at intent. And
fail forward is all about failing at intent, while nevertheless perhaps succeeding at task (hence, no whiffing).
In a travel challenge the PCs can succeed at task and yet get lost if the fiction is appropriate: eg if the intent and task is "We use our compass to strike out boldly north" then a failure - if the fiction permitted - could be narrated as some external influence affecting the compass, so that the PCs succeed in following their compass and yet end up not heading north. That is not "success but". It's failure - but
forward, because now the challenge is for the PCs to identify what it is that is influencing their compass.
In the indie games that Tweet refers to there is no such thing as "the singular and obligatory path forward". Those games are about avoiding railroading, not operationalising it via "success but" narrations of failure.
Equally, in those games, if
nothing that matters to anyone in the game turns on the success or failure of the PCs getting from X to Y, then there is no need to call for any sort of check. Just narrate it and move on. Likewise, if
everything that matters to the game's participants turns on the PCs getting from X to Y, then there is no need to call for any sort of check. Just narrate it and move on.
Once there are clear stakes -
we need to get there to rescue our loved ones, say - then that is where "fail forward" is an invitation to the GM to twist the knife: eg the check fails, the GM narrates some appropriate delay-causing consequence, and then when the PCs arrive at Y from X they find their loved ones in even worse strife than they thought. Note that, in this example, there is a significant extent to which the fact that the PCs are now at Y rather than X is mere colour: the real action is the state of their loved ones, and the narrative logic of the check is not "will we make it from X to Y?" but "will we rescue our loved ones?"
It is perfectly OK to say that failure to arrive at all is in the cards. It simply must be true that whatever scene frame is thus entered serves to advance the story and doesn't thwart it or turn it in a DM-determined direction, at least to too large a degree.
If we are talking about the sorts of systems that are the origin of self-conscious application of "fail forward", then whether or not failure to arrive at all is on the cards depends entirely on the details of the fictional situation and how it relates to what anyone at the table cares about.
In my Prince Valiant game, most of the time the PCs' travel across Britain is simply narrated as occurring. There is no need for any checks, because everything that any participant cares about is premised on the PCs getting from X to Y. The travel is just a backdrop to the events that actually matter in play.
Conversely,
if you are going to call for checks - as [MENTION=1282]darkbard[/MENTION] is intending to - then you should know
why you are doing that. What is at stake? If you don't know that, then you haven't framed your check properly. Once you do know what is at stake, it may or may not turn out to be the case that non-arrival is among those stakes. There's no way to ascertain that possibility in the abstract - it's all about the details of the fiction.
(Of course in some RPG systems, travel always requires a check - which is to say that the system itself always puts some stakes forward as part of travel. Interstellar travel in Classic Traveller is an example of this. But 4e doesn't fall under that description - there is no rule of 4e that
demands a check because the players declare that their PCs travel from X to Y.)