The goal and approach method has, as an additional method, a failure state that is different from the state prior to the roll. As a broad approach, these failure states vary. [MENTION=97077]iserith[/MENTION] has said he might tick a wandering monster clock for a failure in some games, letting the character know they failed and can try again but still changing the situation to become more dangerous. I prefer more immediate changes, such that a failure thwarts the goal directly (so failing an insight check may end the social encounter or cause a damaging social gaffe). Regardless of preference for failure states, the approach will inform the failure state.
Again, I'll provide an in play example from a recent session:
The party knew they were entering an old temple complex full of traps. In the first hallway, there was a trap, a set of false doors that would snap shut to seal the hallway if a pressure plate was triggered. The party had their gloomstalker ranger scout ahead. Mindful of alerting possible enemies, he chose to advance down the hallway without a light source, relying on darkvision. This meant than the DC 12 passive perception check to notice deep groves on the floor showing the arc of closing for the false doors was missed (Passive of 16, -5 for dim light). Had they not, this clue would have indicated something odd and given insight into the nature of the trap. However, the player's chosen approach (use darkvision) to the goal of scouting for dangers, lead to an automatic failure.
The player then chose to move past the doors without investigating them, so, again, the chosen approach lead to automatically missing the trap. The pressure plate was triggered and the trap cut the ranger off from the rest of the party.
The party rogue then moved up to the trap with a light spell to examine it. As the trap was sprung and the mechanism obvious, he automatically succeeded in his action to determine the nature of the trap. (He had good light and stated he was inspecting it visually.) I told him what the trap did and that it was likely triggered on the other side where he couldn't see. He also could see that the mechanism was accessible but under a lot of pressure, so failing to disarm it may cause a violent release, likely spraying pieces like shrapnel as it disintegrated.
Meanwhile, the dwarven battlerager decided this was taking too long and charged the stone doors blocking the hall to break them down. A STR check was called for with failure causing danage for running into a stone door and success breaking the door but automatically causing the same violent release. The dwarf's player agreed and rolled -- success! The door was shattered and a DC 13 DEX save was called against 5d6 piercing damage, half on a save. Barbarians at his level have advantage on DEX saves from sources they can see and it was easily passed. The dwarf's approach given the established wirking of the trap directly resulted in the violent release of the mechanism. Had the rogue attempted to disarm, a success would have both disengaged the doors and avoided the explosion. A failure would have disabled the doors, but with an explosion.
The party has continued on, but now the ranger is using a light source which has already resulted in detection of a similar trap deeper in.
And, finally, my preferred method of requiring a goal and approach with consequences for failurr DOES NOT MEAN that your method cannot. Of course ypu can do this. But, instead of having to stop play to clarify things in situations where "I roll X" is insufficient, I've taken to preferring to bever assume and always have players describe approach. It's a preference that solves things at my table, and works well enough I like to advocate for it. Neither my method nor my play suffer at all if you do it different.
Almost any approach works, by the way, I most certainty don't require anything close to magic words. The trap I describe above had the following notes: [doors close to seal hall on sctivation of pressure plate, DC 12 passive to notice clues, once triggered doors under pressure]. That was it, no magic solve, any reasonable approach by players would have moved play further.