As well said by others, if your group uses passive checks rules, the Observant feat grants a +5 bonus
only when the Perception and Investigation skills outcomes are resolved by a passive check, but
not when they are resolved by a regular check i.e. roll of a dice.
Edit: That said, it is confusing, because the feat bonus (+5) only applies to passive checks, which means the PC is often better off not requesting an active one (and he's better off if the DM often uses passive checks to detect stealth and the like, so the feat can be a bit campaign dependent)! Likely the intent was to negate disadvantage on passive checks and/or provide a free level of advantage (though as written it also STACKS with advantage).
That's an interesting interpretation of the RAI, I haven't thought of that before. It could be a good adjustment to the Observant feat for a group that decides not to use passive checks.
And yes, it is definitely campaign dependent. My players are certainly not better off not requesting a check, but that's because I don't typically just let them "request a check" out of character, I prefer them telling me what their characters are doing, and I can rule
very differently depending on whether they do something generic ("I search the whole room") vs something specific ("I look behind the curtains"). I make this example because it is a no brainer: if there's something behind the curtains and you tell me your PC is looking there, you get an automatic success; if you stay generic, I only give you a chance i.e. a check. And I won't let you complain that your character surely would have been smart enough to figure it out: it made its check, and pretty much failed to figure it out.
Correct. In theory the DM should have already performed a passive check before calling for a roll, but I've found some DMs just don't like passive checks and ignore them. I think the primary reason is that the rules aren't explained on how to use them, with the sole exception of perceiving someone hiding. I think the secondary reason is that with expertise and feats like Observant, a character's passive score can get really high.
Well that's the theory of one way to interpret and play these rules, but the books don't really say that every character should be granted passive checks all the time!
Even if you don't completely ignore passive check rules, one sort-of intermediate approach (between never allowing them and granting them all the time) is to allow them only when a character is doing something repeatedly, which is the only narrative explanation of passive checks explicitly provided by the book.
A similar but more restrictive option, is to grant passive checks when a character is doing something repeatedly AND the player has actually declared it: "I keep searching for traps while we're in this area" would grant passive perception checks for that purpose, but not necessarily to simultaneously notice hidden doors or monsters.
Also, Jeremy Crawford adds the following:
"Passive Perception is an option that a DM chooses to use or not.
That's actually the best past of JC advice. Had "Passive Checks" been clearly labelled as optional in the PHB or even better in the DMG (and then the Observant feat could have been moved to that section too) there would have been a lot less sour feelings and arguments about whether and how to use these rules at all.
Of course, what Jeremy Crawford says about passive perception pretty directly contradicts what the PHB says on page 175, that passive checks “represent the average result for a task done repeatedly, such as searching for secret doors over and over again, or can be used when the GM wants to secretly determine whether the characters succeed at something without rolling dice, such as noticing a hidden monster.”
Basically, passive checks are poorly named. They’re for when you do something continuously over a period of time, or when the DM wants to resolve a check in secret. Saying they act as a floor for ability checks doesn’t make sense, since ability checks and passive checks are used in different situations, at least according to the rules as written.
Completely agree.
It has also been pointed out before that if JC's ruling was RAW, then the Rogue's high level "Reliable Talent" would be useless. So there's a class ability and a feat which in a sense assume two very different ways of using a core rule in order to be worthy. That's pretty much a proof that the whole passive checks rules weren't really though carefully during game design, but were thrown into the game more because of legacy habits.
Generally, I only ask for die rolls when players peform some action that they could reasonably fail, and that has real downsides to failure; if someone wants to search a room for an hour, I see no reason to make them roll, and will probably just give it to them- there's only so many places or ways to hide something.
Another good reason to do this with Perception and Investigation is to avoid having the players slowly creep about, rolling dice for every 10' section of dangerous dungeon, which slows down the game immensely.
I do the same thing and often don't ask for die rolls but just grant automatic success, if the players have the right idea.
But the "player rolling dice every 10'" could be also solved by making a single roll for the whole section (or even the whole dungeon, if there's just a trap or two).
The real reason why many DMs hate passive checks, is because they make the outcome non-random, and therefore make it up to the DM to pre-decide what secret doors
will be found, what locks
will be open, and so on.