Page 133 of the 5e DMG suggests that a typical campaign* awards seven 0-4 hoards, eighteen 5-10 hoards, twelve 11-16 hoards, and eight 17+ hoards. I made a big spreadsheet and figured out what the average results of those tables look like, all added together:

CR Individual Award (per creature) Total GP Value of Cash, Gems, and Art Magic Item Table Rolls
0-4 5 gp 2,630 A x 6, B x 3, C x 2, F x 2
5-10 93 gp 81,797 A x 10, B x 9, C x 5, D x 1, F x 6, G x 2
11-16 947 gp 434,550 A x 4, B x 6, C x 9, D x 5, E x 1, F x 1, G x 2, H x 3, I x 1
17+ 8,470 gp 2,688,200 C x 4, D x 9, E x 6, G x 1, H x 2, I x 4

 

For reference, after all the hoards are awarded (and not counting the pocket change of individual awards) I got a total of 3,207,177.30 GP plus 104 items. The vast majority of the money comes from 17+ hoards, and there are only 46 items found in the first two sets of hoards (and most of it is from tables that primarily award potions and scrolls).

While I’m sure a lot of GMs enjoy rolling up loot at the table, I’m more methodical and also know that I will totally forget to give out sufficient** treasure if I don’t have a plan up front to award it.

For my games, I’m basically chopping this up into 8-13 packages per tier, making sure to give out at least one item in each package, and randomizing the distribution of the GP value a bit. Then, when I decide what each package makes sense for I split up the value into art, gems, and cash (e.g., a goblin hoard might be a ton of copper and silver, an elemental will be all gems, and humanoids with neat nonmagical gear will have that taken out of the budget as “art”). I’m also pre-rolling the items, so I also try to assign the containing package to an encounter that would make sense to have that particular item.

You could obviously also totally divorce the items from the value packages, and sometimes give out multiple items with little or no cash, and sometimes just nonmagical items of value.

 

* I assume this is for four PCs. Presumably you should raise the total by 25% for five, 50% for six, etc.

** Not that 5e really seems to care if you get anywhere close to the normal distribution.