Although Blizzard's Diablo basically invented the Action RPG subgenre and influenced many other RPGs, it wasn't just innovation that made it a great game. Though outdated in many areas, there are other aspects of the game that are rarely if ever done as well in the games that came after (including its sequel). Having gone back and played it recently, these are the things that stood out to me.

You fight an enemy type. Not an enemy level:
In my opinion, one of the worst trends in RPGs is representing power by a little number next to the enemy's health bar. A skeleton that looks and acts just like one from earlier in the game will literally be 100 times as powerful. And it will be the same power as the giant you just fought that happened to be the same level. Sure, Diablo did some reskinning, but it carefully limited how powerful each model could get.

The automatic argument most people make against this is that you need too much art to keep changing enemy types, but this isn't true. In fact, it's one place where the limitation actually leads to better game design.

Instead of a level 20 character fighting two level 20 goblins and a level 40 character fighting two level 40 ogres, maybe the level 40 character will fight an ogre surrounded by a horde of goblins. Or maybe the ogre will be a mage that summons goblins. Since these goblins are the exact same power as those the player has fought previously, their character's progress is highly visible ("I remember when one of these guys used to be hard"), yet they are still challenged. Instead of changing the stats on each enemy, change the context under which they are fought. This will actually add more variety than making a bunch of new models that are functionally the same as the old ones.

Anything had a realistic chance of an awesome loot drop:
The key word in this sentence is "realistic". A .00001% chance doesn't count. Further, don't confine specific loot to specific enemies or areas (other than loosely connecting power of loot with power of enemies). Setting which items drop from which enemies encourages grindy behavior as players seek specific equipment, while loosening this up encourages players to just play the game and enjoy the occasional great drop. If players know that they have a reasonable chance at getting a great drop from any playing session, it greatly increases the excitement they feel when going through the regular game tasks.

If you want to make some items that the player can choose to obtain (rather than relying on randomness), make those items reliable drops (ex: dragon boss always drops sword X) or purchasable so that they player doesn't feel like effort to gain that item is simply wasted until they get a lucky session in.

Fewer item slots made individual pieces of equipment more meaningful:
When you can equip 20 different items that all give stats, a loot drop rarely makes a significant visible difference in your performance. Diablo, if I remember correctly, had 7, and I tended to notice a significant difference most of the time when one of them got replaced.

An alternative if you like a lot of slots is to have a few with very specialized roles that do not include giving stats. Low level trinkets in WoW function like this (i.e. Gnomish battle chicken and Linken's Boomerang do not give stats), but at higher levels, they just turn into stat sticks like everything else you equip or have useful but very passive effects.

Shops had equipment that was both random and good:
Getting gold in Diablo was more rewarding than in any RPG I've played since because the shops, on occasion, would carry really awesome items. Likewise, good drops that you couldn't use sold for a lot, so finding those was still exciting. Naturally, this becomes a more complex issue in MMOs because of gold buying, but if it's not an MMO, make money fun.

Minimal clutter in item drops:
In many modern games, you get so many drops that are potentially good that to feel like you aren't missing stuff, you have to spend a lot of time mathematically analyzing them. I'm not talking about interesting choices (i.e. Do I want the bow that stuns the enemy or the one that hits targets behind it). This is more focused on analyzing which stat stick will increase your effectiveness in doing the exact same thing you've been doing.

In Diablo, drops that were worth investigating or even carrying around to sell were quickly identified, often just by the name, and were rare enough to avoid excessive clutter (while being common enough that you didn't feel too deprived). I've found that action RPGs in particular tend to be overcluttered most of the time.

Permanent cool bonuses could drop or be purchased:
In Diablo, this only applied to characters who used magic (generally the mage, though the other two classes would often carry around a few spells). Finding duplicate spellbooks increased the power of your magic permanently and significantly. The advantage to this is that it provided drops that were ALWAYS exciting with no fear of simply being replaced when something better came along. Naturally, this won't work in an MMO environment, but in any other, this adds another layer of excitement to looting.

Other characters were able to get permanent bonuses in the form of elixirs, but these were a lot less interesting.

Conclusion:
In many regards, Diablo is an outdated game, but there are enough well-designed aspects of this that every RPG designer should give it a second look.