Slight alterations to existing npcs to make them look different, not the kunark method of "this froglok's name is different, He's got 11x the hp!" type crap. Having the concept of tiers already built into your design means your art team already knows, so your starting stuff looks standard fantasy fare, and gets more intricate/menacing/cool looking as you progress. Then, you use a different metric of balance for each tier of mobs. If you have ever played/heard about FF12, it has a gambit system that allows you to effectively program individual characters with predetermined actions/abilities/etc. Map that onto npcs, create a few "archetypes" such as caster, healer, ranged, rogue-like melee, whatever. Then give them a couple of slots that effectively have random abilities from a tier/type specific list. When they spawn, they spawn with a generated skillset, at a predetermined level range. Think quasi-diablo2 elite naming conventions, but with mmorpg type abilities. Could even tie specific naming deals to specific abilities. And yes, the idea is you -want- players to go balls to the wall. Because the more shit that dies, the more content is released for everyone, so that everyone has more stuff to do.
If anything, the more powerful guilds will want the lower tier guilds to kill as much as shit as possible, as it pushes the content to become progressively harder, especially if the more powerful guilds are killing appropriate mobs as well. Then have gear follow the trend. As the bigger guilds kill bigger shit, more things become available for lower tier guilds to do (not the leavings that occured in EQ, but actually opening -more- content instead of blocking it) and more powerful gear becomes available across the entire server. For example, Terraria (omgz sandbox!) has "Hard Mode" which is opened by killing the Wall of Flesh in Hell. Until the wall is killed, you only have normal types of ore and monsters around. When you kill the Wall, three new types of ore/material open up, lots of new crafting options and new mobs are released. Expand that into an mmo.
Server starts fresh, you have gnolls/kobolds/orcs/aviaks/darkweed snakes all over the place. As people kill the populations and spawn bosses, they progressively keep pushing the power levels of the spawned bosses up. Eventually you end up with some dragon spawning and being substantially harder than previous bosses. Big guild kills it and suddenly now the triggers to spawn the previous tier of bosses have their requirements reduced. But now there's Obsidian (resource originally not available in the game) for making new weapons and shit and drakes and unicorns and all sorts of other higher tier mobs now spawn in places. Bosses keep getting killed and the increased spawnrate of lower tier bosses means more shit for lower tier guilds to kill, ramping up the counts needed to spawn the dragon tier bosses much faster so that the upper tier guilds have more content to kill as well. This in turn lets them spawn the next tier of bosses, which when killed in turn creates more lower tier and dragon tier bosses while the next tier will spawn faster because there's more shit for guilds to kill. This also creates new levels of craftable and dropped gear each time a tier is popped, creating a jump in power for the entire server each time something huge is killed.
This jump isn't a gear reset, it is the natural progression of getting better gear by killing bigger shit. And while each progressive boss means harder shit spawns, it also increases the spawn rate of lower tier bosses (to some predetermined cap so killing an orc doesn't spawn 3 dragons or something stupid like that, obviously) so that guilds that aren't cutting edge have even more opportunities to see higher level content faster and can gear up quicker.
Tie in skill growth with this "server power level" along with crafting components and types and the availability of new spells/abilities as well as new content being opened up. Don't penalize the server because one guild is the most powerful in the land. Have people rooting for the top end guilds to kill shit so that there is more stuff to do, not less like what happened in EQ up through PoP.
From a content generation standpoint, once you make a couple of tiers for each type of mob, and give the raid bosses that can spawn (even have bosses that spawn randomly in certain areas, and are not tied into the forced spawn counters) some abilities, at that point you are ahead of the game. You put the "gating" counters behind what people have actually killed, not in the time it takes to kill them. Then you simply can add bigger/more powerful content to the game as you go along, and by using a modular approach to the artwork/content, you can easily keep new shit happening for quite awhile, especially without needing all the scripting and shit that happens in FFXIV and similar.
Another post I can't seem to find detailed the tier system I referred to. It was essentially the idea that mobs of various types are populations. Dungeons with a specific flavor have specific mob types, but in the overworld, mobs are based on population. If, for example, you are in The Commonlands, Darkweed snakes are all over that mofaka. If you keep killing the orcs, eventually they stop spawning all together, and the zone is just full of darkweed snakes. Once the population hits a certain point, you get a rare darkweed snake to spawn somewhere in the zone. If you kill this snake, the darkweeds start despawning if left up for too long and then the orcs start spawning again. You can create synergies like this all over the place, and it all comes down to the spawning mechanic being tied to some very simple database work. Anyway, killing these tier-1 "boss" mobs in zones slowly builds up the counter to an overworld "king" mob who then can spawn. This could be zone dependent, or continent dependant, whatever. This king mob is substantially stronger and requires at least a full group compared to the rare, if not more. Killing the king causes the counters needed to spawn the boss mobs to shrink, meaning they can effectively spawn faster if people are still actively killing stuff. That king mob also adds a counter to a continent/game wide (whatever, easy to expand/shrink) that leads to a dragon spawning. So by playing the game normally, bosses spawn at a fairly randomish/predictably slow rate. Especially in zones with lots of mob types, it would be hard to spawn a specific boss, etc. Anyway...
So the entire server contributes to raid content opening up. It doesn't start that way, you basically have to "level up" the server. The more people actively taking part in content that affects these counters (quests/crafting/whatever can figure into them as well. Pretty easy to get creative with the methods) the faster the raid content opens. But if that many people are actively spawning shit, that means that all the gear/loot/levels aren't being concentrated into one guild, so they aren't going to be decked in gear ready for this dragon raid right out the gate. As people keep killing mobs, eventually the big guild/s take on and successfully kill the dragon. As stated earlier, new crafting shit opens up, the kings spawn more often, new monsters spawn in areas, etc. If the big guild manages to keep downing the kings, then the bosses continue to spawn faster up to a point so that more kings can spawn, putting more group+ mobs into the hands of the rest of the server. Everyone wants guilds bigger than them to kill stuff, because it means more stuff for them to kill and faster gear/level gain period. This dragon now requires slightly less kings to respawn, and in and of itself it added to a number of different counters. Now you're in tier 3, which as people kill the dragon and the previous other bosses, it continues to add up until a demi-god or whatever spawns, and this fashion keeps going on and on. This is just overland.
Then, you tie the dungeons that are thematically related together with their own spawning system, that adds to the counters of the overworld system. The soloers and crap outside benefit directly from the groups farming dungeons, which have their own internal spawning systems and counters. These dungeons have two versions: quest version and open version. You guessed it, quest version is an instance, designed with whatever story people want to put in there and have max player limits and all the shit that happens normally in instances. Quests are rare out in the overworld, because that's just not a safe place for people to be standing around handing out bear-ass collecting quests. Anyway, same layout, just quest specific stuff in the instances and depending on the quest varying amounts of mobs. The open dungeons are just like EQ dungeons, except instead of having static spawns for most the rares, they can basically pop anywhere. Fuck camping. Any number of people can be in there, and the more mobs that die, the quicker the "rare" mob counters pile up. Enough rares die, a 1+ group mob spawns. This guy counts as being an overworld king, so the raiders benefit from people killing shit in these dungeons. I'll stop here because you get the idea by this point, I'm sure.
What this requires is a moderately well kept database, and then artists and a couple of designers to keep developing successive tiers. Again, using modular art and ability skillsets, you can create progressively cooler and stronger npcs as time goes on, with less effort. Then, you still have a secondary expansion team that doesn't function like the old days of "Boom!" and done for a year+. Have them be constantly working on developing zones and new npcs and new content that is released every couple of months, but -still- gated behind the progression level of the server. This would also have the effect of getting a Gates of AQ type mentality that the whole server needs to work together to open up successive tiers of content. But instead of getting an endgame raiding zone and a 20 man, you get a new overland area, dungeons, more mobs (reusing some of the older ones tastefully, of course) and progressively stronger raid mobs that require the server to continue to work together to push past.
You can then have statistics on the database on how fast people are progressing on each server, and dynamically alter the rates of various counters and what not with a more hands on approach if you want people past a certain tier to enjoy your hard work. Or even slow down content production if it isn't being consumed as fast. Anyway, I enjoy typing.