It looks like you're new here. If you want to get involved, click one of these buttons!
One reason for the decline of the subscription model is that some people simply aren't willing to pay a subscription for a game. People who want to buy a game don't like the idea that they're just renting it and will lose access after a while. When you quit a subscription game and consider coming back later, having to pay before you can come back and look around is quite a hassle.
One solution to this is buy-to-play. There isn't a subscription to end, so you don't get kicked out after a while if you stop paying. But this has drawbacks of its own. If you're going to make a game buy to play, you'd better make that one time purchase bring in a considerable amount of money or else you don't get much revenue for your game. Alternatively, companies may tout the game as buy to play, but still try to supplement their income from an item mall and get a large fraction of their revenue that way. For those who want to be free from pay-to-win item malls, that can be problematic.
From a player's perspective, you might think, well then, they can make more content and we can buy access to that, too. But this creates two problems. First is that it's hard to create large amounts of good content quickly. Second is, what if a large fraction of your players decide that they don't need the new content and decline to buy it? Companies try to combat this in expansions by making the new content offer far better loot than the old, but that just deprecates old content.
So I'd like to propose a solution. For simplicity, I'll start with a model that has an obvious flaw, then come back to correct it later.
When games increase the level cap, they usually increase it by a lot all at once. They often add a bunch of new content when they increase the cap, with the idea being that you play the new content to get to the new cap. Then the level cap may stay as it is for a year or two before being increased again. This also leaves some players bored while waiting for the next batch of new content.
What I would propose is that instead of this, you raise the level cap every month--but only by one level. You create new content continuously, rather than releasing a bunch all at once in one big expansion. You keep content available up to several levels above the cap. If the cap is 60 and you want to go into a 65 zone, then have at it, but don't complain if you die in three hits. Maybe the content for the higher level zones isn't entirely polished at first, but few players will go into it well above the cap, and they test it for you and help find bugs that you can fix before more than a handful of players can trickle in.
So what does this have to do with a subscription model? When you pay for a subscription, you get access to everything in the game. When your subscription expires, you still have access to everything, but your personal level cap is frozen at whatever the global cap was when your subscription expired. If you want to increase your cap again, then you subscribe again.
For example, let's suppose that the cap is 30 at launch. If you buy the game at launch but don't subsequently subscribe, then you can play the game forever, but never level past 30. You can go into a level 40 or 50 zone once they're available, but if you get one-shotted, it's your own fault for being a level 30 in a 50 zone. If you subscribe for a while and then quit, what you've previously paid for will never be taken away from you.
However, if you want to always have the maximum level, you'll need to subscribe every month. Make it so that being a level lower than everyone else for level cap plus content is a considerable setback, so that anyone trying to do stuff at or near the cap will need to subscribe continuously. In that way, it's like a subscription model, and a considerable fraction of your active players will likely maintain an active subscription.
This also handles the endgame problem. The endgame is doing future content that is a few levels above the cap. The bigger your group, the higher the level of open-world content you can go after--and the better loot (because it's higher level) you'll get for it. Make any level requirements on gear such that if you're at the current global cap, you can use it, even if in the future the cap would be a few levels higher. But unlike raids from previous expansions, the content never becomes deprecated, as it's the normal content for its level for people who come through a year later when it's no longer near the cap.
Now for the obvious flaw: let's suppose without loss of generality that the cap is increased on the first of each month. Any monthly subscription effectively lasts until the first of the next month and then expires, as in the middle of the month, it doesn't matter if you're subscribed at exactly that moment or not. If it's near the end of the month and you subscribe, then had you waited a few days to start it, you could have effectively had your subscription end a full month later. Telling customers, "Yes, we want you to subscribe--but not for a few days" is a bad idea.
The solution to this is to have many more levels. Instead of increasing the cap by one per month, increase it by one per three days or so, so that the 30-day subscription nets you 10 levels. But that doesn't mean that you need 10 times as much content; you subdivide each previous level into 10 parts, so one level only does you 1/10 as much good, and only takes 1/10 as long to gain. Think of it as adding a decimal point to levels with one digit after the decimal point if you like.