It's almost obvious. After all, every region has one or two, maybe three or four, religions which are dominant, and almost always some isolated communities belonging to different faiths, with little political influence. This has always bothered me in Civ - if a religion enters a city, it's suddenly treated as completely equal to each religion already present. So how about we make a system which differentiates between a city's major religions and the religions that are simply present? This isn't really all that complicated. We don't have to have an actual shifting ratio of believers between different religions in the city. There will simply be a few points where a religion's status can be altered. The whole thing can also be displayed in a very simple, intuitive way: major religions will have icons about 125% the size of current religion icons, and minor religions will have icons about 75% the current normal size. Neither size will be too big for the interface, and the different is big enough for anyone to easily notice. So let's see at what points status can be altered:
a.
When a new religion enters a city
For each city in the empire where the new religion is major, that religion will have a 10% chance of starting as major immediately as it enters the city. Add to that 1% for each foreign city where the religion is major. If the religion is that city's state religion, double the chance. The maximum chance should be 95%, so you still sometimes get some isolated areas where the empire's major religion is not so dominant.
As to the existing religions in the city, if the new religion start as dominant, each existing dominant religion should have 25% chance of becoming minor immediately. Each minor religion should have a 10% chance of becoming major. (All due to the sudden cultural upheaval in the city.) If the new religion starts as minor, each existing religion has a 5% chance of swapping status.
b.
When a city swaps hands
Whenever a city is traded or conquered, the upheaval should cause changes in the religious power balance within the city. This is a bit tricky and complicated. First of all, the new controller's state religion should have a good chance of going major (30%?). Additionally, if the new controller is using Theocracy or Organized Religion, existing major religions should have a good chance of going minor (20% each OR, 40% each Theo?). Lastly, the upheaval should generally cause some changes - a small chance (5%?) for each religion to swap status. Perhaps these three effects should be combined and remixed.
c.
When religious improvements are constructed
Whenever a city receives a new religious building or wonder, that improvement's religion should have a chance of going major, and other religions should have a chance of going minor. The chance should be based on the improvement (wonders should be very influential, cathedrals less, temples even less) and major religions should be rather unlikely of going minor because of this event.
c2. A tiny chance of status swaps should exist whenever a city produces a missionary because it means that city is endorsing a certain religion. The chance should be a universal 1% for each religion to switch, or something along those lines. Building armies of missionaries will make a difference, building one or two usually won't.
d.
When state religion changes
Whenever state religion changes, all cities in that empire should have a good chance of seeing the new state religion going major (40%?) and the other religion going minor (20%).
e.
When Civics change
When the religious Civic is changed to Theocracy, all non-state religions should have a 30% chance or even more to go minor, and the state religion should be extremely likely to go major all across the empire (60%?).
x.
Dying religions
Optionally, when a city gets a new dominant religion (either from a swapped minor religion or from a new religion bursting onto the scene), existing minor religions should each have a chance to just disappear from that city. This way when a religion suddenly starts to prosper it will slowly kill old religions.
y.
Push to determinism
Optionally, some religions could have an inherent bonus or malus to their chances to go major. This way Judaism could be made more unlikely to go major over a large area, and Islam could be made very dominant. This would actually be a very good solution to the problem of Judaism as a multi-national state religion.
z.
The Holy City effect
In a religion's holy city, that religion should always have half the chance to go minor and twice the chance to go major. In fact, when a religion is founded it should be founded as a major religion in its holy city. By the half-and-double effect later on, the religion will mostly stay major in that city (except for scarce and brief periods of minority).
w.
Free Religion
Under the FR Civic, we may want to make all the empire's religions automatically minor for as long as the Civic persists.
Of course all of this is meaningless if there are not special effects to the difference between major religions and minor ones:
a.
First and foremost: political weight
Basically, major religions will be given way, way more weight in considering the selection of state religion. That will already do a hell of a lot to improve realism. A major religion should weigh in at least five times as heavily as a minor religion - Christian Rome flukes will be possible, but not likely. Usually a civ will select as state religion that religion which is actually dominant in its cities. As a side note, the calculation of religion spread should give major religions the same relative weight as they get in consideration of state religion.
b.
Happiness
Major religions should make the people happy, minor religions should not. Minor religions are too small to make people happy on a scale that can be shown in this game. The exception is Free Religion where the diversity itself should make people happy (in principle).
c.
Construction
Major religions should get their buildings and wonders a bit quicker than minor religions. This will not only make major religions more important, but will also help preserve the status-quo by making it harder to build the minor religions' buildings (which would upset the religious power-balance.)
d.
Unit production? (Patriotism)
It might make sense to give a tiny (5%?) malus to build speed on units in cities where the state religion is minor. This would just be a way to show that the major religions, when excluding the state religion in a given city, would be less cooperative.
e.
Trade routes (the counterbalance)
To balance all this, for every three minor religion in a city, that city could get one extra

, representing the connection with the minor religions worldwide. This would make it like in real history - rulers often grudgingly tolerated minor religions because the people of those faiths benefited society in some way.
Finally, to make it all come together, all religions would have to spread much faster, because that way you get most religion all over the place, but not so often as state religions. This is the essential problem in the current model: you can't show how far religions have spread without screwing up the international power-balance.