Allow me to reiterate: a game is considered "alive" if and only if the following two criteria are met:
- Waiting times, as you have said, are low.
- The matches that occur within that matchmaking process are as fair as possible.
Let's say the first criterion stands (which it does not but whatever). The second criterion is so out of reach for COH2, that it's like a dream.
For further info, check out my other threads and Rosbone's. Every match that is produced by COH2's matchmaker would be consider an outlier by all the other major games' matchmakers.
Waiting times are usually (2v2 and 3v3) below 5 minutes from my experience when playing with friends. Most of the games we get are also decently balanced I would say, with some one sided games every now and then.
I can't fully speak for pure random games and 4v4, since I play both of these rarely.
However, we'd need to know the percentage of really unbalanced games. They surely happen, and in 4v4 more than anywhere else. But they also happen in highly populated games that have tens of thousands of players. I can tell by from playing rocket league. Most matches are really good there, but even with a large player base, sometimes the other team is just a decent chunk better/worse than mine.
The only estimate that we can make is look at game lengths. Top200 1v1 is probably the closest base line info that we have to perfectly balanced games. Still, 11,6% of games end with 10 minutes or less, indicating these are either drops/technical issues or unbalanced match making. 10-20 minute games happen fairly frequently, too. Since they also happen often in the top200 games, I'd assume these are rather lost games due to bad playing or one player honestly surrendering than bad matchmaking.
Interestingly though, all 4v4 games have a ratio of 13,9% of <10 minute games. 3v3 has 13,4%; 2v2 12,8%. 2v2 top200 11,6%. And all 1v1 has 14,7%.
Percentages are not vastly different and some could also be explained by drops, afks and technical issues that have not much to do with the match maker. Either this, or the "metric" I chose here is nonsense in the first place.