Co-op, custom maps and mods are in fact similar to multiplayer - they create a stable playerbase. In every statement in my last post you can change the word "multiplayer" into "custom maps" and it will be still valid. The problem is not enough stable playerbase after 3 weeks from release.
Because the game has been released in a relatively barebones amount compared to previous games that have had a ton of content put into the game over their dev cycles. And no, ladder games are completely detached from MP in terms of both design and gameplay for the most part. Relic's biggest mistake was letting dawn of war 3 ship out with nothing else to keep the non ladder players engaged with the game which is why people finished the relatively meaty campaign and then stopped playing.
Of course 5k is a good amount for RTS. It would be amazing amount of players for B class RTS. It is also a great amount of players for over 3 years old game such as CoH2. But it is definitely not a good score for 1st or 2nd most awaited RTS in 2017. This is not a good score for any AAA game just after release.
And here we come to the most sensible conclusion - the DoW3 is a game advertised as AAA, but in fact doesn't live up to this standard. That exactly explains the player drop down and it can be seen on every corner in the game, game streams or reviews.
What metric are you judging the numbers by? DoW3 had a higher peak player number than every other dawn of war game on steam. The only other RTS game apart from coh2 that has a higher number on steam is AOE:HD which is a remaster of AoE2.
BW: remastered is coming out this year, and so is total warhammer. Those games have a far more robust modding community than dawn of war 3. And what makes Relic a AAA studio? They have released 1 new product since they were brought out by SEGA and their support team is tiny. Dawn of war 3 is for all intents and purposes developed by a B-class RTS developer and it has a standard number of players given it's small amount of Co-op content.