ATCkit There are content editors whose job is to select books they believe can be popular with the international crowd or used as experimental targets (there are some we received that we know definitely won't be popular—like Dead On Mars, a sci-fic which I translated. Everyone knew it wouldn't do well because we know what kind of audience Webnovel has. But there are some times certain things that need to be done regardless. The ones selected are already filtered. There is a lot more crap in China.
The web novel industry breeds the habit of readers coming daily, and the readers coming daily also reinforces this requirement. It reinforces them in a (vicious/virtuous) cycle. The ones who return after a break seldom actually pay anyway.
Also, there are times when everyone just gets it wrong. It's like a choosing a script to produce into a movie. Yes, a good director, actor can choose one that they professionally believe that will do well, but there are flops. Likewise, there are success stories that come out of the blue.
Also readership dropout doesn't usually happen at a specific point in the story. It's a continuous drop. Trying to model this as something that can be correlated to the plot is a very, very tough problem. I've looked at similar data. It just looks like noise.
What you said are things everyone in the industry knows. It's not as simple as that. As I said, the readers' wants are also constantly changing. So you are adding a time component to tastes that keep changing.
Also you make it sound like the reason for people dropping might be due to a certain trigger in a plot. And that this can happen to all books that have the same trigger. Like say rape in female-lead novels. Some readers just drop out at rape spots, but in another book, they readers don't. Is it because of the way is written? The build up, etc? It's way more complicated than the way you put it.
In fact, a good content editor can have a 50% hit rate of popular books just from the TITLE alone.
For instance, there was a time when Eastern Fantasy was popular. To meet demand, Webnovel did a lot of those for translations. It later evolved. There was a time when games were popular, so they offered a lot more games. Trial Read is actually a reflection of the customers' preference, no matter what people say about how much a gem something is. There are also a lot more finer details about the kind of plot device is used, but it's all web novel theory. After all, this industry has been in running for 20 years. A lot has been researched by Chinese authors and editors. The general ideas are the same both in China and internationally, perhaps with some minor tweaks due to the different phases of reader preference.