The sad part is that most Japanese game companies make games the way they should be made -- with the independent spirit of an arthouse approach.
We often read that the CEO also wrote the scenario or designed the characters or composed the music in the game... that definitely sounds like a bunch of guys going "Hey, we're doing this for fun. Since I'm the project leader I guess that makes me the CEO, you're the CTO I suppose, etc." That's why it's such a fine line between doujin circle and professionals over there, unlike in the West.
It's definitely different from the western approach where people who don't even have exposure to games get hired from (the auto industry, airline industry, whatever) to take up vice president of marketing, etc.
http://www.andriasang.com/e/blog/2011/02/17/fps_sales/
He refers to 100,000 as being a "mini smash hit."
Those numbers are usually not enough to justify a western localization. I've read numbers such as 250K being tossed as the minimum to break even on localization costs (e.g. for a Tales title). The only one risking it is NIS America.
Regarding low budget stuff such as visual novels, RPGs, strategy, they're actually closer together than you think. SRPGs such as Utawarerumono, Tears to Tiara, and plenty of Planet Stronghold / Wizardry - style first-person battle RPGs are essentially just really fancy visual novels with gameplay mechanics and slightly higher budgets. In fact even sites like Siliconera attempted to lump Agarest War (a well-known console title) as belonging to essentially the same bunch.
Adding a battlemechanic is the common next step in advancing in the game hierarchy (if we consider visual novels as on the bottom of the pile; barely being "games") as long as you have a competent programmer. The biggest jump in budget would actually be 3D... that's why it took so long for GUST to make the leap.