Often it's a vertical slice with gameplay mostly done and assets of medium to high quality.
Since assets are expensive to make this allows you to build a mostly complete product without cutscenes and 50 levels but without too much effort.
You often don't ship this MVP to the public but use it to attract publishers or to demo at game conferences.
[Oped below]
So much of what makes a game good is art and feel based so testing early and often whether you're on the right track is important. Game studios that are smaller have to do this. Larger ones can take the hit if they make a bad call in favor of controlling the marketing/presentation more carefully. I'm not entirely sure how they test. Similar to developers I think most people in the game development industry have good or reasonable instincts about games that would work.
You often don't ship this MVP to the public but use it to attract publishers or to demo at game conferences.
[Oped below]
So much of what makes a game good is art and feel based so testing early and often whether you're on the right track is important. Game studios that are smaller have to do this. Larger ones can take the hit if they make a bad call in favor of controlling the marketing/presentation more carefully. I'm not entirely sure how they test. Similar to developers I think most people in the game development industry have good or reasonable instincts about games that would work.