A good demo for a good game was minor advertising that was dwarfed by good press. If every player wont shut up about how good the game is, their friends would skip the demo and buy anyway.
A good demo for a bad game was good advertising that bit you later. You got more up front sales, but got harder drop offs once word gets around that the demo was all you had.
A bad demo for a good game stuttered sales. Some people would turn away and maybe never come back, and it took time for word of mouth to tell everyone to skip the demo and just buy the game anyway.
A bad demo for a bad game was shit all around.
In the end, this punnett square made it pretty clear that the best option was to make a really good demo if youre game was shit, or you thought you needed the help finding an audience. but if you knew (or “”“knew”“”) your game was good? The demo was wasted time and effort. Either it was a smaller ad bump you werent upset to cut costs for, or you were slowing sales by accident.
Demos are good for us, but suck for the company making them. So they largely stopped making them.