Zeha pointed out that some developers suggest having ONLY a video and NO demo. I’m quite sure there can be found arguments for both views, but I think in the end it boils down to simple math and testing:
First you one can test selling your game by showing only a video and NO demo. After you’ve done this test you can easily see how well (or poorly) “video + no demo” method works. After this (or using separate website) you can test using video AND demo. Now by comparing the results you can tell which method worked better and continue in that direction.
If you have lots of newsletter subscribers, you could construct the test by sending 10% of the people the “video and NO demo” and to 10% “video AND demo”. After this you can tell which 10% performs better, and then you send either video with demo (or no depending which 10% performed better) to the rest 80% of the subscribers.
Making arguments is okay, but I personally think this is a matter of testing and analyzing the results rather than giving opinions.