How to explain... [I have no experience in video editing] To create such a video, I see two different ways:
1. film a movie of real, new, spotless device touched by real person (and cut out any fingers or hands so that they would not be seen, and add these circles manually later as a guide to show where the fingers touched the screen);
2. create a computer 3D-model of the device (maybe, based on high-quality photographs of the device in ideal condition), film a screencast at the device (with debugging-style circles-dots included), and mishmash the rotating 3D model with the screencast so that it would look like a video shot in a dustless space with invisible fingers touching the screen.
The latter can be higher-quality because, unlike the former, it would not require cutting out the imperfections of real world. But the latter also requires attention and precision from the designer of the video. It also might make adding special effects easier.
Of course, it is possible to add those circles with video editing. But it's time-consuming, boring, tiresome work - and it is so much easier if these circles are added automatically. Besides, having these circles turned on all the time could be a nice accessibility feature - to confirm that you have, indeed, touched the screen, clicked it, or double-clicked it. I had once opened about two dozens of MicroB browsers just because of impatiently clicking a link and not knowing whether the click registered or not.
Best wishes. Thank you.