So, for example, Netflix may know that the TV show ‘Breaking Bad’ got X number of viewers per episode, and Y amount of people 22-30 years-old watched each episode all the way through, but Netflix currently does not have a way of measuring what makes audiences stop viewing. They also don’t know what keeps people viewing up through the end of an episode. Do they stop because they find a particular storyline or character boring? Are they more engaged in episodes that focus on one character over another?
In the future of streaming video, generating data using this reaction/timestamp model will be a way for streaming services to ascertain that information and so much more.