5 ms·

This is not correct: - if there are multiple users per netflix account, you underestimate the total number of netflix users. - if the total hour-count inclu

This is not correct: - if there are multiple users per netflix account, you underestimate the total number of netflix users. - if the total hour-count includes users that have only watched 10% of the movie, you over estimate the number of users that have actually watched the entire movie. (by counting in the ones that haven't) In both cases you end up over estimating the fraction of netflix users that have watched the entire movie. If you assume that there are two users per netflix accounts on average, and that 10 users out of 11 will drop the movie after 10% then your estimated percentage will be 4x higher than the real percentage. So, its probably more around 9-10% of netflix users that have seen the movie.

slg 5y ago
>- if the total hour-count includes users that have only watch 10% of the movie, your estimate over estimates the number users that have actually watched the entire movie. (by counting in the ones that haven't) The number Netflix gave us was total time watched. If the average watch time was less than the runtime, that means more people watched the movie for at least some period. That is the number I was highlighting, total viewers and not the number of viewers who watched the entire movie. The existing metrics for theatrical releases is revenue and tickets sold, not people still in the theater when the credits roll. EDIT: To use your numbers, if 10/11 dropped after 10% and 1/11 watched the whole thing, that would mean the average time watched was around 25 minutes. That yields a total of 364m streams. At 2 users per account, that is 182m accounts or 87% of Netflix subscribers watched at least part of the movie. That is obviously a huge increase to that 37% from my prior estimate.
- bnegreve 5y ago
  > If the average watch time was less than the runtime, that means more people watched the movie for at least some period. I'm suggesting some of this total watchtime is not meaningful (<1 min watch time) and may still contribute to a significant proportion of the total watch time. You assume all the watch time is meaningful. You also assume there are no users with multiple accounts. I think these two assumptions are not very reasonable, and can explain why you find your own estimate "literally unbelievable".
  - jobigoud 5y ago
    If there is a lot of < 1 min watch time contributing to the total global watch time it means the share of accounts that started the movie at all is even higher. This is what is surprising the grandparent.
  - slg 5y ago
    >I'm suggesting some of this total watchtime is not meaningful (<1 min watch time) and may still contribute to a significant proportion of the total watch time. How does this contribution a significant proportion to the total watch time? Even the last example that had 90% of users not finishing the movie required 87% of subscribers to at least watch some of it. Do you think the number of subscribers who watched even a second of this movie is higher than 87%? >You also assume there are no users with multiple accounts. Do you mean a person with multiple different Netflix accounts or multiple people sharing a single Netflix account? I have no idea why someone would do the former and I would be shocked if there was any sizable number of people who do that. If you mean the latter, my previous comment used the assumption that each subscriber equals two users like you originally suggested. >I think these two assumptions are not very reasonable, and can explain why you find your own estimate "literally unbelievable". Ok, why don't you throw together some numbers on the estimated percentage of Netflix subscribers who watched any part of this movie?

slg 5y ago

>- if the total hour-count includes users that have only watch 10% of the movie, your estimate over estimates the number users that have actually watched the entire movie. (by counting in the ones that haven't) The number Netflix gave us was total time watched. If the average watch time was less than the runtime, that means more people watched the movie for at least some period. That is the number I was highlighting, total viewers and not the number of viewers who watched the entire movie. The existing metrics for theatrical releases is revenue and tickets sold, not people still in the theater when the credits roll. EDIT: To use your numbers, if 10/11 dropped after 10% and 1/11 watched the whole thing, that would mean the average time watched was around 25 minutes. That yields a total of 364m streams. At 2 users per account, that is 182m accounts or 87% of Netflix subscribers watched at least part of the movie. That is obviously a huge increase to that 37% from my prior estimate.

bnegreve 5y ago
> If the average watch time was less than the runtime, that means more people watched the movie for at least some period. I'm suggesting some of this total watchtime is not meaningful (<1 min watch time) and may still contribute to a significant proportion of the total watch time. You assume all the watch time is meaningful. You also assume there are no users with multiple accounts. I think these two assumptions are not very reasonable, and can explain why you find your own estimate "literally unbelievable".
- jobigoud 5y ago
  If there is a lot of < 1 min watch time contributing to the total global watch time it means the share of accounts that started the movie at all is even higher. This is what is surprising the grandparent.
- slg 5y ago
  >I'm suggesting some of this total watchtime is not meaningful (<1 min watch time) and may still contribute to a significant proportion of the total watch time. How does this contribution a significant proportion to the total watch time? Even the last example that had 90% of users not finishing the movie required 87% of subscribers to at least watch some of it. Do you think the number of subscribers who watched even a second of this movie is higher than 87%? >You also assume there are no users with multiple accounts. Do you mean a person with multiple different Netflix accounts or multiple people sharing a single Netflix account? I have no idea why someone would do the former and I would be shocked if there was any sizable number of people who do that. If you mean the latter, my previous comment used the assumption that each subscriber equals two users like you originally suggested. >I think these two assumptions are not very reasonable, and can explain why you find your own estimate "literally unbelievable". Ok, why don't you throw together some numbers on the estimated percentage of Netflix subscribers who watched any part of this movie?