Notes on playing a game called [NishioGames - RoboMasterS1 assembly section was completed.

  • A meta look back at the video editing process
    • Assembled and photographed for about 3 hours on the night of 8/23
    • I bought Premiere Pro on the evening of 26.
    • Three weeks since the final video of the assembly section was uploaded today, 9/14.
      • You mean one week per hour of original video.
  • It seems to me that taking a long video slows you down because of the “where do I split it up” problem.
    • And we can’t shoot the outro until we split it up and finalize the content.
    • A shorter shot might be more efficient.
    • If you put up two videos a week, that means each original video is 30 minutes long.
    • I’d prefer a pipeline that takes one pomodoro of the main story and an intro-outro together.

image image

  • 40% from external sources, 17% from search

    • Surprisingly large number of searches
  • 9% of impressions are recommendations within YouTube

    • As this increases, the “segment that reads a lot of YouTube videos” is watching their own content one after another.
    • I want to increase image
  • I don’t know why some segments are not showing impressions figures.

  • Browse Features is a segment of the channel’s post-subscription page and a bunch of other stuff.

    • When you subscribe, you’ll see a list of recent videos on the app’s home screen, or something like that.
  • Right now, I’m making 40% on External.

    • Most of them are my Facebook and Twitter posts, so they don’t contribute much to the increase in subscriber base.
    • We need to create a situation where we can consistently earn IMPRESSIONS with YouTube Search.
  • The game’s clear condition is “1,000 Subscribers,” although the View is a prominent indicator, so attention tends to focus on that.

    • Currently 29 people, so progress is about 3%.
    • Acquisition of new users is important
      • But to do so, we need to increase our exposure on YouTube.
      • This is a black box.
  • The next tuning point would be the impression click-through rate.

    • image
    • The choice of title and the way the thumbnails are made make a clear difference.
    • Getting impressions is meaningless if they are not viewed.
      • Phenomenon that the chassis version is not seen.
      • 3, 4, and 5 are all low across the board for assembled videos with a “category + number + content” structure.
      • Even if there is a manga that looks interesting at a manga cafe, if there are only three volumes, I won’t read it.
    • However, as for assembly videos 3, 4, and 5, they are inherently bad.
      • To begin with, it is “the 3rd~5th of a 3 hour video divided into 7 pieces”.
      • Not independent content as a stand-alone video
      • Increasing click-through with thumbnails does not lead to customer happiness.
      • I prefer it when people see “Let’s open the package” and say, “Let’s watch the next video.”
      • I checked and the end screen for episode 3 didn’t specify episode 4, probably because there was time left before it was released.
  • In the beginning, the number of videos was small, so we were fixed and linked to each other, but as the number of videos increased, we stopped.

    • For example, people who started watching this channel with “I played with it”, which has the most views.
      • It’s pointless if you get “played” on the end screen of another video.
    • If it is a continuation, it MUST specify the next video
    • Still, there are two slots left, and until now I’ve been tuning this one.
      • Trust YouTube’s algorithm to change to “recommend the best video for the user”.
      • The other should be “latest video”.
        • As the number of videos increases, the system concentrates search inflow access to the latest videos.
  • Increased search inflow

    • Thinking about voice input search, open difficult-to-type strings to kana.
  • What we would like to target in the future

    • Submission as in the game
    • 1: Increased click-through rate: now 7.1
    • 2: Improvement of impressions via recommendations: 9% now 1.6K
    • 3: Improve search impressions: aim to increase actual number by 677 (6.2%) in the next 3 weeks.
    • If we narrow the range to the most recent 1w.
      • Click-through: 4.6
      • Recommendation impressions: 9.1
      • Search impressions: 14.2
    • Only the click-throughs are getting worse.
      • image
      • The “I’ve played with it” and “I’m going to open it” that were over 10% in the 3W are down considerably.
      • In other words, videos, to begin with, “have a high click-through rate when they are first published, and then it goes down.”
      • Click-throughs in a short period of time are strongly influenced by the timing of the most recent video release and that video, so you probably shouldn’t worry too much about it.
      • Would it be better if it were limited to search impressions only?
        • Theoretically, yes, but we don’t have a large enough sample size for the statistics to make sense at present.

This page is auto-translated from /nishio/3週間振り返り using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I’m very happy to spread my thought to non-Japanese readers.