Lessons learned: Time Dilation Video (Patreon)

A couple of weeks ago I posted a video about “Time Dilation” on YouTube. It reached ~670 views and it  is running out of steam now. My former video on Length Contraction was stuck on ~50 views for the first month and slowly exploded on its second and third months, reaching 150K views. Unfortunately, I think this won’t happen for my last video. In this post, I’ll try to explain why. You could find this information useful, if you are also posting on YouTube, TikTok or similar.

I’ll start with the data I get from the “YouTube Analytics” and give my thoughts and justifications.

Poor video thumbnail

Only 3.3% of the people who saw the video thumbnail decided to click on it.

The main cause for this problem was me rushing to publish the video. I put together the initial thumbnail in 30 minutes and it was poor. See below:

Figure: first thumbnail used for the "Time Dilation" video

The thumbnails that followed were not great either. The other problem is that I published the video just before going on holiday. This gave me little time to fix problems as they emerged (Note: YouTube allows you to change the thumbnail as many times as you want)

Retention

34% of the viewers left the video very early, before the end of the introduction.

I overestimated how long many YouTube viewers are prepared to spend on an introduction. I suspect this is close to zero for unfamiliar channels.  A couple of days after publication, I indeed decided to cut most of the introduction in the video (YouTube allows you to cut parts of your videos.) This helped a bit, but it was too late and not drastic enough.

Instead, the video should have started with a “hook sentence”. Something like “Think you know how time works? Wait and see…” or something like that, but maybe less aggressive.

The video also starts with a message on a black screen. Not very attractive.

A better soundtrack could also help people stick with the video. Also, I should consider cutting the music intro and let it start where it is likely to engage more deeply with the audience.

Engagement

Poor engagement, even after the intro: the video keeps losing viewers steadily at a rate of ~9% per minute.

Figure: Percentage of viewers still watching at the given time.

Poor like count. 58 likes and 4 dislikes on 670 views. Most of the likes/dislikes came from the first hundred viewers.

Nobody commented on the actual Physics of the video.

I received 7 comments out of 670 views. That’s not great, but not awful per se. However, there were no comments focusing on the actual Physics of the video.

No questions, typically means that you lost your audience.

I guess the video had too much content. Relativity is hard to digest. Presenting too many counter-intuitive facts all at the same time may also cause rejection on the viewer. Less may be better than more.

The content should be explained so that the viewer understands and accepts it. The video, instead, just shows what the results of the thought experiment are, without trying to motivate them.

Also, captioned videos are hard to follow. I had explicit comments about the captions being too fast. Certainly, I could have slowed them down a bit. Also, a duration of 4-5 minutes is maybe too long for a captioned video. For videos of this length I should maybe do a voiceover.

Conclusions

The video took 10 days to make, although I worked significantly harder than usual during these days. 40% of this time was spent working on logic and assets that I can re-use in the game. The remaining 60% was work dedicated to the video and not reusable in the game, although it gave me various insights. First it was a materialization of a thought experiment that before only lived in my head. It confirmed what I had thought would happen! Second, it gave me insights on the limitations of my software. These could help in architecting some components of the game (in particular, the dialogue system.)

I honestly think the content of the video is original. Not something you can find elsewhere. Personally, I think it’s pretty cool.

However, in terms of actual success on YouTube… It was poor. I had no additional subscribers on YouTube. After two weeks, YouTube has pretty much stopped showing it to anybody.

I believe very simple changes could have led to a significantly better outcome. A better thumbnail could have easily doubled the viewers. A more catchy start could have led to better viewer retention. Simply cutting the video and reducing its size could also have had a significant impact.

Could all these actions, put together, have led to the 150K views I had with the length contraction video? Not quite, maybe. To get there, I guess, I should have tried to explain things better. It would have required time that I preferred to put into game development, instead.

It is worth noticing, however, that the "length contraction" video went pretty badly on the first publication on TikTok. I then removed the first 15 seconds and it did much better, reaching ~6K views. Readapted on YouTube, the video did very well. This proved that very simple changes can make the difference and push a video 100 to 150K views.

I think I should give TikTok a try, maybe in a month or so. This shouldn’t take too much time and would help validate some of the hypotheses above.