Daily uploading isn't always the answer: a case study
- Thread starter babyteeth4
- Start date
Purple Penguin
Liking YTtalk
We've just started our third channel, and so far it's going pretty well. We have realized how the algorithm seems to favor specific content, and we are trying to keep the theme very consistent. We have also been considering breaking up the main channel into a separate channel for family vlogs (since we have been noticing the same thing as far as consistency and the algorithm favoring certain types of content). We haven't done that yet, though, as I don't want to stretch myself to thin with maintaining uploads on all channels. We have been debating this for months since we were given the advice of breaking the channel up for different content. I'm still hesitant though even though I know it will feed the algorithm. It's so much easier having everything in one spot.
daniel burgio
I Love YTtalk
I'm actually gonna try it because I think my content's pretty good (sounds arrogant, but I have more faith in it nowadays then back when I started), and honestly, it's (insert space here because the system think's I'm swearing) hit or miss getting more than 2 views no matter what frequency I do. Plus, I have tons of videos to go back through and edit.
Last edited:
babyteeth4
Taking over the world... ...one kid at a time!
Yeah our vlogs were the first thing we removed from the main channel because they were doing the worst, so it wasn't as hard of a decision and we already had the second channel (Babyteeth More) set up. Now the vlogs do pretty well on the second channel.
Highfalutin' Low Carb
Loving YTtalk
Do you cross-promote content between your channels?
Stearmans Toy Review
YTtalk Mad
Great stuff! We have noticed this as well. We are getting suggested on new videos but as soon as we have another video posted, those drop.
AquariumCoop
Active Member
I have been watching the discussion. I myself launched a second channel recently, only to 5k subs in the first month. Main channel being at 78k. I was doing it so I could release "grade b" content. Basically things that we shot that after post production just weren't good enough. Or things we shot in batches, we watch the first one not perform as we hoped and then what do you do with the other videos knowing that the formula for those videos isn't the best?
I want to point something out though. Everyone focuses on making the AI/algorithm happy. I think the difference in this change is you're making the viewer happy. And then it makes the algorithm happy. So if you know monster toys do the best on your channel, people tend to watch that, when you offer them something else, say candy haul or something, it's not that it's bad, but it's not what the majority of the content consumers want. It's like a pizza resturant offering a cheeseburger on the menu. Almost purely out of, "it can't be that good, they do pizza here" mentality happens. You might make the best cheeseburger in the world. But you don't get a shot. Now you open a burger shop next to your pizza shop, people are buying cheeseburgers. I think the same thing is happening with viewers, as we segment our content, it lets those choose what they want. Which in turn pleases the AI.
Lets look at an example in the toy realm. lets say parents start up the first video for their kids, and choose for lack of knowing the content better the "monster toy" content. They might always be avoiding candy hauls, because then this makes the kids come ask the parents for candy and they don't want to feed their kids that much candy. Thus the users are training the AI to avoid those videos. You take those same candy haul videos and put em on a different channel, people are coming there FOR candy hauls. No longer will people avoid click on them and in fact click on many in a row etc, teaching the AI that is what the user likes.
I think we all know about the AI/algorithm, but forget that it's sole purpose is to learn from user behavior and make changes. We send the end result of user interactions and the Ai working together. If we can change the user, we can change the AI over time.
I want to point something out though. Everyone focuses on making the AI/algorithm happy. I think the difference in this change is you're making the viewer happy. And then it makes the algorithm happy. So if you know monster toys do the best on your channel, people tend to watch that, when you offer them something else, say candy haul or something, it's not that it's bad, but it's not what the majority of the content consumers want. It's like a pizza resturant offering a cheeseburger on the menu. Almost purely out of, "it can't be that good, they do pizza here" mentality happens. You might make the best cheeseburger in the world. But you don't get a shot. Now you open a burger shop next to your pizza shop, people are buying cheeseburgers. I think the same thing is happening with viewers, as we segment our content, it lets those choose what they want. Which in turn pleases the AI.
Lets look at an example in the toy realm. lets say parents start up the first video for their kids, and choose for lack of knowing the content better the "monster toy" content. They might always be avoiding candy hauls, because then this makes the kids come ask the parents for candy and they don't want to feed their kids that much candy. Thus the users are training the AI to avoid those videos. You take those same candy haul videos and put em on a different channel, people are coming there FOR candy hauls. No longer will people avoid click on them and in fact click on many in a row etc, teaching the AI that is what the user likes.
I think we all know about the AI/algorithm, but forget that it's sole purpose is to learn from user behavior and make changes. We send the end result of user interactions and the Ai working together. If we can change the user, we can change the AI over time.
babyteeth4
Taking over the world... ...one kid at a time!
Yes we promoted our 2nd channel heavily at the launch and also mention it at the end of each episode, as well as when there is something relevant. When we launch additional channels we'll do the same.[DOUBLEPOST=1503361918,1503361641][/DOUBLEPOST]
Good points here. It's hard to admit that not every type of video can be a smash hit, but the flipside is that if you're doing something that people particularly enjoy, the focus of the channel should stick to that, for the good of the channel. You can always make additional channels!
The algorithm does have its flaws though. There have been many times where I checked a couple videos of a certain type (like when I buy a new product and want to see some tutorials/reviews), then I wanted to move on, but the algorithm just keeps suggesting them. Of course, it eventually learns to stop suggesting them when you skip enough of them. But there have been times I've deleted my history to stop the suggestions sooner.
Gamayun
I've Got It
I am still testing subjects of my videos with Al. In my field everything got mixet up. I hope to get solid results soon. I use to make videos in correlation with my most viewed videos and it did not work lately.
