It’s been about two weeks since YouTube rolled out the COPPA restrictions across the platform where creators need to pick either on the channel or video level whether their content is made for kids or not. As this happened in January, the worst month for ads in general, it’s hard to fully tell just how big of an impact this is to creators in terms of earnings from their videos as they are losing personalized advertisements, but we were interested to see what these changes would do regarding views and subscribers.
For this study we were thankful that YouTube provided a flag when we request data for channels indicating if at the channel level they were made for kids or not. You can see the top list of made for kids channels here or if you look up any channel on our website there is a pink badge on their profile that says Made for Kids if the channel is fully made for kids.
To further clarify the above paragraph, if a channel has decided to declare their content as made for kids on the video by video level they do not get this flag (as far as we know at this point) and are not included in this study. It is likely that the number of kids channels NOT included in this study is sizable, just at this point not as easily detectable.
With that being said what can we see?
Social Blade currently collects statistics on 37.9 million channels every day.
Of those channels, 172,227 are marked as Made for Kids (again, FULLY Made For Kids, does NOT include channels that mark on the per video level).
These channels together have over 6 billion subscribers and over 2.5 trillion views.
Now to look at the changes that happened in terms of subscribers and views we looked at the average number of views/subs the channels got in various time periods, 7 days, 14 days, 60 days, and 90 days. We compared the MFK channels to the entire database as well to see if a trend was specific to the kids channels or across the platform. We also compared channels of various sizes, 1+ million subscribers, under 1 million, and the group as a whole. Also it should be noted that due to YouTube abbreviating subscriber counts in Sept this can cause weird bumps in subscriber gains but when you look at the ecosystem as a whole it should average out but the longer the time period the more accurate the number would be.
One last thing to note when reading these findings, if the 7 day number is higher than the 14 day number it means a positive change is occurring, the rate of growth is faster in the 7 day period than the 14 day period.
Here is what we found:
When comparing the 7 day and 14 day averages between the MFK channels and the non-MFK channels of all sizes both groups saw a subscriber loss in the 7 day period. MFK saw a 6.06% loss, and non-mfk saw a 2.89% loss. If we expand this a little bit further to account for YouTube’s abbreviation and look at the 14 day period vs the 30 day period we still see a loss on the MFK channels of 2.27% but the non-mfk channels actually see a gain of 2.45%.
If we change gears and look at views with the same groups, the MFK channels actually beat the non-mfk channels with a 7 day vs 14 day period gain of 1.58% vs a loss of 2.65% on the general group, though expanding to the 14 day vs 30 day does show a loss of 0.96% on mfk and also a loss of 0.15% on the general.
If you just look at the overall picture of the two groups it doesn’t look like there is too much of a change, little percent changes here and there aren’t exactly statistically significant, and in the short term it actually looks like MFK channels have a little bit of a win. But lets dive deeper.
If we segment the two groups into channels above and below one million subscribers things do start to look a little different. MFK channels having 1m+ subscribers only saw a positive 0.4% change in views when looking at the 7d and 14d periods. When looking at the 14d vs 30d they only saw a 0.66% decrease. Both numbers aren’t high enough to really claim much of a difference. BUT if you look at the channels UNDER one million subscribers we see a 12.98% loss in views when looking at 7d vs 14d, though only 0.79% when looking at 14d vs 30d.
Expanding these numbers a bit more, the 1m+ group saw a 2.76% increase in views comparing the 14 day to the 90 day, and the under 1m sub group saw a 10.07% increase. To check these numbers against the non-mfk group the over 1m sub non-mfk channels saw a 0.6% loss in views comparing the 14d and 90d and the under 1m non-mfk group saw a 1.77% increase in views.
So overall what can we learn from this data with some understanding of the space as well. Beyond the numbers this is speculation but there are a couple of guesses that could be made.
First off, don’t panic. The general YouTube recommendation system had been built up over many years and had a lot of training data fed into it over time. We believe the new system pretty much had to start over from scratch and needs to be fed a lot more info to serve up the best recommendations again. So with that being said, only two weeks of data probably isn’t enough yet for it to be able to run with. It’s like a toddler learning to walk when the existing system has already gone through college.
It does appear that so far channels with 1m+ subscribers should more or less see the same views that they were getting beforehand, though subscribers are down. Smaller channels if anything benefited from subscriber growth though more in the non-mfk group than the mfk group.
Smaller MFK channels in the short term do seem to see a huge drop in views, but when you look at the longer period the picture doesn’t look nearly as grim and even looks good. One potential reason for a large drop in views are the smaller channels don’t have as much ability to an algorithmic change and made more changes on their side which could have included stopping making content or changing things too much.
Another very probable reason is.. well just go watch any made for kids video on YouTube right now. The recommendation section is seemly more large channels than you’d expect on standard YouTube. This could just be YouTube training their machine learning with channels it KNOWS are kids channels before it learns more on its own what is best to serve up.
With all that being said remember that at the end of the day YouTube’s primary goal is to keep people of all ages on YouTube watching as much as possible. If the MFK algorithm isn’t performing optimally you can bet YouTube is striving hard working with the limited pieces of data they’re allowed to work with to improve it to keep session time as long as they can.
At Social Blade we’re going to keep an eye on the data as more comes in and we’ll let you know if anything major catches our eye. If you have any insights make sure to share them with us in the comments, our social media, or the Family Video Creators facebook group which helped bounce off ideas to create this blog post in the first place.