watch me at youtube.com/20unc06

Loading...
My Photo
Hi, I'm Danielle (a writer, digital marketer, casual runner, and whatever other labels you want to pick and choose from). I also happen to be transsexual. I have a sneaking suspicion that it'll be a while until I publish my first best seller, so in the meantime, here are my thoughts on everything.

Friday, May 24, 2013

The Analytics Behind One Million Views of Trans YouTube Content

My YouTube channel recently passed 1,000,000 lifetime views (probably on/around May 21). Being a digital marketing nerd, and not having a lot of insight into how YouTube analytics work (and there's not a lot of people talking about it) I thought I'd break it down a little bit to point out some general observations. Obviously, the nature of the content limits is transferability somewhat - but it does offer some interesting discussion points. YouTube Analytics aren't as robust as Google Analytics, so that does limit what I can show you unfortunately.

Before we get started, I should point out a couple of things.
  • This is the "lifetime" views as of halfway through May 24, 2013
  • The first video was posted August 11, 2012
  • There were 55 subsequent videos
  • Finally, the first video's metrics closely mirror the channel as a whole, mostly because it's an enormous outlier. Statistically, this means the entire data sample is horribly skewed. For example:
The top thirty videos (click any image to enlarge and/or view a slideshow of just the data):

Compared to videos 4 through 26 (which is still pretty skewed, but at least legible):


The brief descriptive titles also suggest that certain types of content are more popular than others, as does the fact that the top three - which so clearly dwarf everything else - only have pictures and music (and show a "transformation") and were titled based off keywords mostly. Discussing the nature of the content is a separate beast though. 

I wanted to do an analysis of 1,000,000 views because it's a catchy number, so here's what the total data set looks like in aggregate over the lifetime of the channel content:


This is how the views look at the weekly level:

At the monthly level:


And at the daily level (the one that is most interesting):

Obviously you can see a trio of outliers: 
  • The first jump in traffic (Oct. 22, 2012) is the result of my first mtf transition video actually going a little viral. It appeared on videosift.com and reddit (not sure which came first, but I imagine one led to the other). As a result, views temporarily spiked before trailing off to normal levels.
  • The second spike (Jan. 10, 2013) is the result of being recommended by another girl's transition timeline video. I actually know her via a trans internet forum (and you can check out homegirl's video here), but her video was picked up by HuffPo (and as a result, some other media) which blew it up, it went from being a pretty popular video to millions of views in the span of a few days (we had a discussion about it actually). Simply being recommended by her video spiked my traffic.
  • The third spike (Jan. 25, 2013) is a result of being shared on the Wipe Out Transphobia facebook page (I did not ask for them to post it, and do not like the language around me being "more down to Earth", but I digress). The content was obviously relevant to the audience and as a result they were pretty engaged with the post - sharing it 193 times, commenting 151 times, and hitting 'like' another 804 times. That drove both traffic and engagement on my YouTube channel.


Here's what those same engagement metrics (shares, comments, and likes) look like for my channel (again, not for the video itself):
Apparently people aren't going to be bothered to share the actual video instead of sharing the FB post: That FB post (shared 193 times) only counts as 8 shares in YouTube (because embedding adds an extra level of complexity to analytics). Overall, 114 of those counted 149 shares were via FB.

Comments:

Likes (the Favorites metric also looks similar, just at a smaller scale - there were only 702 "favorites"):

An additional engagement metric that isn't present on Facebook, audience retention, also looks pretty similar (though not nearly as dramatic): 

If you look in the right hand column (average percentage viewed) you'll notice that for the top 3 videos, all montages, perform a little better than the other content. The one exception, the video at the bottom of the list, stands out because it's an attempt to make engaging content: transgender humor. Overall, I clearly haven't improved at creating content - so at least that makes it more comparable. 

What's interesting about engagement (when compared to total views) is that now there's really only one spike over the course of the channel: January 25, 2013 - the Wipeout Transphobia Facebook post. It suggests that hitting a relevant audience (especially a topically relevant one) matters more than just about anything else that you do to try to capture traffic for the video. This should seem pretty common sensical to everyone, whether or not you've ever done SEO or marketing.

However, the October spike does reappear pretty glaringly when we look at the final engagement metric, subscribers (it also made an appearance in the less impressive "Shares" category too):

October 22 was actually slightly more successful than January 25 (they respectively generated 57 and 55 subscribers). Obviously getting the right audience mattered for the second spike, because that's one share on one site - and it performed really well. The first spike (when I only had 10 videos over two months) shows that getting shared organically is still really important (obviously). The thing that is interesting to me (because it is unknown) is why my video went viral two months after it was posted. What happened? Anecdotally, I have shared some content myself on reddit, but when it has appeared anywhere without me it has done demonstrably better.

####

But there's more to this than engagement! Where did all those views come from? So as for traffic sources:

 
This data is pretty easy to interpret, but I should note that because of licensing rights to the music included in my montages, the mobile view counts are probably way down from what they could've been because they're restricted (and the audio can't be stripped after a certain number of views, which these have all surpassed).

The suggested video metric (suggested alongside/after a specific video) has been my main focal point, because it's clearly incredibly important to traffic. And it's some sort of algorithm that I can't figure out yet - probably because I can't even get people to watch longer yet either ;) As for the other traffic sources:
  • I optimized keywords (and continue to try to optimize) for YouTube Search, but the correlation between search and engagement seems spotty to discern. Capturing someone searching for trans related terms should mean higher engagement - but there's a reason search is its own industry and intent is the billion dollar question. 
  • YouTube - other features is mostly (76%) my own video annotations (such as links to other videos, back to my channel homepage, asking people to subscribe) driving traffic around my channel (which seems better than letting them leave).  
  • Recommended video shows up again as the biggest piece of Homepage feeds and subscriptions (37%).
  • The YouTube channel traffic is mostly from my own channel (90%), but I haven't done a lot of work experimenting with optimizing the channel pages themselves (beyond trying to keep viewers within my own videos - which is where I've gone back to try to create more playlists).
  • The other data points are more difficult to break down at the channel level for various reasons. 
The second to last point is around the value of subscribers themselves. YouTube insists they're important and should be one of a creator's focal points, because subscribers should generally engage with your content more than everyone else. This in turn should presumably give it a higher score in the recommended video algorithm (to what extent, I have no idea). That said, subscribers themselves don't appear to watch this content a lot as a group. I imagine that varies a lot depending on the channel and the audience so I'd be very interested to learn more about that. For my channel, the YouTube Analytics gives subscribers credit for around 30k views (but this is tied to them being logged into Google). It also doesn't consider whether or not those subscribers shared my content and/or helped it go viral - so the point is, subscribers' value is hard to measure, but they are clearly important. 

The final data I'll leave you with is the available demographics of my audience (location and age):

The country breakouts are unsurprising, but the gender/age variation is interesting. While a lot of people's first reaction to the fact that my audience is predominantly young girls and older men suggests the latter are probably creepy chasers - I'm going to assume that a good chunk of those older viewers are either closeted trans women themselves or simply late transitioners who haven't updated their YouTube/Google accounts yet.

And yes, I am secretly hoping that this post goes viral and I can collect all sorts of data on it ;) 

2 comments:

Anonymous said...

I'm one of the elder males that frequents your videos and I can assure you that I'm not being creepy.

I suspect it may be more that at our age, especially post 40, many of us are re-evaluating our lives - what we have/have not done - are we happy - do we struggle with what we have become by the decisions that we have/have not made and should we have done things differently.

Speaking from a personal point of view, I am using this information to help look at my life and decide how I want to proceed.

One thing I find interesting is how many women in transition say 'I always knew I was a girl'. Well I've been on this earth 50 years and I don't know what it feels like to be a girl, or a boy for that matter. What makes a boy, what makes a girl. Does it really matter as long as in the end the role you pick for yourself makes you content, or even if you are lucky, happy?

I wish you all the best in your future life and look forward to more of your videos

buy youtube subscribers said...

The more Subscribers (Buy YouTube Subscribers) you have, the more likely it is that new viewers are going to stick around and subscribe to your channel as well.