Misc
A walk through my Wistia video analytics for the Daniel Brusilovsky interview
on Feb 12, 2010 - 10:06 AM PSTThis is part of the behind-the-scenes section of Mixergy, which I call etc.
People who saw an overview of my video data, wanted to see more, so I thought it’d be helpful to look at data for a specific video. Here’s my graph for the Daniel Brusilovsky interview.
A few warnings
- Wistia and I have a business relationship. You’ll see their ads on my site soon. And I got my Wistia account for free.
- My analysis below of the Wistia data is just based on hunches right now. What I’d like to do next is check some of these hunches with A/B tests. If you have a suggestion for what I should use to A/B test my video embeds, let me know.
- Wistia is NOT running analytics on my viewership via RSS or iTunes or MP3 or iPhone or anything except straight-up hits on their player.
Observation #1: There’s a natural drop-off upfront
I always see a drop off in the first few minutes of my videos. I thought it was just caused by ads. What’s interesting about this interview is that I didn’t get a chance to record an ad and there’s still a noticeable drop-off.
Observation #2: People are telling me what’s most important to them
I’ve noticed that I’m getting a good response for my clip posts. Here’s an example:
2,881 view for my full interview with Laura
1,947 views for just a CLIP from Laura’s interview
You can see how a little bit of editing grows my traffic with much less work. The problem is that I can’t always find that clip that people want to see. Data like this makes it easier to find what my audience wants.
Observation #3: 20% of my viewers will watch all the way to the end
This 20% is pretty consistent, regardless of the length of the interview. My interview with Anand of AnandTech, for example was 70 minutes long. 20% watched it till the end. My interview with Tim Sykes was a whopping 80 minutes long, 20% watched it till the end.
I thought that because this interview was shorter and more newsworthy, I’d get a higher retention rate. Nope. 20%.
There are a couple of outliers. Paul Graham’s interview, for example, retained over 25% of the audience. I’m thinking of doing this kind of walk-through for his video too. Let me know what you think.
What do you think?
- Was this post useful to you? Should I do more like this?
- What other data should I look at?
- I want to add a Wistia ad to Mixergy. How can I make it useful?
View Comments to “A walk through my Wistia video analytics for the Daniel Brusilovsky interview”
Leave a Reply
You must be logged in to post a comment.
February 12th, 2010 at 11:04 am
On how you can make the ad useful:
Try something like chartbeat.com did. Instead of just talking about how great their product is, they linked to the dashboard for avc.com, so you can see real results for a real website.
February 12th, 2010 at 11:11 am
I think A/B with/without ads is one place to start, although I've never minded the ads. I'm not sure I'd get rid of them if it resulted in marginally better view numbers.
A question about the mechanics of Wistia: If a person rewinds, does that reduce the count at the point where they rewind, and then up the count when they get to where they're going?
Edit: By the way, I think this is really interesting.
February 12th, 2010 at 11:42 am
I'm always listening the interviews while I'm at the gym, and I'm listening them 100%. :)
February 12th, 2010 at 11:45 am
To answer your question about the mechanics (i.e. what is counted), Wistia creates a video heatmap for every view of the video. The video heatmaps keep track of how many times each individual watches each part of the video. To see an interactive demo of the video heatmaps, see here: http://wistia.com/product/tracking
The graph that Andrew has been showing is what we call the Audience Engagement Graph (AEG). Basically, what this graph does is take all of the video heatmaps from all the individual views and adds them all together. So, if a significant fraction of the audience rewinds to watch a particular part of the video, you see a bump in the AEG from the steady state (a good thing).
Happy to go into more detail if you have more questions!
February 12th, 2010 at 11:57 am
Okay. Chris is seeing if he could do that.
My concern with that is that it'll feel l like I'm tossing a bunch of numbers at you without context, but it's worth a try.
Thanks!
February 12th, 2010 at 2:47 pm
Yeah, I figure that audio and other offline participation is different.
February 14th, 2010 at 1:46 pm
Ben, heatmaps=cool. Keep it up!
February 14th, 2010 at 1:54 pm
I love Mixergy!
March 30th, 2010 at 11:55 am
Hey Andrew,
That short clips are getting a lot of attention is interesting and seems logical. I'm imagining your entire video library snipped up into bite-size clips around specific questions. Like a video answer site for entrepreneurs.
About ads, don't you think there's a huge market emerging for low-cost clever video ad campaigns? When ads are funny, stunning, beautiful or provide real insights, people actually look forward to seeing them.
P.S. Did you happen to get any numbers on your audio downloads? I'm audio-only for all my podcasts so am naturally biased, but I'd be very curious to see if there is a consistent ratio between video and audio downloads.
March 30th, 2010 at 6:55 pm
Hey Andrew,
That short clips are getting a lot of attention is interesting and seems logical. I'm imagining your entire video library snipped up into bite-size clips around specific questions. Like a video answer site for entrepreneurs.
About ads, don't you think there's a huge market emerging for low-cost clever video ad campaigns? When ads are funny, stunning, beautiful or provide real insights, people actually look forward to seeing them.
P.S. Did you happen to get any numbers on your audio downloads? I'm audio-only for all my podcasts so am naturally biased, but I'd be very curious to see if there is a consistent ratio between video and audio downloads.