Playing with the Digital Dare Data
And – as I said – I did it to learn stuff.
The truth is that I didn’t really know what I was going to learn from it. But I knew something would happen.
And it did.
Yesterday, the film took off thanks to Rob Manuel and his lovely B3TA newsletter. These stats are from an hour before the visitor explosion. (Unfortunately, the stats since then can’t be trusted because the counter on the first film didn’t register visitors at all yesterday.)
So what have I learned from all of this?
First of all, I found out the complexities of the maths. To do a four step interactive film, it required 29 separate videos. Here’s the structure:
I also learned quite a bit of technical stuff. YouTube doesn’t like most video formats – so I had to experiment to find an HD one that works. (If you’re interested, I used an H.264 in an MP4 wrapper, with dimensions of 1280 x 720 and the audio encoded in AAC format. Yawn.)
Once it was up I started to get lots of messages from people saying they enjoyed it. That’s always nice. And then, because one of the dares is to contact an old friend, I got a couple of awesome emails from people choosing me as the old friend they haven’t spoken to for a while. That was even nicer. (I also suspect that I may have been one of the people that’s been unfriended on Facebook! Awesomes!)
Most importantly, the comments showed that people like my beard. That makes everything worthwhile.
Very early on (in first hour, in fact) I started to spot patterns in people’s choices. I could see in the stats for each of the colours that red – the colour that I thought would be chosen the most – was in fact chosen the least. Here it is in comparison to the others (these are the stats 48 hours after launch when the first video had received just over 500 views).
So I decided to then have a look at the numbers. Here are the stats for which number was picked for each of the three levels:
As you can see, the first and second graphs are the most dramatic. But I wanted to see if there was a pattern in the numbers chosen, so I added them all together to get this:
We can see that we’ve got some statistically significant variations. The number 6 has almost half the clicks of the number 5. It’s all very interesting but it is still just information rather than insight. I wanted to understand why these anomalies exist.
So I had a look at the positions of the choices on the chatterbox. To make it easy to explain, I mapped it like this:
And then I jiggled the numbers about to get an idea of how popular each quadrant was for each number. This is the result:
Ehm. Wow. Quadrant C is significantly lower than the other three. The rest are all pretty much the same. Let’s compare that to the choices of colours:
Yup. There we have it. They correlate perfectly. For some reason the bottom left quadrant is chosen a lot less than the other three. My guess is that it’s because in the west we read from top left to bottom right. That’s one of the reasons why you generally find a company’s logo at the bottom right of an advert – a kind of full stop for the piece of communication. I’d be interested to see if the results would be different in China where they read in the opposite direction.
Then I wanted to see something else.
When you ask people to go through a process online, you naturally get a drop off with each stage of the process. So I added up all the hits for each level to see if that was true. And, as you can see, it’s not:
You would expect the columns to decrease in size towards the right. But instead, they decrease to stage three and then grow again. Why?
What seems to be happening is that there’s a big drop off from the intro video to stage two for two primary reasons. The first is that the YouTube interactivity doesn’t work on all devices. If you visit it on an iPhone or iPad you can’t go further than the first film. The second reason is that viewers find me physically repellent and abhorrently irritating and don’t want to see any more of my hairy face.
The pick-up towards is the end is probably people wanting to see what the other options are – going back a page or two and exploring the alternatives.
So these are my first discoveries. Even more stuff is coming to light. Maybe I’ll share that too if I find it interesting enough. If you fancy playing with the data, just say and I’ll send you anything you need. Or if you’ve got any other explanations or insights, please share them in the comments.
And I’m already working on the next two YouTube experiments. You’ll find them on my channel very soon.