Video is the new darling of the web, and there are some lofty claims made of it too, the most notably inaccurate of which is that it provides SEO benefits. It's not the video itself that provides SEO benefit, but hosting it on YouTube (Google), and providing a load of metadata around it (text, which the likes of Google can read).
However, this post isn't about extolling the virtues or otherwise of using video on your website. It's more an example of when not to use it.
We were recently asked by a reasonably well-known chocolate brand to create a mobile-friendly Facebook web application. These are the things you often see as the tabs on Facebook pages or external sites that ask you to log in with Facebook.
The crux of the brief was to allow the user to choose a photo of themselves and their friends, move it around, resize it, choose characters for each and end up with a video of their face atop the dancing and singing character. The mobile-friendly element was quite specific - the video had to play on mobiles, and there wasn't enough budget to make mobile apps for even one platform (we often have to work to tight budgets, so this isn't necessarily a problem).
So far so good. You're probably wondering where the problem is? Well, the first is that to make a video of a dancing character with a face, we have two options. Flash, or true video. Flash isn't mobile-friendly but would have made the whole thing a lot simpler (for once!). Video on the other hand works everywhere, but it has to be generated on a server, instead of happening on the user's machine.
So the user has to wait. So what? Now here's where the problems arise really. Each character is 3 videos composited on top of each other (mouth, face, body), and there could be up to 6 of them. So that's up to 18 videos, plus the background: 19. That many videos compositing on top of each other with transparent bits to make sure the bits behind show through properly takes quite a long time, and it turns out it's longer than some people are prepared to wait (and goes up the larger format the video is). Of course, the option to email them when the video was ready, or to automatically post it to their social media places, was presented, and eventually, the latter implemented, but only too late after the first round of users got bored and went off on their merry way.
It also turned out that the servers could handle several hundred requests a second from users, but could only generate one video each on average per minute, which meant that we ended up with a huge backlog of videos that took a few hours to clear the first time around, the only solution being to add more servers and to try and reduce the dimensions of the videos used to generate the end result.
While the number of users of the app was staggering, we learnt one key lesson about how best to achieve a similar result in the future - don't use video! Instead, follow the lead of apps like Elf Yourself and make native mobile apps that can generate animations on the fly. They won't be the same as a true video, but they can be played on any platform you build the app for.