SMX London 2008 – Day 2 Session 3
After a typical conference lunch (chicken and cous-cous – hmmmmm, bland!) we head in to session 3, entitled “Search 3.0 – Video search and blended results”. As a big fan of universal search, I’m very interested to see what they have to say.
First up to the podium was Tom Wilde from EveryZing, who talked about video SEO best practise. EveryZing have a number of ‘products’ that apparently help you get in to universal search – a bold claim considering Google themselves regularly remind us that universal search is only a beta test and won’t always appear – but they definitely seem to know their stuff. Anyway… Tom started with a very nice graph from Hitwise which shows the increase in traffic from search to video sites (and the decrease in traffic to these sites from social networks). So it’s clear that search is sending more and more traffic to video sites, but why and how? The biggest and seemingly most obvious issue is that video itself isn’t crawled – it’s still all about the text, i.e. the meta tags, the title, the description and the comments. So if you can get that right – and extract the keywords and relevant tags from the video by hand – you will have a much easier time getting in to the search results.
Next up was Brian Marin from Performics (the SEO agency that was bought by Google a while back). Brian talked a bit about the limitations of video search – again highlighting the issue that Tom talked about. But he then moved on to Google’s new Audio Indexing (GAUDI) beta tool, which uses speech recognition technology to index the content of YouTube videos. As well as making the indexing of video content a lot easier, it will also allow you to navigate straight to the part of a video that mentions your keywords. It isn’t perfect yet though!
Brian then moved on to images – until image recognition is perfected, you’ve still got to remember all the classics for optimising images, i.e. file names, URL, text surrounding the image, ALT / Title attributes, Page title and Inbound link anchor text. He also talked about ways to optimise images that Google may already have indexed – little things like changing the file attributes (file type, size, width etc) can ensure Google won’t count your image as imagery ‘duplicate content’…! Finally he looked at a few of the images you shouldn’t bother optimising – things like navigational elements, background images etc. He closed with a look at some of the future changes that might happen to image search – scene analysis, visual search, geo-tagging and the like. All very interesting stuff.
Ciaran Norris was the final speaker, who spoke in detail about optimising for YouTube in particular. The basics for optimising YouTube is more views = more ratings = more comments = more views. How do you get viewed more though? Well, very similar to how you get your blog posts read – choose a good title (“exclusive leaked footage that you won’t believe”), use descriptions wisely (moonwalking bear), try to get a good thumbnail (if you can manage it), choose the right tags, encourage commenting / ratings, check your incoming links etc. He also touched on a few of the ‘illegal’ or unethical methods you could use to promote a video, but I won’t go in to those here as they’re all very black-hat. He ended with a video that his company (Altogether Digital) have just produced for Transport for London (“look out for cyclists”) which is a great example of good content – I will let you look it up yourself!