Image search - Google vs Yahoo vs Live Search - a little case study

We recently re-created the Nutch logos and thought we'd keep an eye on the search results in both web and image results.

It's not something that is searched on much and there aren't many competing pages so you'd expect this site to rank ok for that term on all search engines.

Google

Google, as ever, picked up on it pretty quickly - as of 11th March, our blog post was 1st for "nutch logo" on web search and 2nd on image search. The only problem was that despite good and correct use of anchor text, Google images shows the Inter Advertising logo as the Nutch logo. It's a bit strange 'cos the anchor text and alt tags are pretty accurate so I don't really
understand what's gone on there.

Yahoo!

I couldn't find the post at all on Yahoo but I did find a page from CNN that didn't mention Nutch but did have the word "logo" on the page. Once.

Further investigation revealed that Yahoo hadn't indexed any of our blog posts since the beginning of Feb. I hope that their Hadoop platform helps them to actually start to crawl and index the web a little better. My hopes for Live Search weren't high...

Live Search

Despite the fact that Live returned 3 images of a dog, a bird and a wooden box (a hutch?) our blog, not the actual post but the blog homepage, appeared 4th.

Bizarrely, the page with the logos on isn't even indexed but it's obviously got the first few lines of the post from the blog homepage.

Amazingly we were first and second in the Live image search but on www2.interadvertising.co.uk/blog/1.

The www2 subdomain was used long ago and has no links - the www has plenty of links. Google has effectively written off www2 (quite rightly) but Live Search has chosen to go with the weakest subdomain.

Also, blog/1 is not linked to externally or internally so who knows how they've worked that out.

So all in all, pretty much as you'd expect. Google did ok, Yahoo was a joke and Live Search is as mad as a fish.

I can honestly say, I've got better, more relevant and less spammy results from a Nutch sandbox that took me 20 minutes to install the Live Search has been able to despite 4+ years and millions of dollars. How do they get it so wrong....?