There’s been a lot of buzz around Google’s recent announcement of the upcoming ‘Caffeine’ update to its search engine. Some coverage has been insightful of Google’s move towards the real-time web, and of course some of it has been woefully misinformed (as noted elsewhere).

However, almost nobody (bar a lone commenter on Matt Cutts’ blog) has seemed to mention the fact that Google Caffeine appears to serve AJAX results by default, at least in Firefox (oddly not Chrome or IE7 though).

Google has played around with AJAX results before, to some criticism over the potential effect on web analytics of removing keyword data from the referrer string. However it appears Google has done some work on this, and every AJAX result in the Caffeine SERPs is redirected through a URL that sends the required referrer information, as shown below:

This referrer string was talked about on the Google Analytics blog back in April, notable for including the &cd= parameter, which tallies exactly with the ranking of the result clicked on.

So if this is rolled out along with Caffeine’s other changes to Google’s foundations, could this spell the beginning of the end of automated rank checking? I suspect it probably will in the long run.

It’s a logical direction for Google to go in, as it’s one of the longest standing issues with the accuracy of their search data. If Google could stop rank checking bots tomorrow, I doubt they would, because from an advertising perspective, the volume of their impressions for competitive terms could drop dramatically, and advertisers may start spending less on AdWords. For this reason, a soft-launch that slowly begins to disrupt rank checkers makes sense.

However, with Google’s increasing focus on conversions and CPA models, the volume of impressions becomes less important while at the same time the accuracy of their historical search data becomes far more important. If Google can show stronger correlations between impressions, CTR and ultimately conversion rates, they can legitimately charge more for their ads, and advertisers will be happy to spend more as their profit margin increases.

So where does this leave SEOs in a world without rank checkers? Well, most experienced SEOs have been banging on about conversions for years, and SEO’s close correlation with usability, accessibility, information architecture and ultimately findability makes us pretty well placed in a world without rank checking. I suspect Google will add after the fact ranking data to Google Analytics, and there’s already some fantastic data available through Webmaster Tools. I think as an industry we need to start working towards claiming the field of conversion optimisation for ourselves, before others such as the paid search lot, or other online marketing departments with less expertise in this area start claiming it for themselves.

Just came across a very interesting Google search result which used breadcrumb data in the URL area – shown in the screenshot below. Very interesting implications beyond just the display here – it proves that Google is reading and understanding breadcrumb data. Weirdly in both examples below, the page title structure reflects the breadcrumb structure quite closely.

If you’re not already using breadcrumbs in your website architecture, now’s the time to start adding them…

A post on SearchEngineWatch yesterday asked the question “Are PPC Ads Now Counting in Google Organic Backlinks?” A question I was tempted to reject out of hand as tosh (this would be incredibly damaging for Google if this was really true), but after reading through the post I couldn’t discount it entirely without some research of my own.

This issue is more interesting than I initially thought. Basically what is being said in the article is that for one of the author’s client websites, a number of backlinks were appearing in Webmaster Tools that appeared to show links coming from an AdWords/YSM campaign. Not only this, but the website ranks #1 in Google for a phrase only used in the PPC campaign, and which is not in the content of the page that ranks for the phrase:

So after a bit  digging in Yahoo (search for ["ga nc tn & tx land sale" bluegreen] in Yahoo, and see screengrab below), I can see what appears to be happening, is that these particular ads have been syndicated across a bunch of the spammy pseudo search engines you see so many of these days.

In this low-quality syndication network, the JavaScript code they’re using in these sponsored links is shown below:

<a href="[[adurl]]" ONMOUSEOVER="window.status='http://www.BluegreenCommunities.com';return true;">anchor text</a>

I counted only 2 302 redirects on the ad URL redirection chain before the spider hits a robots.txt block. This means Google may have either given up following these redirects before it reaches the robots.txt file, hit the robots file and decided to try another route, or was treating the JavaScript element as a different link entirely.

Now the clue that Google is following the JavaScript code element in this case is that it is the domain, and not the ad destination URL that gets the anchor text credit, accounting for the homepage ranking #1 for the query in question, and not the advertising URL.

This is rather odd, because “window.status” is not a JavaScript link function, and implies Google’s JavaScript crawling isn’t as advanced as they want us to think. Looks to me like they’re just running a regex for anything that looks like a URL inside a onmouseover/onclick/etc command – this requires testing, but if that’s it, surely that’s pretty primitive?

I think it’s fairly clear this is a screw up in Google’s JavaScript crawl implementation, so will likely be fixed in (very) short order. In the meantime, I’m thinking of setting up an AdWords campaign or two…