I was pinging people on Twitter about what to blog about, and Michael Vandemar suggested I take a look at his post about Google web search switching to AJAX. He and some other SEOs have recently noticed that Google search results URLs have been displaying the search query string after a hash mark (#). I'm not an AJAX expert, but some of our developers told me that AJAX typically offers up a better user experience and you can load parts of a page selectively as opposed to the entire page at the beginning. Thus, it seems that Google switched to AJAX-driven search results to create less bandwidth for them, since every refining query will only load the HTML necessary to replace the search results and not necessarily the entire page.
However, there's a drawback to this, as Michael and some other folks have noted. In URLs, hash marks and anything beyond them aren't passed to the server in the referring URLs, nor are they regarded by search engines (this is why we recommend using them for tracking ids in order to avoid duplicate content problems). This means that analytics won't be able to read or record referring keywords/search terms. Instead of differentiating search terms and phrases through the search query string, chances are your analytics program will count Google's referring search traffic as all coming from a single source, Google.com. Obviously this is a big deal--it's incredibly important to know which keywords and phrases are bringing in traffic to your site's pages. Eliminating that informtaion would be like trying to take your newborn baby home from the hospital, only to discover that he/she is mixed into a room with fifty other babies and none of them are wearing identification bracelets.
Michael blogged about potential breaks with the switchover, and so has the Clicky blog. Brian Clark has already noticed some oddities with his analytics:
I haven't gotten the changeover from the search/? query string to the hash mark--when I perform a Google search, I still see the old query string. However, a hash mark query string does resolve for me, so I'm wondering if this is something that'll become more widespread very soon.
Have any of you noticed the hash mark switchover and its resulting effects on your analytics data? What would be a good workaround to this change if Google decides to fully switch to AJAX? Obviously there has to be some fix or else webmasters, site owners, advertisers and Internet marketers everywhere will be pissed...
Here's what we said to SearchEngineLand:
"We’re continually testing new interfaces and features to enhance the user experience. We are currently experimenting with a javascript enhanced result page because we believe that it may ultimately provide a faster experience for our users. At this time only a small percentage of users will see this experiment. It is not our intention to disrupt referrer tracking, and we are continuing to iterate on this project. For more information on the experiments that we run on Google search, please see: https://googleblog.blogspot.com/2006/04/this-is-test-this-is-only-test.html."
By the way, any word on making the SEOMoz comment box draggable in Chrome?
Matt, thanks for your comment and not letting all of this get into total panic! But I'm almost sure that there are lot of those who like to make a mountain out of a molehill in SEO community.
Thanks for informing us, but this post dated as 4/24/2006 and I'm sure thats only a small percent of users that could connect these two things even if they remember it.
"We’re continually testing new interfaces and features to enhance the user experience.
Great. The house is warm now. Can you put the fire out?
Matt,
I was stunned when I opened the source of a new SERP.
- I know this sounds dumb, but what if google tried to index that page (i.e. another website started doing this same tactics) I couldn't find 1 english word on that page, just a bunch of JS.
- The page isn't semantically correct...
**I just read a post about how a half second delay cost google about 20% in profit (the 30 results vs 10 study) so i guess this could add 20% to googles profit while cementing in G-Analytics as the best analytics package...
- Just seems to me that google is being quite on this (besides your canned statement, that was just to searchengineland and a link to a 2006 article; that doesnt reference this specific test at all)
I second making the Comment Box draggable in Chrome. ;-)
Email [email protected] and we'll open up a case based on your request.
So since Chrome uses Webkit (kudos to that), our little draggable interface doesn't support Webkit and thus not Chrome. We'll probably be making the comment box static for everyone rather than fixing it for Webkit if we do tackle the issue.
use handle option in the dragable and other browsers will work!
here is refrence
https://wiki.github.com/madrobby/scriptaculous/draggable
This switch to save bandwidth (or what ever prompted them to do so) is kind of like setting your house on fire to cut down on heating expenses.
Analogy of the day winner. love it.
If no usefull data is passed on in the referral field there is really nothing analytics providers can do to solve it. If there is no data - there is nothing to extract no matter how you tweak it.
Google could, however, easily send traffic through a script that adds usefull referral data - if they want to.
Or maybe they will only do this for Google Analytics leaving it as the only analytics package left on the planet that can track Google searches. That would truely be stupid.
Why stupid? Shareholders generally demand their company do everything to maximise their profits (some do occassionally display some morals though). Wipe out all analytics competition and then you can charge what you like .. that's capitalism.
Not very friendly. But then shareholders aren't generally in it to be noted for their generosity.
With Yahoo! Analytics on the verge of being released to the public, keeping Google results only trackable by Google, it could very well keep Google Analytics users from switching to Yahoo. While not typically Google's way of doing things, I could see why they might do this. Though if Yahoo were to respond in kind, there'd be a lot of frustrated and angry SEO engineers.
Or, as someone as suggested, browsers could starting feeding data after the # to users
Google could solve the keyword/analytics problem in a non-evil way (ie: so that it works for all analytics packages) by redirecting through a page that carried the keywords. Using thecorrect redirect would then populate the referer string with something useful.
Eg: google.com/#q=cheese
clicks through to: google.com/redirect/?q=cheese&url=cheese.com
redirects to: cheese.com
Alternatively, we can create a JS stats package that looks at the browser history to get the keywords.
Bish bash bosh.
Thumbs up for the use of bish bash bosh.
Luv a duck.
Many times redirects will carry the original page as the referrer, iirc, and currently for security purposes I am pretty sure you cannot directly access the browsers history and read it via Javascript.
Technically, if they wanted to they could track this totally behind the scenes, and populate the Google Analytics data that way. However, if they did go that route then GA would be the only analytics package that could tell you what keywords were sending you traffic from Google.
Is that bad for Google Inc?
Disastrous for many, but bad for Google? Who will stop using Google over this?
1) AJAX search is great. WTG Google (although I still don't see it yet). I'm curious to see if they refresh the ads with each search.
2) Google gains no benefit by making it hard to measure your performance. Also as others mentioned, Analytics needs the data anyway, and Analytics is a key part of their Adwords business. It will get taken care of.
3) Even if they goto AJAX search, they can just route all the URLs through a google.com redirect (like Yahoo BOSS) and set the referrers that way.
Still quite interesting :)
Totally agreed, I cant see how would be in Googles interest not to resolve the issue with GA, specifically if the the search terms are coming form their engine....
The hash mark does resolve for me, and yes the ads are refreshed with each search.
Cool, thanks!
I too have been seeing more and more instances of it in my logs.
Nice post Rebecca and great job Michael Vandemar! On top of things as always :)
I don't think this (this being the view that Google is trying to position GA as the only reasonable choice for site stats, which equals revenue opportunity in the shareholders' minds) would make it through anti-trust laws. I think G investors know better than that.
- The web basically runs life in the U.S. (and is trending positively)
- Search runs the web (i think this is still trending positively...)
- Google owns U.S. search (and is trending positively)
Doesn't seem like any less of a monopoly than Microsoft or AT&T back in the day. Those companies may have had more of a stranglehold on their markets percentage-wise, but they only owned product - not information, which has much bigger repercussions (hey, didn't someone just write about that?). Sorry, can't hoard the referral info and charge for Analytics. Of course, if they kept Analytics free...
I understand your line of thinking, but just because Google is currently in the lead and *headed* towards completely monopolizing search, doesn't mean they currently are. Yahoo still holds a large portion of the demographic, and MSN and Ask.com and others to a lesser extent. If they didn't life would be easier (though possibly more dangerous and frustrating) for SEO engineers. And with the upcoming release of Yahoo analytics, I don't think that anti-trust laws would prevent any such action currently. In the future possibly, if Google continues it's current positive trend, but not yet in my humble opinion.
There is a post about how to use this method to your advantage in analytics here https://tinyurl.com/brotub by Mike Plummer this may be what you are actully seeing and not just Google.
That post has nothing whatsoever with what is being discussed here. That's for using hash marks in landing page url's that you have control over, such as where you want AdWords to go. You simply cannot track the referring url fragment after the hash mark if it isn't passed. It's a browser thing. The browsers themselves would have to be rewritten.
this will maybe teach us to read the Google Blog first which was the first thing I do.
Test or not how do you plan on passing referrer info when browsers dont after the #?
I noticed something interesting. Whenever I search from the Firefox search bar, Google resolves with the usual "/?q=search+term". But if I actually type the query into Google.com, then I get the hash mark.
Anybody else seeing this? Do we think this is a limited rollout because we all use Firefox?
I had to sign out (or not use iGoogle) and I get the hashes now when using FF.
The break in analytics tracking (with the new Ajax search results) is due to the way web browsers handle referrer strings.
So, is Google Chrome capable of handling the referrer stings of Ajax search?
change your user agent to OPERA (via user agent switcher) to get rid of this really really annoying "experiment"
oh yeah, this "experiment" kills of all my greasmonkey scripts i did for google serps
Yeah, I just realized it was blocking the onpage SERP data through SEOquake. Very annoying. Thanks for the recommendation!
It will not be long before there is a fix to the problem and a bug to report in another
Matt Cutts,
It was not my intention to buy 30 items at Wal-Mart tonight when I went in for 2. But I did.
Thus, your comment is not very reassuring.
Will Google not come out and state directly that they WILL find a way to retain referrer info in this AJAX system?
(not vague corporatespeak that really says nothing)
as google redirects all clicks on results via a trackings page it would be easy for google to add the refferer there
Matt, what a drastic way to ruin the dynamics of Rebecca's post! Just when everyone was getting ready to pester about Google and the funny way they experiment with AJAX : )
Seriously, what a buzzkill. ;P
I would assume that the Google Analytics team is aware and on the case. I can't imagine Google releasing the Ajax search across the globe without resolving the search query tracking issue in Analytics. Well, we can only hope anyway.
Totally seeing this in my Analytics reports... Man, I hope Google responds to all the coverage this is getting because once this is rolled out to everyone -lookout.
Google might roll out AJAX search as an optional preference, sort of like web history. I know their looking for ways to improve the user experience, some people might prefer the classic search and some might like the AJAX interface better. I agree, if that happens then GA starts moving as the default analytics package for measuring performance in Google - that is if they don't create an API or something similar for webmasters to retrieve referrer data from.
Well, really the interfaces aren't different at all. The average searcher wouldn't have a preference to search with or wihout AJAX, so if Google wanted to implement this to save bandwidth, they couldn't possibly make it an option.
Perhaps a few nuts with tinfoil hats would opt for it.
Can Google really be experimenting with just a small percentage of users if I'm seeing it here in Florida and you guys are seeing it in Seattle? Over the past week, incoming referrals from Google.com have risen from 1-3 per day to about 60 as of yesterday (our average incoming traffic from Google every day is about 1500). GA tells me that those coming from Google.com are in 29 different US states.
I just hope they don't start giving out APIs to track this now potentially obfuscated data only to stop again (as with the SOAP APIs). Granted, rank monitoring is less important than seeing your referring keywords and phrases, but through the APIs, search marketers were trying to monitor rankings in a way that wouldn't bother the engines too much. Now we have to use those human emulated kinds (like Advanced Web Rankings).
Matt says it's not their intention to disrupt referral tracking, but for the sake of a better user experience for the vast majority of searchers, there's no real saying what they might do.
Google are generally in favour of the greater flow of information. If they go ahead with an AJAX / JS / whatever interface, presumably fixing issues like loss of referrer data will be addressed before any general rollout. To do otherwise would be kind of dumb, and that at least is something I'd never accuse them of
I can see some Google engineer coming out and say. We do no harm. Ajax is cool and fast. We think SEO is not supposed to be a factor anyhow. Monopoly us??? Never. Do no evil. And the mumbling goes on. We are SEOs inspite of Google not because of it.