http://support.google.com/analytics/bin/answer.py?hl=en-GB&answer=2790010

Although still in beta, this is no longer a closed test. Has anyone had a play with it yet? On the surface of it, it looks like this could impact the next level up of commercial analytics vendors, as features like being able to integrate offline data are very high end.

Is this primarily oriented at making it possible to get rid of limitations that exist in the current GA such as sharing data across web and apps?

QuoteWhile it's possible to combine several resources into 1 property if you tag them with the same tracking ID, web and app resources cannot be mixed into the same property. App-tracking data can only appear in designated app properties and web-tracking data can only appear in designated web properties.

http://support.google.com/analytics/bin/answer.py?hl=en-GB&answer=2649554

Or is it more about ultimately being able to inject all sorts of data for which GA doesn't provide any metrics at all, regardless of platform?

QuoteIs this primarily oriented at making it possible to get rid of limitations that exist in the current GA such as sharing data across web and apps? Or is it more about ultimately being able to inject all sorts of data for which GA doesn't provide any metrics at all, regardless of platform?

Both, I believe. I'm only just starting with GA, so I don't know how this compares to the stable version. However, one of the benefits of IBM NetInsight was the ease in which you could extract data for a visitor. i.e.:

Visitor comes to website ->
website asks analytics what we know about them (past visits, pages viewed etc.) ->
website displays content informed by analytics data

I could never do anything like this in Yahoo Web Analytics, as the API only supported aggregate reporting - nothing about individual visits or visitors. Am I right in thinking that Google Analytics was/is more like Yahoo than NetInsight?

yeah GA is aggregated data, we hack it into our own system with media tags to get more specific user/attribution journeys, and are thinking about adding a third layer of user specific (and potential identifiable if they give us the data at any point) to totally join up the dots. It's all a bit painful.

I think we might have a beta of this on one site or another - I'll see what the verdict is - we have the attribution modeling beta on several sites and the results don't look entirely accurate, but it is a beta, I assume it will get better and the same probably applies to GUA.

I've been involved with this since the Alpha test.

It has the potential to be amazing, it's completely in line with the announcement of Enhanced Campaigns, it's all about tracking users, not devices. Online and Offline. Loyalty Cards, Point Of Sale in store, all of this stuff should find it's way into Analytics and they also provide multi-attribution modelling as standard, which was only in the Premium $150k version.

You need to understand how to build in the data layer, but it makes certain things a lot easier.

As it is a beta, some things I don't like about it.

No Doubleclick tracking (useful for retargeting via analytics), no split testing, no adsense tracking.

But I am sure those and a lot more will come.

Somebody set up an appliance being switched on and off, and monitoring the weather, so if you can think it, then chances are Universal Analytics can measure it.

http://www.elisa-dbi.co.uk/blog/tracking-wemo-switch-motion-google-universal-analytics/

http://www.elisa-dbi.co.uk/blog/using-universal-analytics-to-segment-by-weather/

that weather one is awesome!

You've confirmed what I was hoping Jim. This doesn't only hit the analytics vendors, but also the business intelligence apps like Crystal Reports that'd be used to aggregate all of this data. This looks really cool.

The tracking code is also really elegant, and plays well with third-party async loaders. I've stripped out Google's async loader as I'm already using LABjs, and this is all that's needed for basic pageview tracking:

(function(i,r){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){(i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();})(window,'ga');ga('create','UA-XXXXXXXX-X','mysite.com');ga('send','pageview');

Really liking UA so far. Latest bit of late night coding, for creating landing pages, with a jQuery prerequisite:

Load up your page tracking parameters as bastardised JSON in a class attribute (or valid JSON in a HTML5 data- attribute with a bit of recoding:

<script id="labjs" type="text/javascript" src="/js/lab-1.0.js" class="{'set':{'campaignSource':'mysource','campaignName':'myname','campaignMedium':'mymedium'}}"></script>

This replaces the usual ga('create','UA-XXX-X,'mysite.com');ga('send','pageview');:

/*! Universal Analytics */
ga('create','UA-XXX-X,'mysite.com');$(function(){var s=$('#labjs[class]').attr('class');if(typeof(s)!==undefined)try{var o=$.parseJSON(s.split('\'').join('\"'));$.each(o.set,function(k,v){ga('set',k,v)})}catch(e){}finally{ga('send','pageview')}});

which pulls out all of the key/value pairs from the HTML and fires them off to Google, replicating the old utm_source, utm_campaign and utm_medium URL parameters. It still fires off a regular pageview if there's no class, an empty class or invalid JSON. Not fully tested yet, but it works in Firefox 20 and IE6, so it's unlikely to have any worse compatibility than jQuery.

Next step is to do the same with the e-commerce plugin.

Browser compatibility:

Standard inline Google async loader:

Chrome 1+
Firefox 1.5+ (broken in FF1)
IE 5+ (not tested in earlier versions)
Opera 8.54+ (broken in 7.54 and earlier)
Safari 3.23+ (earliest version tested)

LABjs as an async loader:

Same as standard Google async loader.

LABjs + Analytics + jQuery campaign tracking:

Breaks IE5, IE5.5, Opera 8.54, Opera 9.64. Works in Opera 10.63+

- Non-jQuery JavaScript will run after loading in jQuery, but not the campaign tracking. It should be an easy fix to get the basic tracking working again, for the sake of completeness.