Back in May 2014, some of the team from SMG headed down to SMX London; arguably the premier SEO conference in the UK calendar. As a forward thinking search marketing company, we like to keep abreast of the latest tools provided by Google. The logic being that if they are willing to share the data, then it’s best to take advantage of it.
With this in mind, we were hotly anticipating the Universal Analytics (UA) seminar. While it was well put together (as all the presentations were) and offered interesting anecdotal observations on using UA, it left us wanting more. Therefore the aim of this piece is to create a more coherent description of Google’s Universal Analytics, for both those who use it and those that don’t…yet.
So the obvious first question is…
What is Google Universal Analytics?
Launched in October 2013, UA has been out of beta (in the public arena) since April 2014, yet to many it’s at best a bit of a mystery; more commonly it’s unheard of altogether.
Google describes UA as:
“Universal Analytics introduces a set of features that change the way data is collected and organised in your Google Analytics account, so that you can get a better understanding of how visitors interact with your online content.” https://support.google.com/analytics/answer/2790010?hl=en
So far so clear… right!
In a broad sense UA is Google’s response to multi-platform browsing which is here to stay but is immensely difficult to track in a meaningful manner. For an excellent piece on the topic of Attribution and Attribution Modelling, visit the blog of Avinash Kaushik, however we are here to discuss Universal Analytics as whole. Tracking onpage and offpage metrics in analytics has long been hit and miss, the main methods being to create specific landing pages/urls (for TV adverts use www.tvad.com, which then redirects to www.thingswewanttosell.co.uk) and bespoke tagging parameters within urls (eshots or general email footers). As we have moved into the second (if not third or fourth) screen era, even the challenge of tracking online activity has become harder.
Previously, users typically used a desktop to browse the web and buy online. This device was relatively fixed and identifiable (via browser cookies) as a user (in a simplistic world where people don’t misuse computers at work for personal use). As smart phones and tablets have become more popular, this has begun to change. Now people can be sat watching the TV and searching for products /information without the hassle of having to boot up a desktop. Then as the search progresses, perhaps it becomes too fiddly to complete on a handheld, so they opt to move their search to the desktop. In the Asynchronous version of Analytics, these two visits are classed as separate visitors and no connection can be drawn between them. The initial visitor with the true source deemed none converting, while the second (desktop) visitor is deemed as “direct” and takes the conversion. Universal Analytics is paving the way to unifying a user identity over platforms (including the high street), as long as an identifiable action is taken i.e a logins on websites, apps or store cards.
When talking about the ramifications of being able to add offline sales and visit data to analytics data, you may be getting very excited and you have every right to be. The inclusion of the Measurement Protocol is truly innovative, in that it allows data from none traditional data to be sent to Google’s Severs and included within your dataset. As noted before, identifiable data is the Holy Grail in this respect, but in reality this is somewhat hampered by privacy laws.
The perfect scenario would be to create a system whereby you entice an offline visitor to sign into an app when they visit your offline shop, this app could stream data such as their position in the shop or you could encourage them to scan items for more data, which would give you insight into their browsing habits. If they then decide to buy, you scan their unique QR code (on the app) into your internet ready till and as such the transaction is recognised as the same visitor. Later that week, they user decides that they want to buy another product but they can’t make it to the offline shop, so they use their desktop to make a purchase, ensuring that they login to their account (achieved by a loyalty program offering discounts etc) and UA compiles their online data alongside their offline actions.
So far, so fantastical right? As I said this is a perfect example and in reality only barely plausible for large companies. While there are a number of solutions for smaller businesses (i.e asking people to sign up to a newsletter via tablet in store) to gather data, affordable and meaningful solutions for SMEs that allow offline/online user id formation are still a step or two away.
The key ideology is a change from simply tracking visits to tracking visitors. As such visitors are given user ids, which is how they are tracked across platforms. Another benefit of user ids is that you can use other sources (such as your CRM, which has been populated with freely given information) and utilise this within your UA reports. This said no personally identifiable data can be sent to Google, ruling out connecting offline sales to online activity via email addresses etc. You can however, use broader demographic data such as age, gender or location. It is worth mentioning that you have to set up the user id functionality and this best undertaken by a developer (according to Google), another step away from full functionality for smaller users.
If you’re reading this with your fingers crossed, hoping that the next item is going to be the reintegration of key words from organic sources…you’re going to be disappointed! Sorry, that was harsh, seriously though that ship has sailed.The way forward is to utilise how people are using your site and enhancing their experience based on this data. To this end, UA features Custom Dimensions, the heir apparent to the custom variables throne, taking customisation of your data into a new era. You can now simply slice and dice your data in a multitude of ways which would have been difficult or impossible in the previous version. You also get 20 Custom Dimensions, each with 20 associated Custom Metrics, with far outstrips the 5 Custom Variables in a profile on a standard account. With Custom Dimensions you can set up advanced reports such data collated exclusively from first time users, allowing you to view content usage and traffic flow trends with fresh insight.
A dimension relates to a single attribute of your visitor and metrics are typically numeric data relating to a dimension. So a very simple example would be if you want to know which gender spends more time on your site, the dimension is gender and the metrics would be page visits and avg time on site. One really great example of how custom dimensions can be put to great use can be found on the Simo Ahava blog, where he describes how he created a report based on weather conditions in relation to conversions.
Another feature that is beneficial for all is the ability to make amendments to the primary configuration, without the need to amend the tracking code. With the UA interface you can take steps such as changing the standard session lengths. Previously, session duration was 30 minutes, meaning that if for example a visitor came to your site and settled on a product but then spent 30 minutes of inactivity on your site, perhaps making comparisons on competitor sites, the session would end and have no conversion associated with it. Now you can tailor session length to match your site offering.
You can also opt to exclude certain referrers, for example the ever present but nonsensical Semalt.com or perhaps your own blog? This is where you will find the comparable ability to the previous analytics, wherein you can identify your own organic results; the most common example is perhaps Yell.com. You can even take the option of excluding specific keywords as being counted as organic traffic. Perhaps you think that all uses of your brand name would be better defined as direct.
Should I make the change?
Google is being very clear that every account will have to move to UA and they are being relatively open about this process, with all the phases clearly defined here, at the time of writing we were in stage 3 of the process, wherein Google is working through properties and automatically upgrading them to Universal Analytics and all features are supported. This upgrade in short means that the property is ready to handle UA; however, you need to upgrade your tracking to take advantage of the new features. If you don’t upgrade the tracking, your account will continue to gather data using the old methodology for two years, after which it will simply stop.
By stage 4 all properties will have been auto updated by Google and the 2 year clock will start ticking. The official Google Phase page is less than forthcoming about specific dates for this but with the rapidity of the moves from phase 1 to 3 (October 2013 to June 2014) it’s a safe bet that it will be sooner rather than later. Keep checking the phase’s page (linked above) or visit the Analytics blog for the latest news.
So the real question is not should I make the change, it’s when do I make the change. Just to be clear Universal Analytics is set to be a fantastic tool, with numerous features. However, there are a few main aspects to consider prior to upgrading. To upgrade to UA the webmaster will have to amend all the tracking (including an upgrade to any ecommerce tracking) throughout your entire website, which is a time and therefore budget consuming task. So it may be tempting to leave the upgrade as late as possible. While this strategy may be prudent for cash flow forecasting or if you have scheduled an update/rebuild of your website (at which point you can make the updates as part of the larger works), there really is no reason to tarry over the update.
This could be a great opportunity to tidy up the tracking code on your site and perhaps install a Tag Manager such as the Google version available, which allows future tags (tracking codes for various elements i.e. analytics, Webmaster Tools or Adwords remarketing) to be added to your site without accessing the actual site code. However, you will still need to install the tag manager code onto every page of your site and as such in the first instance there is no time saving. The benefit lies in streamlining the process for future tag developments.
For many in the earlier phases there was a fear that data would be lost in the move to Universal Analytics and this is reflected in a number of articles still available online. This fear was drawn from the previous phase lacking the capability to deal with:
Google Analytics Demographics and Interests Reports
Remarketing
Google Display Network Impression Reporting
DoubleClick Campaign Manager Reporting
dc.js Javascript library
In a fashion comparable to the Asynchronous version of Analytics but since the progression to phase 3, new data will be processed and reported. As such, please check the date of other articles published on the subject. However, there is still an element of truth for some in this fear. If you have a fully customised analytics profile which includes the use of custom variables (the majority of SME websites won’t, if you are unsure ask your Webmaster).
As the tracking element of UA has been significantly pared down (to a single cookie), there are some personalised elements (notably custom variables) that would have previously been implemented at tracking stage and are now incorporated into the main interface of UA . This means that any personalisation of this nature will not be automatically carried forward and will need to be replicated within Analytics, which may potentially cause a lag or if an error is made, false data in the profile. This fear is propagated by the two step nature of the transition; there is a tangible state within the procedure when your account is neither fully Asynchronous nor Universal. So you can set up the new custom dimensions to replicate your previous custom variables as soon as the first step is complete, but data to test them will not feed through for up to 48 hours after the new tracking is installed.
A solution to the risk of losing data in the transfer can be mitigated by leaving the original tracking on your site, and creating a new property on your site as UA. This will allow you to master the correct setup for your custom dimensions, however as a new property, there will be no historic data available. Leaving you with the choice of running both until Google pulls the plug on the previous data source and gathering historic data on the new UA profiles in the interim (in which case getting started as soon as possible would be the best plan) or you could implement your newly tested dimensions on your current profile and upgrade that way, but this still leaves an element of risk.
On the topic of utilising properties and profiles within analytics, as UA has a host of new features, using it to make comparisons retrospectively will produce skewed data, therefore you may opt to keep a profile that does not feature any of the personalised functions (i.e extended session timeouts) that UA offers (to be a stickler please (please, please) always keep a clean profile of raw data in all your properties to avoid accidents), this will allow you to retain a set of more comparative data (at least for the initial crossover period).
There are a number of great articles already written about how to make the move to UA and as such we will not replicate that here; if you want to find out more check out this Kissmetrics article the official Google version or call Search Marketing Group for more details.
Universal Analytics is unequivocally the future for Google analytical tracking and as such we recommend that you make the move to UA, as soon as plausible for your business rather than wait for your data to be turned off.
The aim of this document was to form a better description of Universal Analytics for those that already use it and those that don’t. To this end the key focus of UA is to move away from a dataset based on visits to your site and create fuller pictures of activity within your business by focusing more on the visitor as an entity. With the introduction of the measurement protocol, UA is a giant leap nearer that goal. With the development of real world systems that integrate well, the visitor as an entity will revolutionise business analytics. In the interim, there are a number of features that offer more customisation than the previous versions have allowed. While the majority have been brought into the reach of users via the interface, the initial setup of many are still within the sphere of requiring technical developers.