How many heads does a hydra have? How to track unique visitors in Web 2.0 Analytics

 In Web Analytics

Paradise Analytics
In the ideal world, web analytics tools are able to track unique visitors without occurring in data accuracy doubts. In this world, every customer has one site with a unique domain (www.customersite.com) and the tracking tool is based on first party cookies, so that it can bypass the browsers’ security settings. In this scenario, the cookies used to track the unique visitors are written from the same domain that is tracked (with Omniture SiteCatalyst, first party cookies tracking must be enabled; Google Analytics and WebTrends use by default first party cookies and that’s a great point of advantage of these tools). In this ideal world, the only elements that could affect / inflate the counting of unique visitors are the cookies deletion (by the users or by cleaning tools) and the usage of different browsers, different PCs or different devices by the same visitor. Of course, I’m assuming that in this world nobody disables cookies and Javascript!

The dark side of the moon
Anyway, even in this ideal world there would be a critical point about the count of unique visitors related to the time range of the analysis. Consider how the unique visitors metric can change its meaning with different time period of analysis: for one specific day, I would know the number of my daily unique visitors, but if I select one week, I could see both my weekly unique visitors and the sum of the daily unique visitors, as well as I select one month, I could see the monthly, the sum of the weekly, or the sum of daily unique visitors. The point here is the scope of the analysis. Another weak point comes out when it is needed to analyze a custom period. How to be sure about the number of unique visitors for example from Monday to Wednesday? Google Analytics is the only tool that is able to count the unique visitors for custom periods, while WebTrends and Omniture SiteCatalyst are not able to deduplicate that value (Omniture provides at least the Data Warehouse feature for the SiteCatalyst users that can extract that value, but the answer to the query is not given in real time).
Anyone would love that ideal world.

The hard truth
By the way, unluckily the real world is more something like this: one site with several domains (e.g. one domain for country, www.mysite.com, www.mysite.it, www.mysite.es) and several sub-domains (e.g. minisites, minisite.mysite.com). The biggest issue in this scenario is the impossibility to activate the first party cookies tracking (every site would write his own cookie, avoiding the accurate count of unique visitors crossing from mysite.com to mysite.it). How can the tools manage this scenario? Google Analytics requires changes in the page code (_setDomainName() and _link() functions) in order to deduplicate visits and visitors. Also, if you want to track two groups of sub-domains of the same domain (e.g. aaa.mysite.it and bbb.mysite.it in a profile, ccc.mysite.it and ddd.mysite.it in another profile), you can’t simply use _setDomainName(“.mysite.it”) for both profiles, because in this way they’ll share the same Google Analytics cookies, showing strange results on reports. There are two solutions to this problem: using the _setDomainName(“none”) function in the tracking code and the _link() function in every link that leads from a sub-domain to one of the other sub-domains tracked in the same profile; or modify the ga.js file (hosting it locally) in order to have sets of cookies with different names (to avoid cookie sharing). Omniture SiteCatalyst gives two chances: the first requires to activate the cross third party cookies (2o7.net), the second is to use the first party cookies only on the main site (e.g. www.mysite.com) and for the other sites write the same cookie but as a third party cookie. This last solution is working fine on our customers.

The worst case of the real world scenario is when a unique visitor browses mysite.com and mysite.it during the same session using the PC on the morning, his mobile device on the afternoon and the iPad on the evening! This world can be more complex than what we can think!

In this complex scenario, the Marketing Department requires to differentiate always more the online experience and to embrace the growing need for online interactions (PC, mobile, iPad). These requirements can be translated into the need to identify the visitor either cross browser and cross device, to make the web experience more customized.

Which are the possible solutions in this complex world? Some customers are experiencing the Local Shared Objects (LSO) commonly called flash cookies, in order to track the visitor cross browser. The flash cookie tracking is based on the concept that when a browser requests a page, it can transmit information not only about itself, but also about the computer it’s running on. In this way, it would be possible to uniquely identify each computer, regardless of the browser used. The flash cookie is written in your computer, and even if you delete your browser’s cookies, you do not delete your flash cookie, so that if you browse mysite.com using Firefox and then using Internet Explorer, you still are recognized as the same visitor. Anyway there is still an issue about this type of tracking: Flash technology is not supported by the great majority of mobile devices, so the tracking cross device would not be granted.

By the way, another solution is recently coming to the light: taking advantage of the wide spread of Facebook, the visitor that uses PC, iPad and iphone to navigate either mysite.com and mysite.it could be uniquely identified by the connection with his Facebook user’s profile. We are just testing this method right now and the following steps can give you an idea of how it works:

– the visitor lands on the home page of mysite.com;

– a popup appears to the visitor, asking the visitor if he wants to Connect the site with his Facebook account (typically rewarding some discount or special offers);

– if the visitor accepts the connection, a welcome message (a kind of “Welcome Visitor, you are signed in with your Facebook account”) will appear on the site;

– all the most relevant data provided by the Facebook profile, as gender, day of birth, geo-location and so on, are collected in the web analytics tools in conjunction with the visitor’s behavior on the site;

– an example of information that can be reached by this method is that many men from Italy are watching the pants showed in home page but only a very small percentage of them is going to buy these pants. In this case, it could be useful a direct email program (also the email address is provided by the Facebook integration) toward the italian men, showing a very advantageous discount to buy the pants. Pretty cool isn’t it?

This integration offers the opportunity to segment your offer based on the visitors’ behaviour on the site and their own Facebook information (social-demographic data, thoughts, and so on). The Facebook visitor ID is the only keystone that ensures that the cross browser and cross device tracking works right.
As you can imagine, the Facebook solution is in conflict with the user’s privacy, and this is why this solution is the most talked-about. There’s a trade off between the greatest accuracy to identify the unique visitors and the privacy that seems to be not granted. If this scenario will take place, the data will not be aggregated anymore, but they would be tied to the single specific user: the web analytics world is switching from an anonymous data collection to a one-to-one data collection.

Where is the web analytics world going to?
The future is to connect the online experience with the offline actions, in order to link the online and the offline customer. A primary key must be identified to tie the two worlds. When it will be found, the golden age will come!

Recent Posts

Leave a Comment