Google Analytics may be free and widely used, but you should be aware of its limitations, says Darren Guarnaccia.
Google caused a stir when it launched a free web analytics service in 2005. Since then, the search giant has steadily improved the service and expanded its capabilities to include features such as customised reporting, advanced segmentation and API-based access to its data repository.
Google Analytics offers valuable reporting on overall website traffic and tracking, but suffers from issues common to many of the page tag-based hosted services — it is also a victim of its own success and design.
So, there are several things to be aware of when considering Google Analytics, and it is important to know not only what the service will tell you, but what it cannot.
Browser blocking
The first thing to understand about page-tagging services like Google Analytics is that they rely on three things: JavaScript, first-party cookies and third-party images. Google Analytics depends on being able to trigger an image on its servers to transmit information back, and this technique is increasingly being blocked by browsers by default.
In addition, more browsers and anti-spyware plug-ins are beginning to block analytics cookies from Google, Webtrends, Omniture and others. The net effect is that your reports from Google Analytics could be missing more and more traffic. So, is this statistically significant?
At an aggregate level, 10 to 20 percent is not too much of a problem, but things are changing in the web world. As today's websites shift away from being passive brochureware to active marketing and sales platforms, the devil lies in those very details missed by Google.
Modern marketers and site managers are beginning to delve into the sort of detail that Google cannot provide. It is a question of scale. Google Analytics cannot possibly be expected to record and store comprehensive details of each and every browser session, over many years, for every single site it handles. The sheer volume of data would be unmanageable.
Yet it is that very level of detail that today's web marketers need to give them a competitive advantage. New emerging solutions now allow marketers to record all the intimate nuances of user behaviour on each page of their website. Details are much more in-depth and may include anything from what site content visitors read, their user profile and how they respond to a poll question to what company they are from to how they completed a web form.
That new level of tracking allows companies to go far beyond looking at macro traffic patterns; now they can follow individual visitor sessions, learn what is working, or not, and pass that real-time customer intelligence to sales.
Imagine your website being able to provide your salespeople with information that will help them to win business. Today, Google Analytics cannot give you that, as there is just too much information to handle.
Interpreting behaviour
Beyond obtaining a detailed knowledge of visitors and their interests, marketers also want the ability to interpret visitors' online behaviour accurately and spot telltale signs that indicate they are ready for a sales contact. The practice of rating leads based on their behaviour, known as lead scoring, allows you to prioritise the ones that are most ready to engage.
On the web, this prioritisation is done by assigning values to website content and activity to show a visitor's propensity to buy, or at least engage with sales. Today, Google does not allow you to weight or score your content in this way, or even to prioritise visitors as leads and prospects.
Google Analytics does provide some good, advanced segmentation features, allowing you to understand smaller cross-sections of your visitor base by geography, returning visit or visit duration, for example. Unfortunately, you cannot drill into some of the most important segmentation categories that marketers are used to doing with other media, such as personality type or industry background.
Overall, Google Analytics is an excellent tool for understanding high-level traffic patterns for your website. But the requirements for demand-generation websites are changing the game.
Darren Guarnaccia is vice president of product marketing at content management system firm Sitecore. A regular speaker, panellist and moderator at industry events, Guarnaccia started his career as director of technology for a large financial company and subsequently ran e-commerce operations for a big regional consulting organisation.
Talkback
I found this article very interesting -- in particular the claim that 10-20% of analytics calls are blocked by browsers or anti-spyware plugins.
As mentioned in the article, analytics depends on a combination of javascript, third-party images, and first-party cookies. (Some legacy customers still use third-party cookies, but this is a small number, and GA uses first-party cookies. Not sure about the other analytics packages.) No standard browser turns any of the above three pieces of functionality off by default. (Safari keeps third-party cookies off by default, but we're not talking about third-party cookies here.) And if a user manually turns off any of these pieces of functionality, the web experience effectively dies. As for anti-spyware plugins, it makes sense to me that something out there exists that might block, for example, any __utm* cookies (GA cookies), but could this be 10-20% of all analytics calls? If so, this would be new news (and very interesting news) to me. Mr. Guarnaccia, it'd be great to learn where you got your 10-20% numbers from. Intuitively I'd think that the real number is far, far lower. Periodically *deleting* cookies is another story, as anti-spyware applications (not plug-ins) tend to do this. But we all know about the short shelf-life of cookies. But this does not mean *blocking* of analytics calls -- only detailed lifetime value (or longterm value) analyses. But the wording of the article implies that browsers and plug-ins are in fact blocking the calls. It'd be great to learn more about where these insights come from so we can all adjust our expectations accordingly.
I look forward to more information on this question.
Regards,
Jonathan
This post has been removed by a moderator.
I'm reminded of a song from my fair lady.
One has to wonder when the English standard dictionary went from Oxford to Urban.
Great question Jonathan,
There's lot of good data about the general innacuracy of page-tag based web analytics. A great primer can be found here http://www.advanced-web-metrics.com/docs/accuracy-whitepaper.pdf. Today, the biggest culprits of these error rates come down to Antivirus products clearing well known analytics cookies, Firewalls blocking the page tagging scripts, and privacy setting in browsers. More an more browswers are starting to block first party cookies if there's not P3P compact privacy policy for the website. There's also the challenges around Javascipt not loading if there's an error, and latency issues, but the bottom line is that we've always known that page-tag based systems were innacurate. That's nothing really new. Because typically you didn't care about being perfectly accurate. We were more interested in being precise, and consistent. Here's an article that goes into more detail: http://www.advanced-web-metrics.com/blog/2008/02/16/accuracy-whitepaper/ So you could look at the big trends and see improvement. And that's where we've been over the last few years. What's different today is that we care about being accurate, because we need that accurate and precise customer intelligence for customer engagement. Sure we still want the overall trends, but we also need to have the detailed data too, and every session not captured is a lead lost, an opportunity missed. For what Google Analytics was designed to do, it is fine. My point is simply that there are things is can't give you, given it's technology architecture, and it's important to know where the gaps are. I hope I've answered your questions, and happy to provide more detail if I haven't.
Best Regards,
Darren
Hi Jonathan,
Just wanted to let you know I've responded to your question, with several links to articles but that has apparently thrown my comment into moderation in ZDNet. Working on getting it approved, but wanted to know an answer is comming.
Best Regards,
Darren
Because Urchin gets its data off of log files, it will report on what GA by itself cannot. So using both at the same time is a good idea if you are concerned by the gap, and if you have an intranet or internal web apps that need reporting too.
Since GA evolved from Urchin they use the same report terminology.
Most complex sites have multiple templates, sub-sites, and 3rd party add-ons. Trying to make sure that each and every page severed has a GA tag is not easy. Urchin helps in auditing content that is being served but tagged correctly.
Urchin 6 is not free.
Cheers!
This post has been removed by a moderator.
Hi Darren
Very interesting post. I work with both the OMS and Google Analytics (actually wrote a book about GA), but do not agree with you all the way. I find OMS really great and it have some fantastic features and functions to optimize a website. BUT, I dont think all you write is true.
You wrote: "Today, Google does not allow you to weight or score your content in this way, or even to prioritise visitors as leads and prospects."
It is not true. Off course you have to make some speciel implementation with the Costum var on page level, and on visitor level, but you are able to get a lot of data on this part.
I enjoy working with both systems, but it is importent not to think "OMS OR GA", but "OMS AND GA".
Cheers,
Jacob