Over the past couple of years, digital analysts have seen one major headline eclipse all other news. And that is of the emergence and now well-established hegemony of Google Analytics.
Google Analytics, or “GA” as folks like to call it, has changed the analytics landscape. It has popularized analytics for folks who hardly know what analytics really is. It has brought analytics into the mainstream for most marketers. No longer the domain of deeply technological vendor specialists, analytics can now be implemented by anyone who signs up for a GA account.
Or Is It Really That Simple?
For some, it’s really that simple. If you have a small site and need to know the basics, you hardly need anyone to do anything for you at all (from an implementation standpoint). GA gives you a tracking parameter where you add your domain name and you plop this into the header of your page and suddenly you are part of the grand analytics experiment.
Where mainstream vendors send you expensive proposals, complicated account teams, service-level agreements, caveats, and contracts, GA sends you the tracking code. And this has proved almost a siren call to many marketers looking (justifiably) to break free of for-fee Big Vendor handcuffs.
But there is an undertone of smug satisfaction in the GA paradigm as well. Not among the knowledgeable users who know the value of analytics as well as its limitations. The knowledgeable GA analyst knows the major stuff – like the fact that GA owns all the data and does not offer a service-level agreement – and offsets this against the fact that it’s free and does nearly as much as a paid tool can do and almost all of what she really needed the paid tool to do.
The unjustifiable satisfaction comes mainly from folks who throw the code out there and say they’ve “done analytics.” They probably complained loudly that the paid tool didn’t do anything for them, and now they’re not paying anything, hardly notice that GA isn’t doing anything for them either. This is because they weren’t before and are not now really engaged in analytics: not looking at conversion statistics, not carefully defining what a conversion even represents for their business, or not checking to see if the data is accurate.
The fact that GA is free makes the insouciance rather easy. In certain organizations, it lets analytics truly become a checked box on the list of “stuff to be done” and, incurring no cost, then implies no need for commitment either. And let me state very clearly that I am not indicting the majority of GA users in this particular misdeed. I use GA for some of my own sites – it makes me neither a better nor a worse marketer to have done so. It’s what I do with the data that makes the difference.
Problem is, some people don’t know what GA does and what it doesn’t do. Knowing even a few of the lesser-known facts and limitations about GA may help dispel the notion that it’s a “set it and forget it” kind of tool. The following items refer specifically to Google Analytics’ free offering, not its more heavyweight GA Premium offering. Many items can be found in an informative book by Brian Clifton called “Advanced Metrics with Google Analytics.”
So here goes:
No. 1: GA ignores data after a certain activity level has been reached.
After 500 hits during a single visitor session, further activity is ignored. “Hits” is an old-fashioned word that no one wants to think about anymore, but servers deal with nearly everything on a “hit” level even if analysts think it’s not useful to reference this as a parameter. A hit is generated any time any item is requested from the web server – a tracking code, a GIF file, a PDF, a page element, an interactive module. And depending on how your site is built, getting to 500 hits might not be so far a barrier to bash into. And then the activity is ignored; it’s not counted.
No. 2: GA samples data after a certain activity level has been reached.
After you reach one million unique dimension combinations, GA samples the data. In somewhat plainer language, this means that once you have that many variables in play – and they do stack up fairly rapidly – GA selectively and intelligently ignores some of the data. It deploys sophisticated algorithms that make quick work of understanding overall patterns and dropping out the data least likely to result in a visible difference. It’s the same principle used by today’s universal image compression files (like JPG) – what you can’t see anyway will never be missed. Again, that is the driving principle. But when acute detail is needed, or where samples are very large and where algorithms may stray very far from true north, then sampling is not so great. It’s certainly something to be aware of if you are running GA on a heavily trafficked site.
No. 3: GA calls key performance indicators “goals” and allows you 20 of them per profile.
This won’t affect smaller implementations and there are workarounds to minimize the affect it may have on even larger ones. But for very large, multiplex implementations, 20 KPIs per profile may seem puny compared to the nearly limitless ones you can set up in the pay-for tools. Remember that KPIs may be any valuable action you define – from time spent on page to content downloaded to transactions to completing a registration. The list is about as infinite as the amount of clicks on all the sites in all the world. Again, this issue can be managed and mitigated and for many it will never be a problem. But enterprises beware: you are probably going to need more than 20 defined KPIs and the free GA tool doesn’t allow it.
No. 4: GA allows you 10 steps per conversion funnel.
So you’ve defined your goals (KPIs, value-triggers, call them what you want), and now you need to track how successful your “paths to conversion” are. These are the pages visitors get to on their way to conversion; and the success of these pages (in other words, whether they move the visitor to the next step in conversion) is critical to overall conversion success. It’s complicated because while you define these, there will always be lots of users who don’t follow expected paths to conversion. They may teach you something else about your content, but they won’t teach you the same thing as you will learn by watching what happens inside your conversion funnel. And GA lets you define up to 10 steps inside this funnel. This makes sense in my opinion. If you are creating funnels with more than 10 steps, you may want to simplify them anyway.
The above four limitations are based on a single guiding principle, and an essential one for a tool that’s free and allows anyone to sign up. That principle can be summed up as “equality.” GA wants to provide everyone with a very good, functioning, accurate set of analytics reports. It does not want those who might greedily gobble up all the bandwidth with their massive traffic to overload the mighty GA servers and slow down your humble reports to meaningless trickle. So it puts some limiters on the engine. It may feel to some like being forced to drive at 55 miles per hour, but it really is a question of resource conservation. Maybe we can live with it a little better if we think of it as conservation. It’s green! Roll with it. Or buy another tool?
The fifth fact is my own. It’s no less a fact than the others, but it’s not likely to be found on a list in a book about GA. Here it is:
No. 5: GA is no better than its implementation.
GA, in this regard, is identical to even the most expensive tools in the market. And we are not addressing the use of the data here – that would be an entirely different and very rich subject. We are talking about building the analytics architecture. About doing more than just putting in the tracking code they give you and looking at the basic reports. We are talking about ensuring the pages are tagged properly, that calls are being made from the page to the analysis engine, and that reports are populating accurately and consistently. Call it “implementation.” Call it “plumbing.” It’s the heart of any analytics project and it needs to be done correctly or the project comes to grief. Try not to leave this job to someone who has 12 other jobs too – nor to your content agency that may or may not really want to be measured quite so fervently as you want to measure them.
- You’ve planned your analytics environment advantageously.
- You’ve gone through a stringent KPI definition exercise and know what your site goals are.
- Your KPIs are logical and match your business needs.
- Your tagging and reporting reflects the choices you’ve made.
This will require mapping business requirements to actual possible reports that can actually be achieved within the tool. It’s just a bit harder than it sounds, and should be placed squarely in the hands of folks who spend lots of time making sure analytics really works.
And perhaps the most important fact about GA is one that shall go unnumbered: that it may be free to “get” but not necessarily free to “get right.”
When measuring the effectiveness of discount codes, retailers often get it wrong. In this article, we'll look at how data-driven attribution can help businesses better understand where discount codes produce the best ROI.
Data. It’s the latest ‘buzzword’ in the digital marketing world when it comes to content.
The term ‘marketing cloud’ has gained significant traction in the last few years as major software companies have sought to monetise the growing importance of technology for marketing teams.
Digital has quite forcefully overturned the entire media industry, causing even the most traditional companies to adapt or be left behind.