The Logical Limitation of Google Analytics

December 15th, 2009 by Nick Goggans

Two distinct modes of scientific thought: “one roughly adapted to that of perception and the imagination : the other at a remove from it”. – Claude Levi-Strauss, The Savage Mind

“Any classification is superior to chaos, and even a classification at the level of sensible properties is a step towards rational ordering”

In a recent New York Times book review of Ken Auletta’s new book, titled Googled, Columbia Law Professor Tim Wu is quoted as saying, “If Google were a person it would have all the flaws and all of the virtues of a classic Silicon Valley geek.” In other words, and perhaps to define “a classic Silicon Valley geek” (I will not use my own definitions), that this, in Auletta’s construct, is a composition of the following attitudes: brilliant, with a tendency towards being socially inept, naive and arrogant. Indeed Ken (may I add rich?).

From my perspective, as one who analyzes web data and frequently uses Google Analytics to do so, I think that this personality set – or line of thinking – has pervaded the Google Analytics product and has potential detrimental side effects to the flourishing of greater understanding of the practical use of web analytics for business or individuals. Now, before I explain this, let me state that I also believe for advanced users the tool is innovating at a strong pace (yet seems to have lost some of the classic “google-ness” – I mean it’s not as simple as it thinks it is). In classic “don’t be evil” positioning, it’s free! and it’s going to help you whether you are an individual or a enterprise. It’s fantastic, it’s easy, and you know you need it right?

Let’s stop there for a second. Why is there a need for web analytics? To answer this question, let’s take a step back and review some basic anthropology (painless), and turn to Claude Levi-Strauss, and his classic, The Savage Mind.

The Anthropological Case for Web Analytics

In a general sense we find that humans don’t do well with chaos. The first need is to create order, and language is one area where this can be studied across cultures. Early anthropologists, for example, tended to believe that the “savage/primitive” cultures reserved language for flora and fauna that possessed a form of utility, such as food, medicine, or ritual. Further study, and differing understandings of earlier data, Levi-Strauss maintains that in fact there are many examples of societies where non-functional flora and fauna are named. What this expresses is a fundamental need to classify, such that the simple need for classification itself (and the performance assigning names) is utility: the creation of order to simplify experience perhaps. Levi-Strauss explains it this way: “…animals and plants are not known as a result of their usefulness; they are deemed to be useful or interesting because they are first of all known.”

Thus, existing in the 21st century, we find ourselves extending into another environment beyond our former village/jungle/desert/city to a digital world where the equivalent “flora”, “fauna”, and “geography” – and connections and associations between – are not easily defined. This makes us unsettled, just as it would if I dropped you off in a foreign landscape where you didn’t understand the language. In a way, our extension into the digital world creates an existential wild west: where am I, and how do I relate and impress this digital environment (whether ‘I’ is a company or individual). So we turn to tools that may help to order this chaos, and find our place.

Thus, we turn to web analytics not really with a practical drive (i.e. in satisfying practical needs like “I need more sales” or “I want to be more popular”), but through an intellectual drive (“where do I exist?” within the rest of the digital world and “how I am connected to it?”).

This is where we get to the solution and the problem with Google Analytics. Google Analytics does a fantastic job at solving the intellectual problem, I know my general results, how people find me, etc. However, it seems to be failing at scaling the next level for many practitioners, that of satisfying needs.

Now, of course Google Analytics is capable of reporting practical needs through setting up goals and monitoring their performance, but this is not how most people are using it. It is this question that led to this entry, which is “Why don’t more people use the goal setting capabilities in Google Analytics?” Is it a design issue? A marketing issue? (Yes, many people do establish goals, some even use these to make decisions, but most users do not from my observation and discussions with other analytics professionals).

So what is the issue? I think it has little to do with the design or marketing – though I initially thought it was a UI issue as the goal setting could be easier and integrated within the reporting pages, but that’s another discussion.

The reason I believe that the use of goal setting features is the exception not the rule of application of Google Analytics came through the combination of Levi-Strauss and reading the book review on Auletta’s book. It is that the Google culture has created a product that allows for productivity of a certain kind of analytic mind, at the exclusion of the other the kind of mind that is more intuitive than methodical, which as we see historically accounts for probably an equal amount of major scientific discoveries as the scientific method approach, which is a modern construct (see examples of intuitive discovery process neolithic innovations to something like finding the right filament for the light bulb (carbon over cotton), which was a true study in not so much scientific method but savvy application of an understanding of what elements in the natural world to try out).

The DNA of Google Analytics Too Structured to Produce True Problem Solving At Scale

Google Analytics, by design, because its DNA is so highly scientific (being a system built by engineers for data analysts), does not easily allow (especially if it becomes industry standard) for what Levi-Strauss calls the “science of the concrete”, (and I call “messing around with tools and things you know to solve problems you’re presented”) described here: “discoveries…authorized from the starting point of a speculative organization and exploration of the sensible world in sensible terms. This science of the concrete was necessarily restricted by its essence to results other than those destined to be achieved by the exact natural sciences but it was no less scientific and its results no less genuine.”

Google Analytics is largely created to presuppose and establish a universal order and lexicon to the study of digitally created causes and effects within websites (and digital advertising). I do not believe that this application of universals can take hold before users are able to apply very basic custom analytics solutions according to their “local/personal” needs, which will have their own lexicon and will at first seen as disorder, but it is in truth the way that this data will be looked at – and is inherently a more evolutionary approach to the application of web analytics to create actions and knowledge, rather than the “intelligent design” approach that we get thrown into with Google Analytics (and trying to establish our own systems into other businesses”benchmarks”.)

Google Analytics is a full digital taxonomy applied to users who may not have a need to classify digital events to the same level of detail that the system allows by default. Users, presented to this deep system of metrics and filters then become paralyzed to use the data to solve “practical data issues”, thus resorting back to simply using the system to look at basic data on web visitors or total pageviews – not leveraging the great segmentation reports available.

As Levi-Strauss explains with language, that a wider sort of classification, does not mean a higher understanding or a more enlightened mind. For example, if I have a word for “tree” but not “oak, birch, magnolia, etc” that doesn’t mean per se that I am less off from both an intellectual or practical sense. Taking this to web analytics, I think the basic problem is too many people trying to sort out ‘what’s an oak;”, ‘what’s a birch?’, rather than ‘what’s a tree’? Or to carry the analogy into the system, instead of “I want to track .pdfs, or facebook entries, etc.” to step back, “what’s a goal?” and then define it.

Thus, in an effort to advance web analytics as a discipline (and yes even a scientific discipline), there is a need to discover more tools and applications that allow a localized/individual (meaning, answering: “What do I need to know?”) not a universal, an intuitive rather than scientific approach to acquiring, presenting, and analyzing web analytics in the short term. It is this step that we must return to, in my mind, to begin to build out to a potentially universal method and lexicon to web analytics. Another way of saying this: get your house in order before worrying about what your competitor is doing. They are probably producing data at this point that is not apples to apples to you (i.e. each company may have wildly different definitions of what a “conversion” is).

There is a need for companies and individuals to understand their digital impression. Further, more questions need to be asked upon the data being collected in websites, social networks, cell phones, etc. What we need is to start asking more questions to measure with what we have – not to assume that the great Google Analytics system becomes our default order for all digital data, which, if it were to occur would bring them even more leverage than exists with search.

Lastly, I want to leave with the notion that the best thinking and the greatest innovations occur at that marriage of intuition and the scientific approach – as seen in the examples below. For our digital age, we are at the beginning and it will be interesting.