Analysis of Web Analytics data only happens in very rare occasions.
At first sight that may sound quite counter-intuitive, as the majority of web sites nowadays is equipped with tracking codes.
What happens to most of the assembled data is this:
Standard reports are crafted (often data extraction is automatized), screen shots from reporting frontends are taken, and the only analytical approach is often limited to the calculation of the percentage increase or decrease in metrics from the last reporting period to the current reporting period: +5,5%, -12,8%, +1,2%.
“What would that have to do with analysis?”, you ask. “You’re right. It’s called reporting”, I’d answer. And we surely could agree that this only forms the first rung on the steep analysis ladder, hopefully leading to insight creation.
“Analysis” is defined as “the division of a physical or abstract whole into its constituent parts to examine or determine their relationship or value.”
This decomposition into “constituent parts” through analysis is somewhat problematic. After all a web site is only a technical structure for which the measured data can be decomposed in many different ways.
The reference point for Web Analytics data is not carved in stone. The measured entities (pages, visits, and visitors) are insinuating a reference horizon for interpreting and analyzing the desirable patterns. Technologists, user experience experts, and marketeers for sure have totally different reference points for analysis based on their perspectives, but assuming that companies and web sites are run based on objectives we must search further for a more common ground.
The “General Management” approach helps us out here, because a General Manager’s (GM) responsibility covers essential parts of company’s operations: revenue and cost. These two topics are thus forming the reference horizon Web Analytics data is supposed to support.
If analysis is not just done for the sake of it but with a perspective on costs and revenues we need to define the service’s boundaries as limits for analysis. These limits are, until today, mostly formulated as technical limits.
Web Analytics tools are deployed on single web sites – the entry and exit pages for a visit do mark the event horizon beyond which analysis cannot reach. Parallel (or: tabbed) browsing is not captured, neither is exposure to print and TV ads, or a face-to-face testimonial given by the user’s best friend three days ago over a drink or two.
So: we are only looking at a small playground within the user realm.
And a particular downside in Web Analytics is that the data doesn’t know anything about negative reflection values a fictive user might formulate verbally like this:
“If product A would cost 10 Euros less I seriously considered buying it. Instead I may either buy product B for a lower price (it will serve me nearly as good as the more expensive one) – or then I could try to find another seller where I can buy product A for less than the price offered on this web site.”
What you will see in your Web Analytics data, reflecting this cascade of considering alternatives is: pretty much nothing! A visit on product pages for product A and B, but no conversion, no sales, and no trace of the postponed or revised purchase decision.
It is no wonder that quite a lot of the efforts put into capturing Web Analytics data these days aims at optimizing the acquisition cost or revenue generating mechanisms. Decreasing acquisition costs and improving conversion rates and values are the core value propositions of Web Analytics tools.
And they perfectly fit the GM’s two-sided interest, too.
What seems to be left behind with this approach is a solid conceptualization and understanding of users’ needs. Formulated bluntly: web site users are expected to function as a servo mechanism of the purchase process.
You have seen how serious this becomes as soon as you once have entered a checkout process in which the site’s navigation got extincted from the user interface. Camouflaged as “to reduce distraction for the user” it is of course a means to reduce user defection – but for the sake of the commercial interest, unrelated to the user.
Still the total cost of purchase, the delivery terms and conditions are to be figured out only once the “Checkout” process is started on many web sites. Misattributing the funnel entry as purchase intent (and calculating “lost revenue” from all the funnel abandonments) is nothing but a negligent misrepresentation.