Web Analytics Applied

What will Omniture do with Visual Sciences?

I suppose I can’t help but to chime in on this news.  Having worked with Omniture for quite a while, here’s what I think they’ll do:

If they acquire VS in Q1 2008, at first they’ll do nothing.  Omniture does not have the support infrastructure to change or integrate the already existing organization at VS, let alone deal with the VS client-management volume that such a change will undoubtedly create.  VS will likely continue to have its own website and GUI through Q3 2008.  The Visual Site tool will also be unaffected.

Then, in a big push, sometime in Q3 or Q4, they will create a kind of VS platform within SiteCatalyst, along the same lines as Genesis, Discover, TouchClarity, and SearchCenter.  It will be a separate tab in the user-interface, thus both promoting the Omniture brand and encouraging the transition to SiteCatalyst. 

I think they will create two levels of enterprise web analytics.  The old HBX will become a “Web Analytics Light” with much the same capabilities it has today, but perhaps without the customization, flexibility, or Visual Site components. 

SiteCatalyst, on the other hand, will become the more upscale product, fusing with both Omniture Discover and Visual Site, dropping Data Warehouse and ASI’s (which are already, according to rumor, going to be phased out), and (hopefully) leveraging VS’s Report Builder capabilities to merge with MS Excel.  Current clients of both Omniture and Visual Sciences will have to make the choice to upgrade to this new version, providing the opportunity for Omniture to increase its contractual revenue.  Those who do not upgrade will likely struggle in negotiations to maintain their current level of implementation and complexity.  Highly customized contracts – current today in both Omniture and Visual Sciences – will increase.

Such a two-level system would allow Omniture to compete with Google, increase its revenue, and still promote a product that will keep it competitive with WebTrends and other enterprise-level solutions.  Merging the two – HBX and SiteCatalyst – might take place eventually, but I don’t see that happening before 2009, and the two-tier system might prove profitable enough to postpone such a fusion.  Certainly, Omniture has been keen to move people away from the out-of-the-box SiteCatalyst implementation and into Discover and its other high-end products, and by having a cheaper, but far less analytical, alternative, they might get current clients to upgrade.  And a “lighter” version of HBX might be attractive enough for small businesses to replace the revenue stream that will be lost, as both current VS clients abandon VS, and prospects shy away from it, because of this merger.

What Visitors are Thinking and the Dangers of the Standard Design

In recent weeks, I’ve been in meetings and in conference huddles where “hard numbers,” quantitative, web analytics behavioral analysis was pitched against qualitative studies from surveys and usability teams.  The mantra goes, “web analytics tells you what a visitor did, and usability/survey data tells you why they did it.”  I’m not going to argue for or against this mantra right now, but I’d like to raise a slightly theoretical point at this “why they did it” question – and a concern for web analytics.

Qualitative analyses. where subjects are asked to test the usability or satisfaction of a website, have the usual challenges of bias, sampling, people lying, but there’s a deeper bias which could be more insidious: people could behave on (and evaluate) a website based on how they are culturally conditioned to behave on a website, by the universe and culture of website architecture and navigation with which they have been familiar for years.   There are whole theories around this kind of social behavior (Bourdieu called it habitus, Giddens called it structuralization).  Basically, surveying a heavy web-user about an unconventional website would be like plopping an ancient Greek in front of the Forbidden Palace in Beijing and asking him what he thinks of the architecture.

Let me give an example: usability studies on subjects in the early 2000’s revealed that people’s eyes are drawn to the top-left corner of a website page.  So many major portals – Yahoo, AOL, MSN, WebMD, Wikipedia, Martha Stewart, You Tube – put important navigation or important information on the upper-left.  Five years later, people surfing the web are used to important information being in the upper-left, reinforcing this paradigm.  If asked to comment on a site where important information was on the upper-right, I’m sure they’d complain.

Another example is conversion funnels.   Almost every purchase-conversion funnel I can think of asks for payment information at the very end of the process.  7 years ago, this was probably a statistically significant increase in satisfaction; now, I’m sure every survey study would say unanimously that’s where payment information should go. 

Other examples: Search (why always top-to-bottom?), Videos (why is it always a box with a >?), Products (always a thumbnail to be enlarged), header/footer (why not lefter-righter?), tabs (why always at the top?), drill-downs (why not drill-rights?).

The danger, I think, is that website designers looking at usability and satisfaction surveys are going to drive the web to a point where every website basically looks the same, much like it’s hard to tell the difference now between a BMW, Corvette, or KIA from afar.  In the last week I’ve been asked three times if there’s an “industry-standard” or “universal best practices” for website architecture or design.  But if such a “Standard Design” does emerge, my fear is that web analytics for the purposes of website optimization will be obsolete down the road. 

The classic studies of module- or link-placement, real estate analysis, funnels, forms, navigation, home page optimization, perhaps even information architecture, will simply become redundant, because there will develop a standard way people expect websites to be organized.  All web analytics will be about reporting, content analysis, and marketing, and not site-architecture or page optimization. 

Maybe it’s too late to do anything about these ur-templates – this may all have been decided in the 90’s.  But what about global websites in Asia, South America, or the Middle East designed by American firms, whose audience has not been similarly preconditioned to the Standard Design and who may have other cultural dispositions against the American or Western paradigm (recall that much of the world reads right-to-left)?

Why not build a site adventurously and creatively, and then use – you knew it was coming – testing on behavioral data to see if the real population respond significantly to these changes?  I’d love to see a website with a bottom-nav, tabs on the side, internal search where you scroll right-to-left (think of the advertising potential), new ideas for navigation (I’ve seen some good Web2.0 examples).  Let such a site launch in Qatar, and rather than sending surveys to focus groups there, use cheaper behavioral data to see if it rivals the Standard Design?

I’m not a web designer, nor do I have that background.   But as a web analyst, I have seen a certain standardization in many site templates which often are driven by focus-groups and qualitative analysis, after which traditional web analytics applied to site or page optimization is reduced to miniscule changes in the ordering of top-nav elements or placement of banner ads.  And thus loses its value-proposition for us.

Web Analytics Theory?

I was at Bentley College today with Joseph Carrabis and Judah Phillips, where we were interviewed about the practice of web analytics by the business school’s usability studies team. I haven’t been in an academic setting in quite a while, but a ghost of academia came up when we were asked whether there was a “theory of web analytics.”

I know the word “theory” well from graduate school. Every academic discipline has “theory.” Take cultural anthropology: there’s structuralism, post-structuralism, Marxism, neo-Marxism, positivism, processualism, post-processualism, even functionalism. These theories underpin the conclusions drawn from cultural-anthropological research. But at Semphonic, when I tell Gary I’ve got a “theory” about something, it usually means something like “I think the s.VisitorNameSpace variable is not configured right, causing the cataclysmic bug on site X.”

I suppose we’ve got “Functionalism,” but most of what’s in our white paper is methodology, not theory. The “theory” of Functionalism might be something like “All pages on a website have a specific function.” This sounds like theory – just as the theory of Marxist anthropology might be phrased as “all human activity is the result of class struggle.”

But what other “theories” are out there for web analytics? Here might be some examples:

“The value of all websites is a combination of ad-revenue, ecommerce, call-avoidance and branding.” (valueism?)

“The final KPI for all websites is expressed in $ terms.” (pecuniarism?)

“Success of a website is measured by the psychological engagement of the visitor.” (psycholigism?)

“The only way to measure the success of a website is through actual visitor behavior.” (behaviorism?)

“The homepage is by far the most important page on a website.” (homepageism?)

“Website activity can only be studied in conjunction with the offline business.” (integrationism?)

“Every website is different, so there can be no universal theory of web analytics.” (nihilism?)

Can there be, or should there be, a web analytics theory? Perhaps it’s a combination of all these “ism’s” above. Perhaps we’re just not at that stage yet. Perhaps there can be, but it will be relegated to obscure scholarly journals. Or, perhaps all web analytics practitioners implicitly have a theory behind all the web analytics they do, but it is never formalized. Who knows – maybe I’ll enroll at Bentley College.