Oleg Tulin

Oleg Tulin | 2009-06-26

I often read or hear the popular misconception about web analytics that the main rationale for its implementation is that it can help us identify and diagnose problems with our site's content, navigation, usability and other aspects of its design. Identify and diagnose problems, that's it.

The truth is, however, that this is only a minor one among reasons for implementing web analytics. The main reason is that... well, briefly speaking... web analytics is essentially, and to the fullest extent possible, a business intelligence unit for any business that uses online channel (it's XXI century -- modern, Internet-driven marketplace is right behind our computer monitors).

However, despite I myself tend to overlook that minor reason, "identify and diagnose problems" still applies. It caught me by surprise, recently.

That happened when I was watching visitors sessions on one of the sites that I keep an eye on (it sells electronic equipment for DJs and other musicians). I suddenly came across one session that made me look at it closely. For 21 minutes visitor was repeatedly downloading same file from website. Even worse, he or she was using not only onsite download links, but also download links for that file published by other sites (offsite links).

Thinking for a while, I came to the conclusion, that someone is desperately retrying failed download. I looked at reported response status -- it was OK (200). Puzzled, I came out to download that mysterious file myself and see what is wrong. The file was downloading (200 was reported right), but checking downloaded file revealed that for some reason it was corrupt.

This site undergone CMS change recently and, thus, there was significant probability of surprises -- so I was actually not very surprised. Previously, we had tested that site for broken links -- all links were fine. Still, link testing tool do not go all way to test integrity of downloaded files (understandably, since it is technically not feasible for obvious reasons). Who then is to detect screwed download? Site visitors are! And then -- from visitors' behaviour -- we are, the visitor trackers, if we watch our visitors closely. That is what happened to me.

You can see yourself what happened in that session and nearly feel visitor's desperation on this vivid screenshot (downloads are grayed). The most vivid sign of visitor's desperation is that he or she has even found offsite download links for that same file using search engine behind the scene.

For web analyst like me, who routinely analyzes visitors sessions and have trained eye, it was not too difficult to spot something is wrong with that session. But the key thing is that I trained my eye doing... guess what?... yes!, business intelligence work for client websites. I trained myself while doing exactly what I mentioned as major reason why business would want to implement web analytics. As it happens, working on major task helped me to do a minor task -- identify and diagnose problems with client site's content.

What kind of business intelligence, you may ask? Various. For example, often I was analyzing the behaviour of individual buying visitors (customers) for specific product. Or individual sessions of site's (product's) revisitors. Or those visitors who have demonstrated most interest in the specific product offer. Often it was real-time. From the given screenshot and above story you can get the feeling of how such studies are done.

Of course, studying sessions of visitors from some "focus group" is only part of the story, kind of "fine tuning" and hypothesis testing. Statistical methods play major role here, when thousands of visitors/sessions are analyzed at once to produce various performance indicators. But analysts cannot interpret those aggregate figures right, unless they themselves have followed visitors/customers through numerous individual sessions and well understand what motives drive those visitors/customers.

In all such studies, the key goal was to identify and interpret the typical behaviour of customers from specific market segment in relation to offers and product information. Do they care about product info placed on the site or come in already well-informed? Have they already made decision to purchase or are still deciding? To what extent price is important for them? And so on. Effective marketing of products and services via online channel is entirely based on answers to this type of questions. Know your customer, period -- this is the only way for business to succeed. Doing business online makes appeal of this motto even stronger. That is why this, and nothing else, is the major task of web analytics.

Comments | Tweet this | Read Blog

This blog is intended to share thoughts with people interested in web analytics, website metrics, analysis of online visitors/customers behaviour, online marketing, online business and (distantly) related matters.

Oleg Tulin, web analytics expert/consultant (olegtulin at gmail.com)

Read me in Web Analytics Blog