The WAA has released much needed web analytics standards, but are they sufficient? Read ThinkMetrics' CEO's evaluation.
On August 23, the International Web Analytics Association (WAA) finally released its first set of standards. You can obtain a copy here.
The Standards Committee has been working on these for years, although you can't tell from the result, which is patchy. Some of the standards are extremely well thought through, and some are just plain flakey. However, while far from perfect or complete, the standards the association has created do take web analytics another (small) step forward.
Prior to the release of these standards, all we had was a few vaguely defined items from JICWEBS, developed by the advertising industry, and blatantly biased towards maximizing revenue rather than reporting objectively. The new standards provide what the WAA considers to be a foundation upon which later standards can be built.
The WAA has not provided any identifying name to this set, so I am calling it the Level One standards, in the same way as we have DOM Level One, DOM Level Two, and so forth. The WAA Level One standards are divided into four groups; Building Block Terms, Visit Characterization, Conversion Metrics and Content Characterization.
Building Block Terms are those fundamental measurements that are necessary in order to describe any form of web activity. These are the measurements that are then used in calculations to create other metrics. There are seven such terms. They describe the visitor, the page and the visit.
The Visit Characterization section contains 13 metrics that can be used to describe a visit. These relate to the path people take through a site; how they enter, which route they take inside the site, how they leave. Visit Characterization metrics distinguish between an Entry Page and a Landing Page and provide definitions for five different types of referrer. There is some extremely useful information in this section.
Only two metrics are provided in the Conversion Metrics section: Event and Conversion. Obvious absences are Abandonment and Abandonment Ratio. While calculating your conversion ratio is the starting point, if you want to do anything about it you'll need to assess your abandonment rates.
The Content Characterization section is an attempt to provide some metrics for the analysis of content. The metrics provided are Page Exit Ratio, Single Page Visits, Single Page View Visits and Bounce Rate.
I must confess to being disappointed with this section. I think it is thin, the definitions are typically much shorter than those elsewhere, and I am not convinced as to the logic. According to the WAA, a visit bounces if and only if it involves a single view of a single page. If someone reloads the page, it is no longer a bounced visit. I think this is questionable because, in my view, the time between reloads is critical. If someone reloads a page within a second or two then leaves, I would count that as a bounce. According to this standard, it's a visit. However, multiple meaningless reloads happen all the time on websites. They are so common that some metrics systems don't count additional page requests if they happen within one to two seconds of the first.
In my view, the primary indicator of a bounce is time, not the number of page views. The Bounce Rate is an attempt to assess how many people arrived at the site but failed to enter. This is because the first thing people do when they land on a new site is assess that site to see if it contains what they are looking for. No one reads every site landed on. Therefore, the first thing any site must do is convince new arrivals that the site is worthwhile. You can't sell anything if no one enters the store.
If someone skims five pages in less than 10 seconds then leaves, I don't think that person entered the store. If someone spends 15 minutes reading your Home Page and then leaves, I think the person entered, even if not going in very far. Not everyone assesses sites by looking at a single page. Some people work on the basis that pages are of different quality, so they scan a few quickly to get a feel for the site. I would have liked the option to cross-reference time with page views when it comes to assessing bounces.