Thursday, August 14, 2008

WASP case study: SiteCatalyst hierarchies

As I'm getting closer to WASP v1.0 release, I have started to work with some of my clients to put the tool to the test. So here's first real life case study of WASP!

The client: Quebecor Media Inc

Quebecor is one of Canada's largest media companies. More specifically, Quebecor Media Inc. encompasses nearly 200 websites with very diversified ebusiness requirements, strategies, and technologies. In terms of web analytics, it's what I would call a "nice playground", with sites ranging from multilingual media portals to bookstore, dating, classifieds, jobs, search, etc.

This case study looks more closely at, a (French) TV channel website with information about shows, schedule, artists, contests, live and on demand video, etc.

The web analytics tool: Omniture SiteCatalyst

QMI took the right approach of defining guidelines and requirements for their SiteCatalyst implementation. This way, a common baseline is shared across all sites, easing deployment across various business units (despite different technologies and different development teams), standardizing the terminology and aggregating traffic into roll up accounts.

The challenge

Although most sites are template driven (CMS, eCommerce catalogues, etc.), there are always certain pages that needs to be tagged in a more manual way. This is often the case for transactions: subscriptions, checkouts, etc. But with dozens of templates, some of them might be missing tags or using the wrong version of the implementation code.

SiteCatalyst hierarchies

While looking at the SiteCatalyst Hierarchy report, we uncovered some levels had a lot of “Unspecified” values. We focused on the "network|vosemissions" subset of the site (see image), which was crawled using WASP. Omniture knowledge base article #1941 attempts to explain the "Unspecified" value but here’s a simpler explanation:
Unspecified hierarchy: If some pages do not set the hierarchy variable (hier1) to its full extent, all those page views are summed up into "Unspecified" so the total of pages views at that level match up with the total shown at the level above.
Since the hierarchy variable is often populated automatically, a simple error or unexpected condition might cause the value to be missing or invalid (think about special characters, blanks, javascript errors, etc.).

    WASP in action

    It took WASP less than 15 minutes to crawl the first 1,000 pages of that site section. By using the WASP "Save as CSV (Comma delimited)" option and filtering on the Hierarchy column (h1) for pages with only "network|vosemissions", we were able to identify a couple of pages that didn't have the 3rd hierarchy level.

    Templates such as artists bios and specific show episodes were using an earlier version of the implementation code.

    Problem solved!

    Why WASP is better?

    You could browse your site manually and look at debuggers and proxies until your eyes end up in the same socket... You could use a very expensive tool and try to get this level of detail after several hours... Or...

    >>> Install WASP now! <<<

    That's what it's all about: making it easier to do quality assurance of your web analytics implementation.