Being the seventh-largest newspaper company in the United States presents an interesting challenge for MediaNews Group Inc.

The company, which owns such newspapers as The Denver Post and The Oakland (Calif.) Tribune, as a whole maintains more than 200 Web sites, 47 of which are daily newspapers. Getting accurate and timely Web traffic statistics for such a large operation is no easy task. Each newspaper site no doubt would like to have traffic figures available for its sales associates and editorial team, while making access to the data available to corporate as well.

Log processing

Traditionally, Web data is analyzed through the processing of log files. Log files are generated by the Web server(s) and can be set to collect characteristics such as length of visit, IP address, Web browser type and version, pages viewed and a trove of other data.

With popular Web sites generating page views in the millions, these log files quickly grow into the multiple gigabyte range and analyzing them can take computers days of continuous number crunching.

One alternative to this procedure is to outsource Web analytics in an application service provider type of arrangement. This is the approach that MediaNews Group took when the company adopted Omniture Inc.’s SiteCatalyst application.

Implementing an offsite analytics product involves one of two approaches, depending on how the site is built. In the case of a static site, where there are hundreds or thousands of fixed HTML pages, typically a script is written to parse every page on the site and insert the required JavaScript code.

Modification needed

Most sites, including those of MediaNews Group, are dynamic and database driven. In that case, the Web publishing system is modified to introduce the specified code in each page on the site, in effect inserting it into the “wrapper” that surrounds the live data. Once the code is in place and being actively called by a visitor’s Web browser, all traffic characteristics are tracked by SiteCatalyst and are available through a browser-based interface, often within an hour of being launched.

Omniture Chief Technology Officer Brett Error said that one area where SiteCatalyst excels is path analysis.

“I’m actually disappointed in my industry that we haven’t done a very good job helping Web marketers properly exploit this very rich source of information,” he said. “A typical report will have a list of every single session on the Web site and all the pages that were visited. The first one or two data points might be interesting, but with so many different things to do on a site, the data gets diluted very quickly and there’s no way to really base decisions on it.”

Graphical ‘tree’

SiteCatalyst presents this data as a graphical branching tree, using varying line thicknesses to represent the relative amount of traffic between pages.

The newest version of the software, 9.0, also features other graphical reporting tools such as Optimum Path ClickMap, which overlays actual clickthrough data onto an active Web page in a color scheme that allows the identification of most-popular links and paths through a site. Visitor identification, segmentation and transformation architecture, known as Vista, is a new technique that allows the segmentation of visitors based on user-defined rule sets. For example, visitors could be segmented into registered and non-registered groups. An editor could segment out a particular story category such as sports and track user behavior only as it pertains to that grouping of stories.

Tweak no more

According to Jim Shugarts, operations manager for MediaNews Group Interactive, before adopting SiteCatalyst the group was doing individual analysis of server log files.

“I’d keep tweaking reports and tweaking them, and our log files just kept getting bigger and bigger,” he said. “That came with growing traffic and with the company buying more papers. So we were looking to find an outsourced solution that would reduce the costs as far as machines on our end and allow us to get away from storing all these reports.”

Having multiple Web sites on one measurement system helped ease the collection of data. Individual papers have access to get their monthly statistics. The company is organized into regional ad areas, so ad managers have access to geographically limited data as well. On a system-wide level, MediaNews Group does a corporate-wide analysis.

In addition to the analysis of the traffic data at regular intervals, Shugarts said newspapers will often want to track the performance of individual stories as they hit the site, or individual columnists.

“At The Denver Post, for example, when the Avalanche (hockey team) won the Stanley Cup, we could go into the sports section, keep hitting refresh and just watch the traffic grow live.”

Having migrated entirely to this ASP arrangement, Shugarts was able to discontinue the accumulation of log files entirely. As webmasters know, this has the potential to speed a server’s response times, as the server isn’t hitting the hard drive with writes to the log file every time a file is requested.

Shugarts said the process of migrating over to the ASP model was quite simple.

“The code was very easy to implement,” he said. “We did it by using an include file (a file that automatically replicates itself onto selected pages of a Web site), so we knew what section and what pages they were going into. It was probably the first thing I’ve ever used that was out of the box and then literally up and running the very same day.”

Pricing for SiteCatalyst varies depending on the number of sites, the size of the sites and the amount of site traffic.