The need to measure Web site results is self-evident to any direct marketer. But exactly what to measure is still a bit of a mystery to many. Traffic volume is interesting but not very informative. Purchase activity is important but doesn’t capture behavior leading up to the transaction. Referrals from paid Web advertising and click-throughs from email campaigns illuminate only small pieces of the picture. Activity for individual pages, viagra such as where visitors come from and where they go next, pharm gives a tremendous amount of detail but is hard to put into context.
The closest thing to a standard approach for understanding Web behavior data is probably the “activity funnel”. To set this up, the user specifies a target behavior, typically a purchase, and identifies activity stages leading up to that behavior. Pages on the Web site are assigned to different stages and page activity is summarized by stage to show how customers progress through the buying process and where they drop out before reaching the end.
ClickTracks (ClickTracks, 877-773-2249, www.clicktracks.com) takes its funnel analysis very seriously. It subdivides each stage into pages or page groups, so users can see which pages are most effective at getting customers to move on. (Actually, ClickTracks shows correlation, not movement, because it counts all visits to each stage regardless of sequence.) It also assigns the customers to segments, so users can see how different customer groups are affected by the different page groups. The system presents this information as statistics and bar charts alongside images of the pages themselves. Differing background colors distinguish strong from weak performers at a glance. Exit statistics show the destination pages of visitors who leave the funnel. Users can compare results from two time periods to see trends or assess the impact of site changes.
If it sounds like there’s a lot going on here: there is. A fully decked out ClickTracks Funnel report, comparing multiple stages, page groups, and customer segments across two site versions, takes careful study to interpret correctly. But once you get the hang of it, ClickTracks presents a great deal of information in a visually efficient and attractive package.
Of course, this report is only as good as the underlying data. Assembling Web data raises two specific issues: gathering results in the first place, and classifying them into meaningful categories.
The most direct method of gathering Web results is to extract them from server log files. These capture the details of every page served by a site. Raw log files are massive and need considerable refinement to extract the information that marketers care about. They are not necessarily complete, because pages cached outside the server may be presented without a log entry being created. On the plus side, log files are created automatically and easily accessible.
The alternative to log files is page tags: bits of code (typically JavaScript) on each Web page that send a message to an external system every time the page is viewed. Tags avoid the problems with cached pages and can capture data about page contents that log files cannot. The messages can easily be directed to external hosted services, so little is required of in-house computer staff. But developers do need to add the tags in the first place. Tags can also create real or perceived privacy issues. (For a more detailed discussion of these differences, search the ClickTracks support page for ‘pros and cons’.)
ClickTracks, unlike most Web analysis vendors, offers both log file and tag-based approaches. The company reports that most of its clients use log files, although tags are gaining popularity. Clients can run either the log file or tag-based versions in-house, or subscribe to a tag-based hosted system.
The second challenge with Web data is assigning customers and Web pages to categories. This is particularly easy in ClickTracks. Both the log file and tag-based versions of the system store all data originally gathered about each event. Users can then decide after the fact which data elements they will use to define categories. This makes it easy to try different classification schemes and to define multiple, overlapping segmentations. Definitions can be based on patterns or common elements as well as specific values. Customer segments can be based on particular events, such as having viewed a certain page or purchased a certain product. Of course, such segments can be applied across multiple sessions only if the Web site contains persistent cookies or some other identifier to make the connections.
Although the Funnel report is arguably the heart of ClickTracks, there are several others. Site Overview presents key statistics including number of visitors, most viewed pages, common search queries, costs for paid search, common entry and exit pages, and average time on site. Search shows the number and performance of visitors by keyword, and also the site’s ranking for each keyword on Google, Yahoo! and MSN. Navigation shows detailed information about individual Web pages, including where visitors came from, how much time they spent, where they went next, and how many clicked on each link within the page. It superimposes the link information on an image of the page itself and can show statistics for different customer segments and date ranges.
The Campaign report can integrate with Google and Yahoo! to read costs for paid search campaigns, with SubscriberMail and IntelliContact for email campaign statistics, and with Yahoo! Stores and MivaMerchant for ecommerce revenue. Campaigns from other sources can be identified by parsing identifiers with ClickTrack’s standard segmentation methods. A separate Fraud Detection report highlights campaigns with unusual visitor behavior. Robot report counts visitors from search engines to each page.
The system also generates a simplified log file for users wishing to create their own reports from scratch.
ClickTracks was introduced in 2001 and has over 8,000 customers. The simplest version of the system starts at $49 per month hosted or $495 for installed, while the fully-featured Pro version starts $179 per month or $3,495. The installed version of the tag-based system starts at $13,995.
* * *
Leave a Reply
You must be logged in to post a comment.