Tracking user engagement with websites

Conrad Bennett, Technical Services Director of web analytics provider WebTrends, discusses the flaws in various forms of web measurement.

Nielsen recently announced plans to measure time spent in order to track user engagement with websites, while ComScore also intends to enhance its measures this year.

There has already been some debate in the industry around this growing focus on finding out how much time people spend on sites, as well as other potential measures of "engagement".

This comes as a result of an ever-increasing number of sites beginning to incorporate Web 2.0 technology - from user-generated content such as blogs and forums, to flash and audio and video, and is primarily driven by attempts to track the value of investments made by advertisers.

Web 2.0 does require a shift in the metrics site owners use to judge the performance of their website.

With many Web 2.0 sites, for example, page views as a measurement of user behaviour no longer delivers accurate results, since the concept of pages is meaningless on sites where users might be carrying out several activities on a single page (such as uploading their own content, commenting on other videos, and directing people to their blog).

Rather than measuring pages, website owners will want to measure unique visitors to find ways of segmenting their user base into distinct groups based on their behaviour, behaviour which will of course vary wildly between sites with differing objectives.

There is also a significant challenge in meeting the requirements of advertisers, who want simple at-a-glance metrics that are comparable between sites.

Advertisers also want big numbers, as do site owners when they are engaged in financial discussions with advertisers.

All the standard metrics - page views, visits, and unique visitors - have flaws in this respect, either because they can be manipulated with site design, or because there are insufficient standards in the industry.

The fact that page views are less valuable as a form of measurement in a Web 2.0 site is actually a red herring, as their worth as a metric is highly debatable in any site - Web 2.0 or not.

This article for example could easily be split over two pages when posted on a website - this doesn't mean the site is more successful.

Visit duration is no better, because it doesn't measure actual interaction with the site, and whether the user is simply struggling to find what they want because of overly complicated site design.

On the other hand a particularly efficient site may have people in and out in minutes; again this is not a reliable sign of a site's success or lack of it.

Perhaps advertisers do want users lost in a site, clicking back and forth and encountering their ads more frequently and for longer periods of time.

However, this is certainly not a key objective for site owners.

I am sure though that advertisers wouldn't take such a short-term view as ultimately happy customers are returning customers.

Unless you have a monopoly on whatever you sell, it needs to work quickly, smoothly and with absolute clarity - which opens the door to another debate on the practicality of Web 2.0 technology for e-commerce and B2B business sites.

Site owners typically now maintain one set of metrics designed to report only the numbers required for audit purposes, and another set for actual analytics purposes in the organisation to understand how users interact with their websites.

The only way to truly understand what is happening on a site is to sit down and spend time looking at the metrics.

Unless advertisers have the time or inclination to do this, how will they ever know what consumers are up to?.

Back to top