ComScore Suggests New Web Metric: Site Engagement

[PaulPA]

While this article article [washingtonpost.com] is written by a ComScore employee, the concept of measuring user engagement with a site is interesting.

One argument for new metric:

Lately there has been controversy around the efficacy of the page view as the gauge of Internet consumption. A new technology called Ajax enables content to be refreshed on screen without serving a new page view.

What is engagment:

Engagement represents an attempt to capture the quality of a consumer's exposure to content. Not all exposures are engaged exposures, and we want our measurement of ad media to account for the difference. It goes beyond "how many?" and "how much?" to ask "how good?"

How it will be measured:

At ComScore, we plan to introduce a engaged duration metric this year. We're able to track engagement with a Web entity with software that allows us to "see" exactly what applications or sites users are interacting with. Even if users have three or four browser windows open at once, we can track when a particular window becomes active.

[Receptional Andy]

It's an interesting idea, although I would have liked some more technical detail as to how they're actually going to implement this. And while page views is bad for Ajax, it isn't so bad for other things.

A page view is generated when one computer loads one Web page one time, regardless of whether the user stays on the page for two seconds or a half hour (indeed, regardless of whether there is anyone in front of the computer at all).

This cuts both ways though: if I spend half an hour reading a news article in depth, they've got no way of knowing if I was in front of the screen or not (measuring scrolling is a possibility with javascript, I suppose).

Nonetheless, it's good to see web analytics continuing to advance from the humble beginnings of looking at requests in server log files to more human-orientated metrics.

[PaulPA]

I believe this approach could work to some extent if the methodology uses customer panels and sampling. This will allow ComScore to install the proper monitoring software to develop the metric.

However, as many people are aware, such methods tend to miss a lot of people especially those accessing the Internet from tightly controlled environments such as many workplaces. It seems at best it could work for measuring retail sites which are probably accessed more frequently from home then work but even then the undercounting could be quite high (e.g. missing much of Cyber Monday traffic a large portion of which comes from workplace computers)