Judah,

For what it's worth, I'm pretty sure that (in contrast to javascript-based web
analytics systems), the audience measurement services are able to capture visit
duration for the last page visited on a site with their tracking software. I'd
have to double-check with my client reps to confirm...

David

-------------- Original message --------------
From: "Judah Phillips" <judahphillips@...>
Hello Jim,

Y'know I share the love and respect. I also hear ya loud and clear about
the differences in understanding the impact of a duration metric on "macro"
and "micro" level site analysis. In fact, your past, genial posts about
this issue are quite thought provoking.

Simply put Nielsen is providing a new metric for "measurement" not for
"evaluation." Such categorization makes the announcement easier to accept.
I get agitated in the evaluation of a site based on a time-metric *without
context*.

If I may briefly paraphrase the "macro"... what you are saying is
that measures of "frequency" on the web are so noisy, that using "duration"
as a proxy for "frequency" makes sense in that it is comparable across media
types because it is inexact in similar ways. Therefore, because of historic
precedent, a time-basis for media buying becomes a de facto "standard" for
approximating audience "weight" on the web. Thus, with a Total Time Online
metric, a buyer of electronic media can apply existing models to purchase a
known quantity of time (i.e weight) and predict a known result (i.e I buy X
weight, I get Y sales).

That makes sense to me if the theory holds true, but, darn, it is
suboptimal, especially as a measure of Web 2.0 sites. It's not just single
page visits that aren't counted, but the time spent on the last page in the
session isn't counted either. If the majority of your "engagement" takes
place on the last page in the session, then the "total time" metric fails to
account for that too. The time data is trimmed at the beginning and the
end. Take for example Zillow. In my personal experience, I engage the last
page I request, panning, zooming, checking my neighbors home values, then I
leave. In the current model, my time on the last page is not captured.

One could assume that rich internet applications and sites that serve
RIA transfer more megabytes then "web 1.0" sites. Does that mean I should
measure and rank on "megabytes transferred" as indicator of Web
2.0engagement? Is it a useful measure of "weight"? No, it's not, and
I'm not
sure "total time" is the solution either. That said, from the macro level
and business model perspective, I get why Nielsen is moving in that
direction. I also understand why they are backing off a bit on the
importance of the metric for "evaluation" and talking about how they are
still tracking page views and other metrics too.

I'll still give Nielsen props for making people think
about "the numbers" and using their influence to bring light to the
challenges of new media measurement. At the end of the day though, it's
still tough data to believe since I have no insight as to whether the panel
suffers from coverage error and selection bias, regardless of the technical
challenges to measuring time on the web.

Judah

[Non-text portions of this message have been removed]




[Non-text portions of this message have been removed]