Your browser is not supported. Please update it.

25 July 2007

Nielsen ‘time spent’ web metric, old hat or searing breakthrough?

Nielsen Media was widely quoted last week as saying that it would introduce the new concept of counting how long a person spends on a web page, as a key new metric for future web site measurement.

The only real affect has been for everyone to screw up their face quizzically and ask what Nielsen thinks it is doing? One person screwing up his face and asking that question was Conrad Bennett, technical services director at WebTrends, who talked to us this week about his company’s specialty, measuring web site metrics.

‘This all comes from the gradual death of the page view as a meaningful metric as Web 2.0 techniques find favor. Publishers everywhere try to conjure up extra impressions by arbitrarily spreading an article over 3 or 4 pages, so page views have come to mean less and less. But in Web 2.0 content tends to refresh within the page, with the result that people spend more time on a single page, at the expense of page views.’

Advertisers are not so much the issue here although they are keen to have standard metrics, but they have feedback from advertising servers every time an advert is served, every time an advert is clicked and they are increasingly trying to get to grips with newer technologies like how to measure the delivery of a video or audio advert, and ensure people have experienced them.

This is more about what a publisher can say about his or her own site, especially to an auditor ‘ that it has X million unique visitors, consuming Y number of page views and Z minutes spent on the site. Bennett’s view is that there is never a single statistic that describes a site, but that increasingly the most important of them is the number of unique visitors.

‘The problem is that there are so many ways of calculating this,’ Says Bennett, ‘you can count IP addresses, or drop a 1st party cookie into the machine, but even that only counts a unique user as one browser on a particular machine.’

‘The duration of a web visit is not so meaningful. There is a visitor timeout of around 30 minutes, so when people fire something up and have it open all day, refreshing the data every hour or so, they show up as multiple 30 minutes sessions, when they’ve just looked at the cricket score for two seconds once or twice. And it’s not a new metric, it’s been around for years.

‘All that you can really do is offer a range of metrics, not one chosen to make your site look good, which combine to give a flavor of your web traffic,’ said Bennett.