Posted July 10, 2007 9:40 am by with 30 comments

Tweet about this on TwitterShare on LinkedInShare on Google+Share on FacebookBuffer this page

Nielsen/NetRatings is going to dump page views as an official measurement of web site popularity and instead start measuring the total time spent by visitors on a web site.

“Based on everything that’s going on with the influx of Ajax and streaming, we feel total minutes is the best gauge for site traffic,” said Scott Ross, director of product marketing at Nielsen. “We’re changing our stance on how the data should be” used.

Nielsen will still provide page view figures but won’t formally rank them. Ross said page view remains a valid gauge of a site’s ad inventory, but time spent is better for capturing the level of engagement users have with a site.

Ok, so riddle me this Batman. With the rise of tabbed browsers, how do you account for those of us that leave a web page loaded in a tab but perhaps only engage the page for a couple of minutes? I’m always opening up interesting sites in a tab, with the intent of going back to them at some point in the day. Sometime I never do! Anyone know how Nielsen will account for these types of visitors?

  • Dean

    I have felt the same way about the “time spent” measurement and its potential overinflation. You can certainly stop a session after a set a time of inactivity but even that will result in inaccurate data.

    For the record I have been on for 3 weeks now so I am sure someone over there is giddy over the “time spent” data 😛

  • Scott

    The Nielsen meter is on the desktop of its panelists and can determine whether the browser is ‘in focus’. Only one application (browser, Word, IM, etc.) can be in focus at a time because the keyboard/mouse/etc needs to know where to direct user requests. For a browser, only one tab can be in focus at a time for the same reason.

    So if there are multiple tabs open, only one site is getting credit. If there is more than 30 minutes of inactivity, time is stopped and only 1 minute of time is credited.

    Hope that solves the riddle for ya 🙂

  • LOL @ Dean

    @Scott – hmm, sounds good, but I’d be interested to see if it works the way your described. What if I pull up a site and then grab coffee for 29 minutes before sitting back down?

  • rob

    I’ve never liked the metric “time spent on site”. It only helps in some situations. If it’s less than a couple seconds it tells you they came in and out fast – not relevant content for the reader. If it’s in the half-minute range for an article, you can guess they probably read it. If it’s in the minutes and there are several page views, then you can guess they stuck around and checked everything out. When it’s really high, I wonder if it’s a helpful stat. In other words, time spent on a site is only helpful in the context of page views.

  • Andy,

    Glad to see some push-back on this weak move by Nielsen. The search for one-size-fits-all metrics is a bit ridiculous to begin with, but “time spent”? For a ranking metric? And AOL gets the same credit for people who are on AIM all day and/or night as Google gets for a single search (which is ridiculous both ways if you think about it)??

    Sure, the page view has been dead for some time, but are “undead” metrics any better?

    Scott: Thanks for the tip. That’s pretty cool… I didn’t realize you could track it like that. Still, measuring time spent isn’t for everyone, and it’s an especially poor tool for ranking, IMO.

  • Pingback: Adventures in Blogging » Blog Archive » Quit obsessing over pageviews or your blog will end up in the Dead Pool!()

  • some combination of regular clicks (w/o long delay between( + time spent on site is probably much better metric than page vies.

    while it’s not perfect, it’s a lot more accurate than sites that engineer page refreshes or split data & behavior across multiple pages just to pimp the page views.

    clicks or time or both are likely pretty close to the elusive “attention” metric.

    – dave mcclure

  • Agreed. Time spent is definitely not the right approach. I really don’t understand what the huge deal is… if Ajax-style pages are the issue, then programmers can easily pass events via JavaScript to the Analytics package.

    The activity of the subscriber is key… and, of course, the conversion! both can be captured, it simply takes a little more work. Of course, if browsers and web analytics firms got smart, they’d capture any http request… be it in the page or in the browser.

    Some help from the folks who build this stuff would be nice!

  • Pingback: links for 2007-07-11 « ‘Cross The Breeze()

  • This is really a problem for time management. Scott it is interesting. But multiple tabs dont get counted.

  • I don’t think that the number of users that use a tab based browser compares to those who don’t. Numbers still tell us (in most cases) that internet explorer 6.0 is most popular. This may, however change in the future.

  • Bushido,

    ~75-100 million active Firefox users, and you still think that “most popular…in most cases” is good enough to cut it for a ranking metric? It’s been a decade since I took statistics, but that seems like a pretty mean deviation 😉

    No matter how long we pick at this, the Web isn’t meant for comparative metrics.

  • What happened to Nielsen sending me a dollar to tell them what TV shows i’m watching? I’ve got a paypal account…Show me the money:)

  • There is a place for new method of counting visitors. And the one who will do it first will be winner.

  • Jay

    Unfortunately, defining a universal “engagement metric” (sounds like what Nielsen/NetRatings trying to do?) is like finding the holy grail. So while we all are trying to crack the code on how to best measure engagement, Compete created an important sister metric a few months back called “Attention” (freely available on

    The Attention metric considers total time we collectively spend online and determines what percentage of that total time was spent on a given site. For example: If MySpace has Attention of 12%, it translates to: Of all time spent online by all U.S. internet users, 12% was spent on MySpace. You can also interpret the 12% as “the average internet user in the U.S. spends 12% of their online time on MySpace”.

    Since “Attention” is based on time, logically the more time we spend on a site, the more attention we give it. We think of Attention as finite – a pie-chart – so the sites that are increasing in Attention over time are performing well along this metric.

    We think true Engagement on the other hand is more of a spectrum that requires attitudinal inputs. The idea of a “universal” engagement is nice, but in practice it should generally be specific to a firm or product. What constitutes engagement for one product may be very different to even its competitors, let alone firms in other industries.

    At Compete, our view is that marketers need universal measures in order to put their own performance into context – relative to rivals, peers, or anyone else they want to compare themselves to. While we don’t present Attention as the king of all metrics, and certainly not as a one-size-fits-all metrics, we see it as an additional piece of the puzzle, and a step in the right direction.

    @Andy: Compete accounts for tabbed browsing by paying close attention to the sequence of pages being viewed. The window/tab/page in focus gets the time credit. So if there are multiple tabs open, only the page open gets credit. If there is more than 30 minutes of inactivity, time is stopped. Sounds like Nielsen does the same.

  • Pingback: links for 2007-07-11 | The Marketing Technology Blog()

  • I guess with all of its flaws, there’s really not a better way of tracking this kind of stuff – unless we hack into everyone’s web cam and watch them. 😉

    Thanks all for your thoughts.

  • It is hard to measure if it’s not a real time web application.

  • Real Pageview time should follow a normal (gaussian, or bell curve, distribution). In science we all the time strip out observations that sit significantly outside the normal curve – it’s called removing outliers.


  • Pingback: » Time spent metric doesn’t work for blogs, page views are still relevant A View from the Isle: Social Media, Blogging, Tech, and Opinion from Tris Hussey()

  • I leave tabs open for months. Somebody’s gonna get some really good scores with a metric based on the time a page is open.

  • Scott

    Hi all –
    Great conversation. Here are some more clarifications:

    1) For multi-tabbed browsers, we only credit the browser that is in focus (e.g., in the forefront). So if someone is visiting sites through multiple tabs, only one receives credit. If the whole browser is not in focus (such as when an IM, Microsoft Office, etc is in focus), no browser tabs are credited with time spent.

    Even with multiple monitors, only one application is in focus because your keyboard and mouse actions need to know where to be directed.

    2) If a panelist is inactive for more than 30 minutes, we dial back the time spent for the last page or application viewed back to 1 minute.

    3) We are not saying time spent is the sole metric for comparing traffic volume across sites. We still report on page views and we also report on sessions which provides info on the frequency of visitation. But while we’d say Unique Audience is still the best comparison across sites, we feel Time Spent is now the best comparison of traffic volume across sites since Page Views are less relevant for sites employing AJAX, Streaming, Applications, Gaming, etc.

    Best regards,

  • On a different note, does anyone know the rate of users using tabbed browsing? I know plenty of folks who have FireFox or the new IE but never use tabbed browsing.

  • I know quite a few people who don’t even know about tabs yet. Because of tabs, I can view a site for days at a time without actually viewing it for more than a minute. There should be a maximum inactive time cut-off worked into the measurement to account for tabs. Even on a blog with 10 posts on one page, I will click referenced links to investigate topics… and usually follow a rabbit hole from there. 5-10 minutes of inactivity on the original site is a good indication that I have moved on. But then there are flash sites with no page refresh… and AJAX also, so who knows. And that hacked webcam notion makes me paranoid.

  • Pingback: Nielsen//NetRatings Drops Page Views as Ad Metric at Catalyst Search Matters Search Marketing Blog()

  • I was going to add to the question raised by Carson Coots:
    sites employing AJAX, Streaming, Applications, Gaming, will typically have users engaged for more than 30 minutes with no page refresh. Can the stats deal with this?

  • Vince,

    I used to be in charge of developing meters for NetValue and then NetRatings, so I think I can answer your question.
    Users activity is only counted when there is mouse/keyboard activity.
    So, it is possible to measure time spent very efficiently, using:
    – users activity
    – browser visibility
    – network activity, etc.

    A “content” (I don’t want to use “page”) gets credit when it is *visible* on a computer that *someone is actually using*.

    All the issues mentioned (tabbed browsing, coffee break, word document window hiding the browser) are taken into account.

    I’m not allowed to tell more about the NetRatings meter, but I can provide with more details about alenty’s site-centric technology that gets the same quality as user-centric meters…

  • Pingback: Rules Of Engagement: Time Spent And New Metrics - Gaming Under Construction()

  • Pingback: An effective online ad? | Written By All Of Us | A Slack Barshinger Blog()

  • Pingback: Peelbacks | Written By All Of Us | A Slack Barshinger Blog()