Measuring Visitor Engagement Take Three: Time Spent on Site

Analytics · By On · 3 Comments

This is the third in a series of posts in which I dis­cuss the poten­tial “best” mea­sure­ment for online adver­tis­ing. While some audi­ence mea­sure­ment firms believe time-spent-on-site is set to take over the wan­ing page view as the most effec­tive mea­sure of vis­i­tor engage­ment, I believe they’re wrong.

I believe it is actu­ally impres­sions that should become the stan­dard by which both buy­ers and sell­ers of online media will begin to nego­ti­ate their buys. And per­haps, in the future, another met­ric like clicks will gain more foot­ing in the non-search adver­tis­ing world (clicks are already the stan­dard in search advertising).

For a quick sum­mary of my argu­ment, read the first in the series. For more on time-spent-on-site, read on…

With its announce­ment, audi­ence mea­sure­ment firms have sug­gested that time-spent-on-site is a more equi­table way than page views to mea­sure vis­i­tor engage­ment in a Web 2.0 world.

The ratio­nale is that Web 2.0 tech­nolo­gies like AJAX do not adhere to the tra­di­tional page metaphor, so time spent on site is a bet­ter form of engage­ment mea­sure­ment. (For a descrip­tion of why pages that use Web 2.0 tech­nolo­gies are dif­fi­cult to mea­sure using tra­di­tional page views, see Mea­sur­ing Vis­i­tor Engage­ment: Brave New World or The Emperor’s New Clothes.

To that end, Scott Ross, direc­tor of prod­uct mar­ket­ing at Nielsen com­mented, in a Beet​.tv inter­view that, based on every­thing that’s going on with the influx of AJAX and stream­ing, total min­utes is the best gauge for site traffic.

By way of exam­ple, Ross ref­er­enced, in the same inter­view (or read about it in a Com­put­er­World arti­cle from Oct. 7, 2007), a com­par­i­son between MySpace and Youtube, where MySpace out­num­bers YouTube’s page views by a fac­tor of 10 or 11, but the ratio of time spent on MySpace is about 70% less — at roughly 3 to 1.

Sites like AOL and Yahoo also cat­a­pult to the top of the “engage­ment” list, dri­ven by inter­ac­tive appli­ca­tions like Instant Mes­sen­ger and email.

With all of this in mind, it becomes increas­ingly evi­dent why time spent on site has attracted so much atten­tion. In a way, audi­ence mea­sure­ment firms didn’t have a choice. They needed to move beyond the page view to mea­sure engage­ment, or the “qual­ity” of the visit (but not nec­es­sar­ily the qual­ity of the visitor).

But some­thing else hap­pens with time-spent-on-site. Audi­ence mea­sure­ment firms regain the high ground. Why? Because time spent on site is actu­ally dic­tated by the pan­elist, not by the site they visit. In other words, time spent on site is mea­sured by a meter that sits on the user’s com­puter — and unlike page views, sites have lit­tle way to manip­u­late or cor­rupt the number.

At this point, you could argue that this is great — kill two birds with one stone. Improve engage­ment or quality-of-visit mea­sure­ment, and offer a met­ric that can’t be dra­mat­i­cally biased by sites themselves.

While I haven’t seen any arti­cles men­tion this angle of the busi­ness (con­trol vs. not), it’s nonethe­less sup­port­ive of the claims that this is Brave New World ter­ri­tory and bet­ter for the indus­try as a whole.

But as I men­tioned ear­lier, I actu­ally disagree.

Time-spent-on-site has been avail­able for years as a stan­dard web­site met­ric.  In fact, it has changed very lit­tle since the pre­his­toric era when server log files were used to mea­sure site traf­fic (yes, it was avail­able back then).

So is it really a step for­ward for audi­ence mea­sure­ment — an indus­try that has recently come under fire for major chal­lenges in their under­ly­ing method­olo­gies and data accu­racy?  And is it really a step for­ward for adver­tis­ers and pub­lish­ers — who have like­wise strug­gled to cre­ate an effec­tive mar­ket­place for buy­ers and sell­ers of ad inven­tory (hence the mete­oric rise of third-party ad networks)?

I think not.  Next time, I’ll write about impres­sions, and why using impres­sions offers a sig­nif­i­cant oppor­tu­nity for adver­tis­ers and pub­lish­ers to cre­ate a more effi­cient marketplace.

  • Tammy A.

    Time spent on site is also one of those tricky met­rics because it can be arti­fi­cially high. Back in the day, I used to watch the aver­age time spent on site met­rics for a par­tic­u­lar site I was man­ag­ing. While inter­est­ing, it could also be mis­in­ter­preted. More time spent on the site doesn’t always equal a hap­pier vis­i­tor. It could be they couldn’t find what they wanted and had to spend more time search­ing. It could also mean they were on your site and sim­ply left their browser open and went on doing some­thing else, which arti­fi­cially inflates the sense of engage­ment they had. (Of course even­taully the 30-minute rules comes in, but I think you get my point.) Should be inter­est­ing to see how this plays out.

  • David Ris­date

    Though I agree with your basic premise about Time Spent being poten­tially not a step for­ward, I fear you approach this from a very col­ored per­spec­tive (being employed with a site ana­lyt­ics firm).

    Time Spent, as mea­sured by an audi­ence mea­sure­ment firms, is accu­rately cap­tured by soft­ware on a users com­puter. How­ever, com­pa­nies that use page-tagging meth­ods to track behav­ior are unable to pro­vide a com­plete pic­ture because the last page of a visit is never included in time spent cal­cu­la­tions (since there is no sub­se­quent server call to com­pare time­stamps). This is sim­ply a fun­de­men­tal chal­lenge in page-tagging and accepted as being a ‘flaw’ in the process. How­ever, this means that Time Spent mea­sures from com­pa­nies like Core­met­rics and Omni­ture are inher­ently incorrect.

    What’s odd in your post­ing above is that you say you dis­agree with the push of Time Spent, yet you don’t men­tion why. Pithy state­ments such as “I think not” don’t pro­vide any detail as to why you are dis­agree­ing with the indus­try. And you don’t make any men­tion at all of the mea­sure­ment flaw noted above and how it puts com­pa­nies like yours in a more chal­leng­ing posi­tion to find alter­nate meth­ods of mea­sur­ing ‘engagement’.

  • http://www.fanhub.com michael k

    Just fin­ished read­ing all three ‘vis­i­tor engage­ment’ entries and now have more ques­tions than answers! I run a resource site for trav­el­ing auto rac­ing fans with a decent amount of con­tent (i.e.: city guides, venue feed­back, images, restau­rants, lodg­ing, etc.). The site employs both tabs and Ajax, so it seems that impres­sion num­bers will never give me the com­plete story when it comes to users. Because of this, I have been using the ‘time spent’ as a major KPI. Fur­ther, when it comes to adver­tis­ing, it seems that impres­sions and ‘time spent’ can’t be separated.

    If I am com­pet­ing for ad dol­lars against another, sim­i­lar site, it seems like there has to be a way to lever­age ‘time-on-page’ in addi­tion to merely impres­sions. If both sites have the same num­ber of over­all impres­sions on the Tal­ladega Super­speed­way page, but only aver­ages 30 sec­onds per page; how­ever, my users where on the page for over 3 min­utes, wouldn’t I be leav­ing ad dol­lars on the table if I can’t lever­age that fact? If noth­ing else, it seems like my site’s higher time-on-page num­ber will con­vince the adver­tiser that my site is more wor­thy of their dollars.

    If impres­sions were the met­ric adver­tis­ers used to com­pare the two sites, wouldn’t they sim­ply ‘go with their gut’, pick the pret­tier site, or even, flip a coin to deter­mine which site to adver­tise on?