Picking up the baton of Mat’s metrics theme, a Beeb article asks whether we need to reassess completely how we measure internet traffic. The net metrics folks over at comScore have pointed out that a minority of “hardcore” (because it’s closely related to porn, except not at all) web users regularly clear all cookies from their PCs, thus leading to an overestimation of the number of unique visitors to a website. comScore call these nefarious types “serial resetters” (though, surely, “cookie cleaners” would have been both much cuter and had more potential for being spun off into a Saturday morning childrens’ TV show?) and suggest that we all shift over to a TV-industry style panel-system for gauging representative numbers of net users. (Not that TV panels are perfect – the UK BARB system works in such a way that it pretty much means that if your family is one of their panels, every time someone walks out of the room where the TV is, it represents 400 000 people switching off, which is a tad harsh, I’d have thought). It’s not just users who are, ahem, gaming net metrics – so are some websites themselves (because, yes, websites are people too). The methods are hardly sophisticated, but they can make for good PR. Notoriously slimy US tabloid site The Drudge Report (cannot bring myself to link to it due to the possibility that it will mean I end up in hell) recently released this bit of puffery:
Clever Drudge!!! As the lovely kids over at Valleywag pointed out, it’s a HUGE number (good PR) of page loads (which, to the average Joe means visits) – but it ain’t unique visits because the Drudge report is set to automatically reload every three minutes. As the kids say, it’s an “outrageous traffic trick”, and one, I’d add, that just indicates how tricksy assessing web traffic generally can be.
Which is why the sagacious personages over at Nielsen NetRatings are suggesting that we (you, me, and the whole wide world holding hands in a Diet Coke kind of way) start using the length of time spent on sites as a method of gauging popularity. The upshot of which would be the following top10 league table:
RuneScape – 6hrs 32minsElectronic Arts Online – 3h 07mBebo – 2h 37mFacebook 2h 28meBay – 1h 55mKing.com – 1h 53mAdventure Quest – 1h 35mFox Interactive Media (MySpace) – 1h 11mClub Penguin – Ih 10mCartoon Network – 1h 09m (Source: Nielsen)
Dwell time is a useful and interesting metric. On the other hand, Google has an average dwell time of a couple of milliseconds. The NetRatings release actually offers lots of different ways of assessing “engagement”.
Total page views
Pages per visitor
Total visits
Visits per visitor
Total time
Time per visitor
Google wins every one except “Total time” (where it comes second) and “Time per visitor” (where, as you say, the teenagers are in the ascendant — “Heh heh heh *snort*! He said Ass-End-Ant! Heh heh heh”)
It scared me a little until I realised that these were average times over a month.
[...] Cookie-Based Metrics Can be Flawed – comScore released a study this week indicating that metrics like unique visitors/users, that depend on cookies, are overstated by up to 150% because 3 of 10 U.S. Internet users frequently delete both first- and third-party cookies. Thus, if cookie-based analytics shows 10,000 unique U.S. visitors, the true number may be closer to 4,000. The Press Release is available here, with further analysis found here. [...]