At a presentation last night to the assembled journalists (and a few randoms like me) at the Knight New Media Center online politics seminar, Politico.com editor Bill Nichols mentioned in passing that in March, the Politico site had six million page-views and two million unique visitors. That ratio really jumped out at me, since for most sites, the average reader hits two-ish pages per visit — sometimes painfully less, and for sites like YouTube on which people browse extensively, significantly more.
Doing a little math, Politico’s averaging three page-views per UNIQUE VISITOR for the entire month — in other words, their average reader only looked at three pages over the course of the whole month of March. Not the healthiest of ratios! Why does this matter? One classic measure of a site’s long-term success is its “stickiness” — the extent to which it keeps people on the site for long periods and also keeps them coming back for more. So far, at least in terms of traffic, Politico’s not looking so sticky.
By comparison, over the course of March, e.politics averaged about seven page-views per unique visitor (though from a few orders of magnitude fewer visitors…), and this is a site on which many stories can be read fully on the front page and so might yield only one page-view from several “article-views.” Other sites with a loyal readership (thanks, kids!) probably show a similar pattern.
Now, take this all with a grain of salt, since it’s hard to make any solid conclusions without looking at Politico’s traffic numbers in more detail. For instance, they may have a hard core of 20,000 or 50,000 readers who come back over and over, and a much larger pool of people who hit the site once from an outside link and then never show up again. In that case, they’re probably doing just fine, since they’re building a cadre of regular readers who’ll keep coming back. And when I asked Bill about it, he said that the Politico folks weren’t too worried about the traffic ratio, since they’re still a relatively new site and were also taking steps to provide enticements to keep readers on the site longer and more often. If I were those guys, I’d make that a priority. In its short life, Politico’s generated quite a splash and stirred up more than its share of discussion, but all that’s no good if the readers don’t keep coming back.
Update: Politico.com’s publisher responds to this and to the later article on niche vs. mass audiences. Read his note.
– cpd
Are you sure he didn’t mean 2 million unique visits? It’s very difficult to accurately track real-world individuals visiting a website over the course of a month because that technique relies on long-term cookies. Those can expired or cleared on purpose, or when people switch computers or browsers.
More likely, he meant that 2 million unique visits resulted in 6 million page views, which averages out to a very realistic 3 pages per visit.
The Politico uses WebTrends, which I’m pretty sure track only visits and not visitors. You use StatCounter, which looks like it tries its best to track visitors over a period of time, instead of individual visits.
Well, it all depends on what you mean by “unique visitor”. Everybody has a different length of time.
Usually, though, a unique doesn’t span multiple days. I visit today, I visit tomorrow, that’s two unique visits.
So, three pages per visit isn’t crazy.
Hey guys, good questions — Kari wrote hers before Luigi’s was visible to the public. In his initial statement and in his response to my follow-up question, Bill clearly said “visitors” rather than “visits,” though I’ll drop him a note to make sure that I didn’t mis-hear him or that he didn’t mis-speak. He seems quite savvy about web-related issues, so I’m sure knows the difference between the two.
As for the general question of distinguishing unique visitors out of a stream of visits, there are a lot of ways to do it and it varies by stats platform — by IP number and by cookie are the most common, I believe. I’m sure Webtrends will measure unique visitors — people pay good money for it b/c it’ll make your traffic stats stand up and bark.
BTW, I have Statcounter installed as an experiment and mainly rely on the analytics package that comes with my site hosting arrangement. I’m planning to install Google Analytics as another experiment but haven’t gotten around to it.
Thanks again — I’ll let you know what I hear from Bill.
–cpd
One more thing about stats — don’t take absolute numbers too seriously, b/c stats always have a whiff of voodoo about them. One visitor can look like ten and ten like one, depending on how their requests are routed (through proxy servers and such). For that and other reasons (spoofing, etc.), there’s always going to be a certain amount of uncertainty — the only thing you know for sure is your own server stats, i.e., which files were requested and from what IP number. Most experts I’ve read advise paying attention to trends in the data over time rather than to absolute numbers.
Even when you’re using numbers from your own server, different stats programs may interpret them differently, depending on how they define things like “visitors” and “visits.” That’s one reason I’m planning to install several different trackers, to do a side-by-side comparison. Geek alert!
–cpd
Good points all around.
One correction: I’m a “he” – not a “she”. Bio here.
Another point, cpd, it’s not just proxy servers and spoofing that can lead to the ten-to-one IP problem. If it were just those rare cases, we’d ignore ’em.
But this multiples problem also occurs in any house or office where multiple people are using a single wireless node. In my work, the fact that the entire Oregon State Capitol is on three IPs for staff and three IPs for the public wi-fi means that hundreds and hundreds of visitors to BlueOregon.com wind up looking like just six uniques.
And the reverse is also true for AOL dial-up users — since AOL can hop and skip a single dial-up session across multiple internet-connected IPs. That one person making a single visit looks like many uniques.
RE: using the wrong gender-specific pronoun — goddammit, that’s twice lately. I’m now going to refer to all of you as “it” from now on. –cpd