The double whammy of my loose talk of attention-whoring below and my avowal over at Oliver’s that I am not nor have I ever been a hit-slut has got me to thinking, as I am wont to do after too much coffee.
For someone who swears not to care whether he’s the Hit King Of Bumfuzz Nebraska or not, I do check my referrers and webstats a fair bit, and am always tickled to see one of those spikes that indicates I’ve mortally annoyed yet another group of harmless citizens. Again. Other than comments, which I seek most assiduously, because I believe in this two-way sh-t with a passion (unless of course you want to criticize me, in which case go stick your head in a pig), it’s about the only way I can tell how the heck I’m doing at this non-zero-sum game.
But I wish someone would explain to me how this hits and visits and pageviews sh-t works. I still keep those two little icons ticking over at the bottom of the page because I’ve had ’em since I started on Blogger way back when, and I’m nothing if not a slave to continuity. We also got a webstats package set up on the server a few months ago, and that never ceases to confuse the hell out of me.
For example, here’s my numbers (gimme the numbers, Harry!) for Friday of this week, a pretty much average day for this month.
Sitemeter says : 260 visits/460 pageviews
Nedstat says : 340 pageviews
Webstats says : 10669 hits/484 visits/1135 pages
What the hell do these numbers actually mean? Why are they so wildly different? Am I a f–king superstar yet? Will I become rich and famous, to go along with fabulously handsome and extraordinarily well-hung? Will I start making $6K a month, like whatsisface?
Not bloody likely.
The only stats thing I ever pay attention to is the neat little monthly graph from the Sitemeter gizmo, anyway. But I am genuinely curious as to how on earth these different numbers can be reconciled, what they actually mean, and if they reflect in any way at all the actual number of people who visit this site and shake their heads in bemusement at my latest textual antics.
I sure as heck don’t know. Vanity is the cheese in the submarine sandwich of social intercourse. But if you understand this stuff, I’d sure love a quick tutorial…
Category:
Metablogging Join the conversation! 9 Comments
Comments are closed.
Wanting a large audience for what you would do for a small audience is fine. Wanting a large audience so badly that you change everything you do to get one is not fine. You can be excited (or disappointed) by your statistics without becoming Mr Eternal Sweepsweek.
This Week On The Eternal Sweepsweek : I set myself on fire!
It’s something that very few seem to talk about openly, that I’ve seen, but I’m pretty sure there are very few of us that don’t pay at least some attention to the nuumbers of people that hunker down around our virtual hearthfires.
The only time I really see much mention of it is in a derogatory kinda way : ‘he/she’s obsessed with hit counts!’
I’m just trying to figure out what the heck the numbers mean. And I’m only being partially disengenous, honest.
Let’s see if we can shed some light on the meaning of the numbers… the Webstats stats are raw and unfiltered. They record every single instance of a resource being grabbed by any visitor, including you (hence mt.cgi being the #2 most visited page). Because it includes all the images and things, a lot of hits can be recorded for a single page view, especially considering your use of graphics to denote categories and the like.
Sitemeter and Nedstats try to filter and analyze your stats, but they have limitations. As they are only called by being included in a page, they do not record hits to any page that their code isn’t included on (which gets rid of all your movabletype files and such). They also do their best to figure out which hits come from the same people and group those together into “visits” and the like. It is a highly non-exact science, so the numbers are bound to come up different.
That said, I would take the number of visits reported by Sitemeter as a rough estimate of the number of visitors you get, then perhaps reduce it by 25% to account for people like me who read your site at work and at home and generate two different sets of visits for a single set of eyeballs.
Personally, I love statistics. Sure, some of it is stroking the ego bone, but I get a kick out of thought that people from so many different places are stumbling upon my stuff, and stats are the only way I see of finding that sort of stuff out.
Well, a lot of them *blush* come from balmy Portugal…
Just to add my two cents to what John said above:
Running a web server usually results in a log of some sort. Said log can track many things, depending on what level of sensitivity you set it to [1]. Now, based on this log, different packages try to rip it down into statistics that people can actually understand. This includes things like hits (which I understand to mean the number of requests for a given domain), visits (which I understand to mean a series of pages on your site visited by the same user in a short amount of time; try to figure out a consistant number there), and pageviews (the number of requests for a given page file, be it html, php, etc.). Now, when you consider the number of variables that play into this pile of definitions, you can understand why the different packages can have issues. For instance, consider this: every time you update your blog, you can ping weblogs.com. This causes weblogs.com to request certain pages from your site. This isn’t a real person viewing it, its a bot checking things out. What does that count for? Since its a request, it should count as a hit, but is it a visit? The bot views the page just like a person would, so should it be a pageview, even though the reader isn’t human?
In order to answer those questions, the stats package would have to check out the type of each request, and in some cases possibly the user agent. Given the large number of user agent strings out there (can anyone say browser sniffing?) its hard to come up with a way to nail all of them and discount a bot from a browser and vice versa. This can lead to a number of variations in how your stats appear. Also, as John said, tkae into account that you view your site pretty often in the course of updating, etc, and that some packages try to figure that into their counts, and you have a pretty crazy game going on; nevermind the search engine hits, the “myownblogindexerrobotpieceofcodethingythatrunsamuckaroundthenet” hits, and the fact that certain browsers allow for people to configure the useragent string themselves.
The bottom line is that you need to take everything with a grain of salt. I personally think that a referrer script like you have running is a good way to keep your finger on the pulse of who’s visiting and where from and how often.
If you have any questions, let me know. I’d be more than happy to try to clarify.
By the way, I’ve been reading for a few days now. Keep up the good work.
[1]http://httpd.apache.org/docs/mod/core.html#loglevel
> Well, a lot of them *blush*
> come from balmy Portugal…
A few come from barmy Poland, but (thanks to the magic of corporate networking) the stats probably think I’m in Denmark.
I’d check out that there Miguel’s site, too, but I don’t speak the Portuguese or whatever it is those people speak on Spain’s porch. (Is that where the name comes from? In American, it means porch-guys or something?)
Thanks for the mini-tutorials, guys.
And Eeksy, you’re firing on all cylinders tonight! I enjoyed your comments over at Blogroots as well. Certainly more coherent than my ape-like hootings.
at my last job, I used to tell my boss that interpreting web stats required a pointy hat and a bubbling cauldron. 🙂 jason does well with the details, I’d merely add the practical comment that the stats are best used as a relative measure: how do they compare to previous time periods?
> And Eeksy, you’re firing on all cylinders tonight!
Oh, sure. Now Matt and Meg and RCB and all those other folk who like to write about blogs and make a little cash from it will have me rubbed out. I’ll have negative stats tomorrow. I’ll be sleeping with the fiches and other obsolete data storage methodologies.