Real world statistics

Traffic

Traffic data is useful for the economics of blogging, namely determining how much money can be made from advertising and occasionally which hosting plan to use. To aid calculation, the rounded statistics are:

Unique visitors

1300

Pages viewed

6700

Bandwidth

80 MB

The software makes the distinction that this is for “viewed traffic” which excludes traffic generated by robots, worms, or replies with special HTTP status codes.

So what are the financial implications? Well, the numbers vary, but one estimate says If … you get 3,000 page impressions a day [you] make about $270 per month. As my blog is getting about an order of magnitude less traffic, I wouldn’t expect to be make more than a dollar a day from my blog. With unmetered LAMP hosting for less than $10 per month, though, even that low amount of revenue could cover my costs.

Patterns

There doesn’t appear to be any meaningful pattern in some of the data that is available. For example, there is a seemingly regular rising and lowering in traffic levels, but it is out of synch with the cycle of days of the week. The most intense days were 2007-06-10, 2007-06-19 or 2007-06-27, depending on which measure of traffic you are using, and the most intense day of the week and hour of the day for visits were Wednesday and 08:00, however you measure them. The overwhelming majority of visitors were from the USA or EU, with the visitors from Australia (in third place) accounting for an order of magnitude less traffic.

Apparently, one host was particularly keen to access my blog, with over 1000 page views, the second most keen only accounting for about 300. The fan’s IP address is related to Panama, according to WHOIS, but this information is again not particularly useful or meaningful to me. Apart from this Panamanian, other avid visitors are robots, the three most bandwidth-intensive in descending order being the MSNBot, Yahoo Slurp, and then the Googlebot.

Profiling

Fortunately there is more good data to be found in the monthly summary. Firstly, although slightly to my chagrin, 89% of visitors stay for less than 30 seconds, although the average is apparently 165 seconds. Trawling the AWStats forum I found I was right to be sceptical about how these durations are measured, and with the average visit consisting of 2 pages (and thus perhaps 50% of visits involving just one page for which it is impossible to generate a visit duration) this stat could be severely under-counting long visits.

In apparent contradiction to the statistic that about 3000 of the 7200 hits were for an “HTML or XML static page”, the most popular two pages were for the root page of the blog and the wp-comments-post.php page (presumably popular with spammers) with 2600 and 2000 pages respectively. It appears, therefore, that URLs which don’t have a specific file extension at the end are treated as being static HTML by the software. The hit count for other pages is useful, though, because (after changing the permalink settings in WordPress) the stats list which of my posts are most popular. Unfortunately the URLs contain the post number (as I find post slugs to be an inelegant programming solution to generating URLs) and if I visit the pages to find out which posts those numbers refer to, I will change the stats. That means I have to use the administrative backend to do the conversion, which is a little cumbersome. The relevant piece of information, though, is that the post on “Logical fallacies” from two months prior was most popular in this period, with over 300 views.

I would love to give lengthy details about operating system and browser usage now, but with “Unknown” accounting for the second and third most popular choice respectively in those two lists, the statistical significance is lower than I would hope. Without trying to divide the unknowns between the remaining categories, I can only give the other percentages as lower bounds. For instance, at least 76% of visitors were running Windows and 8% Linux, which suggests a more Free readership than other technical sites. Strangely, though, the numbers next to the browser names were 54% for Internet Explorer and 19% for Opera, leaving Firefox down at 5%. To fill the gaps, one might suggest that this large number of unknown users were using Firefox on Windows, but they seem a strange set of people to be hiding their details and visiting my blog randomly.

Origin

Once overcoming the shock of how many people actually visit your blog, the writings of a complete stranger to most of them, there is an aftershock caused by discovering how people find themselves at your blog in the first place. In my case, for 94% of visitors there is no referer information, but there are about 190 visits that came through search engines, and 20 that came from other pages on the net. Unsurprisingly (although coincidentally), 94% of the visitors from search engines used Google, but the second most popular search engine was one I had never even heard of. I’ll mention the search terms these people were using below, but the links from other sites are just as interesting. Admittedly, many of them are friends’ websites or sites which should probably count as search engines, but there was one in there from a French site containing a small list of pages that dealt with ISO standards. I’m glad that aspect of my blog is being appreciated, especially across a language barrier.

As for those search terms, many found my blog because it contains the rather rare name of a piece of PHP graphing software (which I won’t state here, for fear of the duelling banjos effect). The term “PHP” itself was quite common in the searches used to find this site, too, as was “mplayer”, but the most popular single query was for the one word which appears after the tilde in this URL, which I’m sure people muse on when they read a random page from this blog. My curiosity is really piqued, though, by the list of one-off search queries which were used to find my blog. Some are unbelievable, either because I didn’t realise the sorts of queries these pages could go some way towards answering, or because the queries seem to bare no relation to my site. None has prompted me to change the focus of this blog or the subjects I write about; I only feel encouraged that by writing about what interests me, I have helped people on their various journeys.

A question remains, though, which this one month of stats can’t answer: Does traffic to your blog scale linearly with the number of posts?