Yesterday, our sister-company's MD dropped by with a question. He wants to make his website sound good, and he was wondering whether quoting hits rather than visits was reasonable. I thought a little, looked a couple of things up, and decided that provided he doesn't describe them as anything else, quoting hits was fair; anyone with any right to be judging would know how poor an indicator that was, and ask for something else (at which point they'd be suspicious of the fact that he's too embarrassed to give figures for page views and sessions, even though he clearly has them).

Today, our Accounts Manager asked how his website was going. He quoted his hits figure, and she was impressed; I disappeared ominously behind my cubicle wall, doubled up in laughter as I was.

So I had a quick look at the logs of my own sites. I don't normally bother analysing them: the forums are successful if the users are happy, the blog is successful if ... well I'm not entirely sure when the blog would be successful ... and so on. The figure I managed to dredge up was that in the last 30 hours my blog had about 4% of his total lifetime hits. Since many of those are from my moblog script, I used this as an example of how unreliable the statistic was.

Turns out I was looking at the wrong logs. I have two instances of Apache running: the front one serves 'light' material and proxies heavier requests back to the other. One instance logs all requests, while the other only finds out about requests for PHP or CGI scripts.

In a good fortnight a week or two ago my larger forum had 80% of his lifetime hits. Hits are still a useless measure, but it's nice to come out on top every once in a while. It won't last if he keeps gaining users like he is, but if need be I can always rig something with artificially high hits, since that's what he wants to compare.