search analysis icon Web search analysis

Our "search our web pages" feature logs information about the people who use it and what they searched for. The data collected are the date and time, the remote host, operating system and browser information, which page was searched from, what word or phrase was searched for, and the number of matches. The remote host is normally the web address of the user's proxy. The operating system and browser ID are provided as part of the dialogue between the user's browser and the web server which is responding to the search request.

We have accumulated enough records in the last few years to do a simple analysis of these results. After removing our test records and a few other spurious lines, the search log has 222 records. For simplicity, where a single user did multiple searches, these are analysed as independent records.

Operating system used: This plot and the next one indicate that Microsoft has been allowed to take over the "advanced" world. Us Mac users can only look on in wonder as you all complain about how many millions of dollars the latest virus or worm has cost industry (mind you the figures look a bit rubbery, since they depend on input from IT departments and accountants), and how much was really spent on the "millenium bug", and wonder why no one seems to blame Microsoft, let alone punish them by buying someone else's software. operating system comparison

browser comparison Browser used: See rant above, although we are aware that a number of browsers can be configured to report themselves as Internet Explorer. We suspect that almost all PC users just use the browser (IE) which came with their machine, and do not make an active choice. As web developers we have all the major browsers on Macs and PCs, and try to test our pages on all of them to make sure that the "browsing experience" is satisfactory for as many people as possible. Our "default" browser is Safari, which comes pre-loaded with Macs running X. Despite only being version 1.0, it is relatively stable and handles standards-compliant pages very well. In addition it has several useful features for the professional - for example "empty the cache" is directly available as a menu item and the browser opens an 'activity' window which shows all the files requested and allows individual frames, pictures, etc. to be easily loaded in a new window. Firefox is also very useful, mostly because of the numerous plug-ins which it supports, especially HTML code validation and broken link checking.

User Domain: Since we are Australian, no great surprise here - just under half the searches were from Australian domains. Given that a few Australian surfers will have a .com domain, the real Australian/international split is probably very close to 50/50 domain comparison

search source comparison Search source: If a search is carried out from the main page, it looks at the entire site, but there are also options to search just the major sections of the site. The food pages are probably the most visited, and it is clear that many more searches are initiated from within these pages. Just to show that we care, after we saw that someone had unsuccessfully looked for 'conversions', we created a cooking units conversion page. The travel pages are also well "patronised" but they provoke less searching.

Search string entered: There are always a few people who will push any button they find on a web page (or anywhere else). In fact very few did this on our pages as can be seen by the small 'empty' slice. But the truly amazing finding is the number of people looking for ways to cook crayfish (including a couple who obviously tried but could not spell the word, or maybe just could not type it). No other search string came close to the number of 'crayfish' searches. Sadly, we have no crayfish recipes, but obviously we will have to try to find some to satisfy this huge unmet demand. There were no searches for 'sex' - reputedly the commonest search string in major search engines - but if anyone tries it they will probably not get a lot of satisfaction from our pages. Strange food searches included 'hedgehog' and 'nikkujaga' - hopefully the former does not refer to the little flea-ridden spiky animal, and a Google search of the net returned no matches for 'nikkujaga' - is there such a thing? string entered comparison

match success frequency Success rate: Around half the searches failed to find any pages containing the search string. While this was often because no such page existed, on many searches it was clear that a bit of spelling and/or typing accuracy, or a more foccussed search approach, might have produced a better result. For example, some strings entered included 'loquate', 'SAHARA DESERT' (from the food page), 'shakespeares steak pie', 'japanes food', 'am in Ghana ,want any nearest mexican embassy', 'dessurts', 'Caryfish' and 'calafornian'.