Web Server Logs and 2007
It’s that time of the year again when I disclose what I’m doing with the data from my web server logs. Yes, your user data. Based on the number of people who read Privacy Policies, I sometimes wonder if this is needed. I sweetened the pot late last year and offered a gift, but that didn’t help.
After the AOL data fiasco of last August, I thought there would be more interest user privacy. I put in a line in our Privacy Policy that offered an Amazon gift certificate to the first person who read the entire document. How many people do you think responded? As I’m going through the logs for February, I can see that 32 people read the page. Perhaps, these are people looking for a new offer from me. If I go through the logs and trace back the IP addresses, I wouldn’t be surprised to see these linked to search engine spiders and not real people.
What’s Changed over the Year
I can’t say much is new in terms of your data. Certainly, there are more of you than last year. We’ll take that as a good sign. Some percentages have changed a bit. If you’re a fan of Alexa rankings, we’ve improved. We don’t give much weighting to these online ranking sites as we find most of the stats are off the mark when it comes to our site. (Sorry Geoff M., I know you’re a good person.) As example, Alexa has US visitors at about 48% yet our logs show between 65% and 68%. The only time it dropped that low was when we got some coverage in a Japanese IT journal.
The biggest change has come in the tools used to analyze your data. You’ve probably gathered that I like to try different tools. Currently, I’m testing three new ones.
ClickTracks Optimizer – We’ve used the free version called Appetizer for a while and like it. We’re using Optimizer as we’re evaluating it for another company. It’s always easier to evaluate products with data you know. It does an outstanding job of letting me see where people are clicking on a page. Using the tool, I could see that people weren’t using the Tag Cloud on the top navigation bar and I decided to pull it. This site overlay is a feature I could never get to work in Google Analytics.
The program also allows me to segment users. As example, I could create a segment for all users that were referred from Google. I could even create a segment based on the 32 people that visited the privacy page. The program has an amazing number of reports, but it could also be one of those tools where you get so deep you lose track of your initial goal. As example, why do people from the United Kingdom spend longer on the site than the average user? Or why do men read more about Firefox, but women read more about Cornell notes? OK, I made the last one up as nothing in the data logs suggests sex. The best I can tell you on that subject is what Microsoft adCenter labs reports about our site. They say 59% of you are males. I have no idea.
103 Bees – This is an interesting service with some nice features. One reason I look at logs is to see what questions people asked the search engines. Sometimes I’ve covered the subject. Other times you realize this might be a topic I should consider. 103 Bees makes it easy to spot the questions people posed to search engines that resulted in the user viewing your site. I can quickly look at this list and say, yes I did an article on that subject. Now, if I could only figure out who asked the question about free Vista books, as I might want one. Of course, I can’t tell who asked the question.
This system works by adding a snippet of JavaScript code to my template. Like Google Analytics, if you have JavaScript turned off, I don’t get any data. I also played with some other products where users could ask a question from your site. I installed one system on our test site but found the system too invasive. I also had a hard time unsubscribing from the emails so decided to yank it.
HitTail – You’ve probably heard the expression, “the long tale”. It was popularized several years back by Chris Anderson from Wired magazine. He wrote a book by the same name that is a great read. This service identifies your tail and offers suggestions for articles. To put this in perspective, the service has identified 22,750 unique keywords on this site. I didn’t even know I knew that many words. Of those, the top 10 keywords make up 10.4% of my search traffic. I would show this graphically, but it doesn’t fit. The top entry is “update firefox”. If you were to type that entry into Google, you would probably see my article appear on the first page.
Although showing on the first page has its rewards, the main benefit is again seeing what queries people are posing to search engines for future articles. The screen snap below shows some recent queries. The bold items represent what a reader typed into the search engine query box whether it was at Google, Yahoo! or some other search engine. They then saw one of our pages in the search results and clicked the link. I can find this data in the web logs, but this service makes it easier to read. The items that are gray show links from referring sites that aren’t search engines. As example, you can see several referrals from the Lifehacker site about two articles on our site. As an aside, Lifehacker is a site you should check out.
One of the nice features of the service is I can click an entry and duplicate the search results. For example, I was perplexed by the query for free product key for Microsoft 2007. I knew I didn’t have articles that provided keys. It turns out, it was the article I did on Microsoft Office Trial. It’s also interesting to see the differences in queries based on the user’s location. Even though English is a common denominator, there are differences between countries and the words used.
Ads and Affiliates
There was some news in this area although it may have gone unnoticed. I was annoyed with various ad models on the site and pulled off the code serving ads. I outlined my thinking and asked people for their opinions. It was clear that people didn’t mind the ads. I took the middle ground and pulled the code that embedded the ads in the article. You’ll notice the ads are on the sides, which is where we started.
I also experimented with other ad models. Most weren’t ready for prime time in my opinion or confused the user such as systems that show ads for specific underlined words. I thought this might be confusing for some people and I didn’t find the ads relevant to the content. I also tried several affiliate offerings where I liked the products. These items also appeared on the side. The revenue returns weren’t enough to justify the space.
The current test is with Google Image Ads. I have to say some of these I like and they fit well with the content. Others make me laugh where the ad keywords overlap with a keyword in the article but are used in different contexts. Sorry, but this out of my control. As example web logs and log stairs. Yes, in an ideal world, I would love to have a log cabin overlooking some waterfront or mountain property where I could write and review my logs. Just goes to show you no system is perfect unless you go out and dedicate yourself to contacting companies for ad space.
How Much Data?
One of the questions you might be asking is whether these tools overlap and if they are all needed. The first answer is easy as there is overlap. Even with overlap, the different tools can show different results or from a different perspective. I’m sure this time next year; I won’t have all of them. That’s all part of the testing process. Another factor is cost in terms of money or time. Some of these tools are in beta and when they go live and announce their pricing structure, I may opt not to use them. Equally important is the cost they have on the system. Most of the online tools also come with a performance hit. They need me to add several lines of code to the template for them to work. Depending on various factors, this may result in slower response time for a viewer. If I notice a service taking too long, I generally pull it so it won’t lead to a bad user experience. The other point to stress is none of these tools truly represents you. At best, these tools try to quantify certain attributes about you.
Related Privacy Articles
Last Updated (Sunday, 30 September 2012 14:19)