Friday, February 02, 2007

WA: Jim Burchell, Omniture

Jim Burchell of Omniture came and spoke to us a little about the history of web analytics, and then spent the remainder of the time answering any questions we had. He's worked for three web analytic companies over the last 6 or 7 years: WebSideStory, WebTrends, and currently Omniture.

We discussed the difference between analytics that analyze the web server's log files (analyzing all the HTTP requests) and putting JavaScript on your web page. Log files are less accurate, they tend to either inflate or deflate certain numbers. For example users whose ISP is AOL tend to be aggregated behind IP addresses. With only examining the log files, that one IP address will only appear as one user to the web server, and will only count as one unique visitor, greatly deflating the actual number.

With the JavaScript technique, page views are counted a lot more accurately, as they can actually count each time a page is viewed. On the other hand, the log files have a harder time distinguishing between page hits and page views, and what should actually be counted as a page view, the numbers tend to be a lot higher than normal. This particular problem causes problems for companies like Omniture, because when they institute their method, it appears that their client company's web traffic is actually decreasing.

A question was posed on the difference between Google analytics and what Omniture does, while this may seem a naïve question to ask, I think Jim answered it very well; I don't doubt that any one in the class had a clue as to the answer. He said Omniture provides depth. For example, take someone in Oregon who looked at some brown shoes on Nike's website, but didn't end up buying them. Omniture could tell you the different regions where certain people looked at a certain product as well as their age, which provides the company with very accurate data with which to remarket to their customers, making their marketing endeavors more successful.

No comments: