Recently I discovered a great article from the de facto usability guru, Jakob Neilsen, entitled Traffic Log Patterns. The article consists of two parts, the first being traffic distribution on a website, the second being traffic distribution as related to search engine queries. In this post I’ll tackle some commentary on the first of those parts, traffic distribution on a website.
In Traffic Log Patterns Neilsen reveals the distribution of traffic on his Alert Box website, how the home page receives the majority of traffic while deeper pages in the site structure receive proportionally less traffic, but, that cumulatively the deeper pages receive more traffic than the most popular pages. Neilsen then contrasts his current Alert Box traffic distribution with his traffic distribution from 10 years ago and finds that, though the volume of traffic has grown, its distribution has remained consistent.
To demonstrate, Neilsen plots the distribution of traffic on a two-dimensional x-y graph and calls the resulting shape a Zipf distribution. Since I have no idea about Zipfs, I won’t argue with him. The Zipf distribution holds true for incoming traffic from external websites and for internal traffic on the Alert Box site. The Zipf distribution represents a consistent pattern of traffic on the web.
For me, the main value of the article is the graphs, as they provide a great visual representation of website traffic that’s usually seen in tabular form. If you’re looking for a way to understand and communicate what you’re seeing since you’ve been following my Quick Guide to Reading Webstats, Neilsen’s graphs and discussion of traffic distribution do just that.
The funny thing is Chris Anderson discovered the same distribution of data in a very similar web context a few years ago and gave it a catchy name: The Long Tail. The name spawned an article in Wired magazine and a newly released book of the same name. I recommend at least reading the article and getting up to date on the long tail blog Anderson keeps as well. I have yet to read the book, but if I find it worthwhile I’ll add to this post.