Last week's release was based on weekly numbers of US visits to google.com vs. facebook.com, expressed as percentages of the total number US visits to websites. Are visits the best way to measure popularity of a website? Yes and no. Analysts love visits because they are an easy way to measure using click stream data, but unfortunately the definition of a visit is somewhat arbitrary.
It is unclear from the article how exactly a visit is defined, but the Web Analytics Association defines a visit as
"an interaction, by an individual, with a web site consisting of one or more requests for a page. If an individual has not taken another action (typically additional page views) on the site within a specified time period, the visit will terminate by timing out."The standard timeout for visits is 30 minutes. This means that if I refresh my Facebook News Feed every 15 minutes for 8 hours, I've made one visit to Facebook that day. If, however, I refresh once an hour for 8 hours, that's eight visits because the time between refreshes is greater than 30 minutes. So, if you only look at visits to measure popularity, you would say a site that people check frequently throughout the day (a single day long visit per person) is less popular than one people visit sporadically during the day (many short visits per person). Do people visit Google and Facebook in the same way?
At Alexa, we take a broader approach in determining popularity as measured by the Alexa Traffic Rank. The Alexa Traffic Rank of a site is based on the global number of unique visitors to a website, and the number of unique pages seen by each visitor. Visits, due to the vagaries in the definition, aren't even part of the equation. The Alexa Traffic Rank is also a 3-month trailing average, so for at least the top 100,000 sites small changes on shorter timescales are averaged out. By construction, the Alexa Traffic Rank measures popularity based on the number of people who see the content on a site, and how much unique content they view, measured over the past three months.
Obviously, the Alexa Traffic Rank is very different way to measure popularity than the number of times people have visited a site in the past week. But even subtle differences in how something is measured, or the interval over which it is measured, can alter the results. For example, whether or not to include AJAX, Flash, or other Rich Internet Applications when defining a visit can significantly change the final number. This is an important thing to remember when comparing data across different services. If you do not understand what you are looking at, then you can easily be led to wrong conclusions.
So, which is a better measure of popularity, visits or the Alexa Traffic Rank? Well, the Alexa Traffic Rank is an excellent indicator of popularity because it reflects the number of people visiting a site and the amount of content they are exposed to. But I wouldn't choose a single number, there are a lot of other things to consider when comparing the success and popularity of websites.
Pageviews can give you a clue as to how people are interacting with a site, as can bounce rate, time on site, etc. Also, audience profiles, upstream sites, and percentage of traffic from search, and downstream sites can tell you who the visitors are, what routes they are taking to find the site, and how well they are being served. But even that isn't enough. The objectives of sites vary significantly, so what are good numbers for one site might not be good for another. For example, bounce rate is generally uninteresting for a blog because people tend to either check the home page or find content via search engines. For a retail site, however, a high bounce rate might be very telling. On that same retail site, a large number of page views might indicate people are highly engaged and doing research on products, or that they are having problems with navigation and are clicking around blindly before leaving. So to truly measure the success of a website, you need to understand what the goals of the site are and then identify the metrics that make sense for measuring those goals.
In case you haven't checked already, the Alexa Traffic Rank for Google is still #1 and Facebook is still #2. Which site do you think is the most popular? How would you measure it?