What is a visit?

The topic on what constitutes a single unique visit has been discussed so many times but until today the debate is still ongoing. A journalist once called me up to clarify the idea of a unique visitors for a story he’s researching on and all of the people he’d asked told him a slightly different definition. I added that the whole “site visit” definition is dependent on the tool that’s measuring it.

Over at the Alexa Weblog, Geoffrey Mack tackled the topic on various angles:

First, let’s tackle the biggest elephant in the room, what is a visit? In the simplest terms, a visit is when an individual person visits your site. But beyond that simple definition there is no agreement as to what a visit actually means.

Consider:

Cookies. Some sites use cookies. Some sites don’t. Cookies help identify when the same user comes back on the same computer with the same browser. That’s good. But sometimes people use multiple browsers and multiple computers… I know I do. Cookies don’t help there.

Log-ins. Some sites require their visitors to log-in. This is a great way to track visitors, but few sites actually require users to log in each time they visit.

Repeat visits. What if the person goes away then comes back? Is that 1 visit or 2? What if the person gets up for a cup of coffee and comes back 15 minutes later? What if that person comes back 5 times in a day? There is little agreement as to what constitutes a visit in these cases. How does your log program count them?

What if that person comes back with a different IP address or logs in as a different user?

Crawlers? The Web is littered with crawlers. Sometimes the vast majority of pageviews on Alexa come from crawlers. Does your log program recognize them and remove them?
Raw Logs vs. Web Bugs. Do you analyze your raw Web Server logs or do you have some embedded javascript bug on your page that logs the visits with a 3rd party service? The differences between those two methods can be vast. If you are curious, try both and you’ll see.

Fraud. It is easy to create fraudulent visits and pageviews in logs. Unfortunately, it can also be very profitable — planning to sell a domain or trying to increase click-throughs? Just start clicking around. Clear your cookies and do it again. Go to a different computer and do it again. It ALL shows up in the logs.

This is just the tip of the iceberg. The point is simply this. A visit is NOT a visit, even if you are using the same log analysis program. You can’t reliably detect fraud or crawlers or many of the other factors mentioned above that have a drastic impact on your reported visitor number.

While some use their internal stats log like AWStats (too bloated) and others use 3rd-party analytics (Statscounter, Google Analytics), the huge numerical gap between these tools is the reaosn why Alexa is trying to independently measure relative traffic rankings.

Still, I can’t shake the fact that my blog which generates a couple hundred thousand pageviews a month has a higher Alexa ranking that a forum I also own and churning out close to 6 million pageviews a month.