Unique Visits and Visitors
Note: See terminology for definitions of common terms.
The term visit implies that someone arrives at a website, stays a while, then leaves. While this is a nice way to describe the situation, it is not quite technically accurate. For most types of web surfing a visit is made up of a series of individual file requests (hits). Each time a client requests a file, it is a completely separate event.
To calculate the number of visits, the website's host server must utilize server log files. Log files contain all the information about every single hit on the website. In order to extract the number of individual visitors from this information, webmasters use a log analyzer.
A log analyzer will look at a log file and identify all the hits which came from the same IP address. If these hits occur regularly within a short period of time, the analyzer assumes it is the same person. When there has been no activity from that IP address for a certain period of time, the analyzer assumes the person has left (and records the last page hit as the exit page). Any new hits from the same IP address will count as a new visit.
There are a number of potential problems associated with calculating visits this way:
- If a web page takes a long time for the visitor to read, the analyzer may think they are a new visitor when they finally finish the page and request a new one.
- Some people will download many pages at once (especially those who like tabbed browsing) and read them over a period of time. This can mean that they don't generate any new hits for so long that the analyzer assumes they have left. Then when they do access a new page they are counted as a new visit. If they don't request a new page, the length of their visit will be underestimated.
- Some ISPs and anonymizer services give the appearance that lots of people have the same IP address. Log analysers will count all these people as the same visitor.
- If a website's audience is mostly from the same area or uses the same ISP, there is a greater chance of confusing visitors.
Despite these problems, estimating visits does work surprisingly well on the whole. It's just important to understand that it is an estimate and certain factors may influence the statistics.
Most log analysers can be configured to suit different websites. For example, you can set the amount of time before a new hit from the same IP address is counted as a new visit.
Unique visitor means a single person who visits a website any number of times. For example, one unique visitor might make several unique visits to a website over a number of days.
Calculating unique visitors generally means counting the number of unique IP addresses. The analyzer assumes that these are separate people.
Calculating unique visitors has two main problems:
- In many cases the same visitor has lots of different IP addresses over time.
- In some cases multiple visitors appear to have the same IP address.
Here's an example: For five years I operated a website targeting a New Zealand audience. At the time there was one major ISP which had a very dominant national market share. This ISP used a system which made all their customers appear to have the same IP address, which meant that a significant portion of the site's visits appeared to be from the same person. However I knew from our user login records that there were several thousand unique visitors per day.
Other Calculation Methods
There are various other ways to track visits and visitors, for example:
- If a website requires users to login, they can count login sessions.
- Cookies can be used to track visits.
- Some types of web services create a "persistent connection", where the client and server are in constant contact.
There is no way to measure visits or visitors with 100% accuracy, but as long as you understand the limitations it is possible to make useful estimates.