Visit/Session
From WikiWebAnalytics
Contents |
[edit] What is a visit?
Short version: A visit is a time-grouped series of page views made by a single user.
Explanation: A visit or session represents a single period of a users interaction with a web site. Since these visits or sessions are really created by a web analytics tool, there is an arbitrarily defined period of inactivity, often a half hour, that is used to signal the end of a session.
Analogy: Someone comes to your house: They look through your CD collection (how old school), read your Web Analytics Books, go to the toilet and eat the dinner you cooked them. Each of those activities is a distinct event but the whole you call a visit.
[edit] Why are they useful?
Visits are fairly crude in that they just show you how many people are turning up, not what they are doing once they reach your site. As such they form the basis of conversion rates (X turned up but only Y bought anything). However they are also the most easily understood metric, and in a corporate environment represent the limit of what the average non-web manager will tolerate in terms of 'tech-speak'. Thus there is a tendency to fixate on this metric; its the analysts job to lead the business users into deeper insights.
[edit] Visits over time
Even the most basic web analytics tool will enable tracking visits over time. Watching your visits over a period of weeks or months will tend to show a series of cycles or trends:
- Hours of the day: Usage will peak at different times of the day depending on what your site is doing and it's geography.
- Days of the week: Likewise with days - you will find your site has a rhythm.
- Seasonal shifts: Site visits will go up as deadlines approach, Xmas presents and bought and so on.
Once you are used to tracking these cycles the thing to look for is the exception, and part of the fun is trying to track down the sudden surge/drop: Did your site get on Slashdot and experience the 'Slashdot effect' and then fall over? Was there a major news item that pulled people away? Or did you finally get something right on your site!
[edit] Issues
[edit] Accuracy
There are several ways that visits can be inaccurate, but the main reason is one of arbitrary session expirations:
- If a site has it's session time-out at 60 minutes and a user leaves their browser on a page and then comes back and clicks on it 61 minutes later, it will count as two visits.
So there is an (historical) assumption that 30-60 minutes is as long as your user will be interested in your site. With the advent of the 'attention economy', tabbed browsing and keeping social media sites open all day long, this may not be the case for your user group. In the second scenario this inaccuracy will be mitigated by having users explicitly logged in, in which case you can make (further) assumptions about what constitutes a new session. As with all stats it helps to know what your Key Performance Indicators (KPIs) are. For instance your site might be a family blog, so several visits a day might be satisfactory as each of those visits will have a high value; but several visits a day for an ecommerce site is not enough.
[edit] Definition of visit
Please note that the exact definition of a visits duration will change depending on the stats service or software you use: You should know yours.
Many web analytics tools allow you to modify the visit definition.
[edit] History
The first work at determining the right threshold for session duration was Lara D. Catledge and James E. Pitkow, "Characterizing Browsing Strategies in the World Wide Web." In Computer Networks and ISDN Systems 27. Elsevier Science, 1995. The methodology used is quite interesting, see the cite seer trace and the original paper. The conclusion was 25.5 minutes.
