I don't think you can capture every single page viewing, but I think you might be able to capture enough information to be worthwhile for analysis of website usage.
Create a database table with columns for: web page name, user name, start time, and end time.
On page load, INSERT a record into the table containing data for the first three fields. Return the ID of that record for future use.
On any navigation, UPDATE the record in the navigation event handler, using the ID returned earlier.
You will end up with a lot more records with start times than records with both start and end time. But, you can do these analyses from this simple data:
- You can count the number of visits to each page by counting start times.
- You can calculate the length of time the user displayed each page for the records that have both start and end time.
- If you have other information about users, such as roles or locations, you can do more analysis of page viewing. For example, if you know roles, you can see which roles use which pages the most.
It is possible that your data will be distorted by the fact that some pages are abandoned more often than others.
However, you certainly can try to capture this data and see how reasonable it appears. Sometimes in the real world, we have to make due with less than perfect information. But that may be enough.
Edit: Either of these approaches might meet your needs.
1) Here's the HTML portion of an Ajax solution. It's from this page, which has PHP code for recording the information in a text file -- easy enough to change to writing to a database if you wish.
<html>
<head>
<title>Duration Logging Demo</title>
<script type=”text/javascript”>
var oRequest;
var tstart = new Date();
// ooooo, ajax. ooooooo …
if(window.XMLHttpRequest)
oRequest = new XMLHttpRequest();
else if(window.ActiveXObject)
oRequest = new ActiveXObject(“Microsoft.XMLHTTP”);
function sendAReq(sendStr)
// a generic function to send away any data to the server
// specifically ‘logtimefile.php’ in this case
{
oRequest.open(“POST”, “logtimefile.php”, true); //this is where the stuff is going
oRequest.setRequestHeader(“Content-Type”, “application/x-www-form-urlencoded”);
oRequest.send(sendStr);
}
function calcTime()
{
var tend = new Date();
var totTime = (tend.getTime() – tstart.getTime())/1000;
msg = “[URL:" location.href "] Time Spent: ” totTime ” seconds”;
sendAReq(‘tmsg=’ msg);
}
</script>
</head>
<body onbeforeunload=”javascript:calcTime();”>
Hi, navigate away from this page or Refresh this page to find the time you spent seeing
this page in a log file in the server.
</body>
</html>
2) Another fellow proposes creating a timer in Page_Load. Write the initial database record at that point. Then, on the timer's Elapsed event, do an update of that record. Do a final update in onbeforeunload. Then, if for some reason you miss the very last onbeforeunload event, at least you will have recorded most of the time the user spent on the page (depending upon the timer Interval). Of course, this solution will be relatively resource-intensive if you update every second and have hundreds or thousands of concurrent users. So, you could make it configurable that this feature be turned on and off for the application.