I remember the first time I saw Omniture ClickMap place overlays over the links, buttons, and images on a page. It was impressive; even to a web analytics newbie, it was clear that the ability to see the most effective areas of a page in terms of clicks and other metrics so visually can bring obvious value. However, I also remember thinking that I had no idea how ClickMap does what it does. It literally seemed like magic. Unlike the data available in the SiteCatalyst reporting interface, which can be seen and easily deciphered and counted coming through in image requests as pages are viewed, the ClickMap data collection and reporting functionality was—and is, to many users—completely mysterious.
My goal in this post is to explain some of the intricacies of ClickMap data collection and reporting, in so doing, I will also offer a few tips to improve (or fix) ClickMap tracking on your site.
At a relatively high level, ClickMap does something like this:
When the page loads, the
s_gi() function is called. That function attaches an onclick event handler to the document.body object so that ClickMap can track any page element that is clicked. The function called when a link is clicked writes data about that link to a cookie (s_sq), which the SiteCatalyst code then reads on the next page view (or image request) going to the same report suite as the page containing the link in question; this data is sent as part of that image request.
(Note that custom links/file downloads/exit links will pass ClickMap data immediately; they do not require a subsequent page load.)
The object ID will either be taken from the href property of the element, or from the onclick event handler if an
s_objectID value is specified there. If the former, note that the entire value of the href property is used as the object ID. (See below for more information on this!) The objectID is written to the s_sq cookie, and is read and passed on the next image request as described above.
The page name (or URL, if the page name is not defined) of the page containing the given link is passed in the “pid=” parameter of the image request. ClickMap uses the page name to locate the data regarding links on the page; thus, if the page name changes—either permanently or subsequent to the image request on the page—ClickMap may be unable to look up the data collected in order to build the overlay on the page.
To determine the object location, SiteCatalyst code uses the
document.all() method to assign a numerical ID to each distinct element on the page. This numerical ID is passed in the “oi=” parameter on the image request. When building overlays, ClickMap searches the page for the numerical ID stored in the data for the given page.
How to improve ClickMap tracking/reliability using object IDs
If you’ve ever had concerns about your ClickMap data—Is ClickMap tracking all of my links correctly? Why do certain prominent links not display many clicks?—you’ve probably heard about the
s_objectID variable, which is implemented in the
onclick event handler of your anchor tags and other clickable elements. Omniture strongly recommends implementing this variable, set to a unique ID for each link on the pages of your site, whenever links appear not to be tracked correctly. Why? Because it gives you complete control over the granularity of links on the page.
Here’s an example to help explain what I mean. ClickMap struggles with links that only differ in their query strings; this typically happens on sites that use templates to generate pages dynamically, with the values in the query string determining the content on the page (such as a generic product detail page where value in the query string tells your site which product to display). Thus, you might have several links on the same page that all point roughly to the same place:
Based on the explanation of basic ClickMap functionality given above, we know that the code will try to use “http://www.yoursite.com/mypage.php?content_id=123456″ as the object ID. The problem is that the query string will ultimately be stripped out by default during data processing. (This setting can be changed, but this also affects other aspects of data processing.) The result is that, when displaying data, ClickMap is left trying to figure out which “http://www.yoursite.com/mypage.php” link is which.
Of course, this isn’t a problem if the
s_objectID variable is implemented correctly; it allows you to tell ClickMap that links are distinct. You can use any method to generate
s_objectID values, and a unique value should ideally be applied to each link on each page of your site, although these values should be the same across distinct page views, visits, and visitors (i.e., they should not change from one page view to the next, or from one visit to the next, etc.). The implementation might look something like this:
<a href=”http://www.yoursite.com/mypage.php?content_id=123456″ onclick=”var s_objectID=’mypage_link1′;”>
The result is that when the ClickMap plug-in tries to place overlays, it has a much easier time differentiating links and attributing clicks correctly.
There is one more HUGE advantage to using the
s_objectID variable, which is that it helps ClickMap work around changes to page content and layout that would otherwise impact the tool’s ability to overlay links correctly. As mentioned above, an important element of ClickMap data is the link location. I have seen dozens of cases where a change to page layout has altered a link’s location enough that ClickMap struggled to find it when displaying data.
Even seemingly minor differences between the way a page looked when ClickMap data was collected and the way it looks when data is displayed can alter the number of clicks that are shown, because ClickMap thinks in terms of the page element index returned by the
document.all() method, and, particularly as designers, we typically don’t. We may add five seemingly innocuous links to a left navigation menu, thinking that all we did was consume what had been white space. However, in the world of ClickMap, what we have actually done is bump everything else in the code down by five elements. The page layout may be essentially unchanged, but ClickMap is impacted nonetheless. Using the
s_objectID variable helps to overcome this by telling ClickMap exactly what link to overlay with any given data, (almost) regardless of where it may now exist on the page.
(Closely related to this is the fact that
s_objectID helps in resolving ClickMap discrepancies across browsers. For example, ClickMap does not have access to the element locations in Firefox as it does in IE. While it is still able to track data in Firefox, it is not uncommon to see discrepancies when comparing against clicks reported by the IE version of ClickMap. Using s_objectID overcomes this just as it helps to overcome changes in page content: by removing the need for ClickMap to figure out the element location on its own.)
How changing page names affects ClickMap
As I described above, ClickMap uses the page name to organize link clicks, so that it doesn’t confuse the data for two identical links that occur on separate pages; the current
s.pageName value is passed in the pid= parameter on the image request. When you visit your site and run the ClickMap plug-in, it first checks the
s.pageName value on the page, then uses this page name to look up the click data. However, because ClickMap data for a given link click in passed after the page loads, it is possible for the
s.pageName to have changed between the time that the page loaded (and the page view data was passed into SiteCatalyst) and the time that the link was clicked. The result is that ClickMap data will be associated with a page name that is different from the one that the plug-in will detect when you view the page, making it impossible for ClickMap to display data for the page.
This is particularly common on sites that set the
s.pageName variable within the
s_doPlugins function in the s_code.js file. This function is called when the page loads, but it is also called whenever a link is clicked (in order to set the ClickMap cookie). Let’s say you have the following code within
s.pageName=s.pageName+" : "+s.prop7
Now, assume that on your home page,
s.pageName originally receives a value of “Home Page,” and that
s.prop7 stores the two-character country code of the home page version that is being served up. On the page load, the final value of
s.pageName might be “Home Page : UK.” But when the link is clicked, and
s_doPlugins is called again, it would change to “Home Page : UK : UK.” The link click would be associated to this latter value, but the ClickMap plug-in would see the former value, and would therefore be unable to tie the actual link clicks to the page.
There are a few ways around it. Certainly the best is not to set the s.pageName variable within the
s_doPlugins function. I’ve never seen a case where a user actually wanted the
s.pageName variable to change when a link was clicked, so setting the variable outside of
s_doPlugins is preferable. Another option would be to save a copy of the page in question as “web page, complete” using your web browser and then hard-code the
s.pageName to the altered value (e.g., “Home Page : UK : UK”), and run the ClickMap plug-in on the saved page with the altered page name, so that the tool is able to find the link data that was tied to the altered page name.
Segmenting ClickMap data
This isn’t an implementation tip, but it’s an important feature nonetheless. What many users may not know is that ClickMap data can be segmented using ASI, which Adam Greco described here. When you set up an ASI segment (e.g., “visits where visit number is greater than one,” to see return visits in their own report suite), the ClickMap data is included.
Using the example just given, if I set up an ASI slot to view return visits in their own report suite, I would then be able to visit my site using ClickMap and select the ASI slot from the “Site/Segment” drop-down menu to see click data for return visits only.
Hopefully, this information will help you developers understand ClickMap “under the hood” so that when you are planning a new or upgraded implementation, you have a context for understanding what you’re seeing (or will see) in ClickMap, as well as a framework for troubleshooting issues that may arise. ClickMap is a giant topic, and I’m sure this won’t be my last post on this product. Still, as always, please don’t hesitate to leave comments, ping me on Twitter (OmnitureCare), or e-mail me (omniturecare at omniture dot com) with any questions!