Enabling Customer Segmentation

pic
Share this article

So, following a recent “back to basics” post, a reader requested that I write something describing how custom traffic props are actually useful.  How about customer segmentation achieved through traffic props?

Custom traffic props are used for counting traffic-based activity, to gain insights into user behaviour across our sites.  By default, custom traffic props only come enabled with the metric Page Views, which, by itself is, admittedly, not a lot of good.  But, with a couple of minor customisations, you can get valuable insights from the props very quickly.

This post is actually about customer segmentation – in our case measuring student activity versus staff activity versus “anonymous” activity.  In your case, it’ll probably be customers vs. non-customers.  It’s about understanding what they do differently (not how they convert necessarily, because for us, they’ve already converted). In a retail, financial or ISP environment, it might be about how customers interact with your content, or your help site versus non-customers.  What do they do differently?  What can you do to improve their interactions?  How do they use search differently?  Do they spend more or less time with you?  Are there signals that are pre-cursors to abandonment?

You don’t have to stop at the customer vs. non-customer segments either.  There’s plenty of other customer segments that you can apply throughout the brand funnel, as they move from being just a browser to a high value loyal customer.

segmentation_digital_strategy

And each major segment can sub divide into sub-segments.  For example, the Loyal Customers can be High or Low value loyal customers, they could be in danger of abandoning you, they could be a brand advocate, highly active in social media.  If you can segment them and re-identify them when they return to your site, you can present them different offers and different content to move them to the next “stage”.

All you need to do is reset the value in the traffic prop each time they move to the next level.

Customer Segmentation

We have segmented our traffic by user type for about two years now, to understand the differences in our audience types across a number of key areas on our site, such as search terms, content, tools, average visit duration, key activities and so forth.

We ultimately re-designed our staff and student main pages, in part based on the information from the analytics.  We looked at:

  • key sites those segments were consistently going to (we have over 500 separate sites on the Murdoch domain) ;
  • key content within those sites;
  • search terms that consistently popped up
  • primary entry and exit pages, bounce rates, page bounce rates
  • key paths to content
  • average visit length
  • average time on page
  • average pages per visit
  • page velocity for each segment

We then built out a set of new pages designed for each audience, with the most common content and destinations specific to each audience type.  We’re about to expand upon this by launching transactional-based mobile site, designed for the iPhone, iPod and iPad series.

All of this came from using a single custom traffic prop – prop25 – User Type.

And some very interesting things came from those traffic numbers too.  Things like the volume of traffic from each segment:

staff_student_page_views

And the seasonality of searches, not to mention the differences in keywords (note the differences in terms, especially the seasonality spikes across the student based search terms below:

staff_student_search_terms

From the search terms alone, coupled with top destinations, we were able to put together a Search As You Type functionality across our site, that aids faster access to the most popular content destinations, based on the keyword, or partial keyword that is entered.

sayt

A couple of quick wins

Enable Pathing

The first thing to do is turn on Pathing for the custom traffic prop, via a request through live support.   Even though you may not actually use pathing on the prop, pathing enables the following additional metrics:

  • Average Page Depth
  • Average Time Spent on Page
  • Entries
  • Exits
  • Reloads
  • Single Access
  • Visits

You may also (probably should) request that you get Unique Visitors turned on for this prop – but you’ll have to do that through you account manager.

Set up correlations

The second thing to do is set up some correlations between this prop and others, for instance:

  • User Type and Search Term
  • User Type and Site Section
  • User Type and PageName

Remember: correlations only work when the two props are set at the same time and passed into SiteCatalyst at the same time.  If they are passed at different times, your correlation won’t work.  So, you need to ensure that you pass in the User Type on each request, which is easily accomplished through the s_code.

So, now to the nitty gritty part

Firstly you’ll need to ensure that you have the following plugin:

  • getAndPersistValue

You can download this from the help section of the admin.  Follow the instructions to place this in your s_code.

You’ll also need an eVar, which shouldn’t be a problem as we now have 75 of them, courtesy of SiteCatalyst v14.9.  Note that if you’re intending on using props or eVars that are numbered higher than 50 (the old amount), then you’ll need to update your s_code to version H22 or higher.

The first thing to do is locate specific areas of your site that you can identify a user type.  This might be on the logged in page, or it might be a specific page of content only for a specific type of user.  In our case, we set prop25 (User Type) when the user either:

  1. Logs into our portal (either staff portal or student portal)
  2. Views either the Staff homepage or the Student homepage

We have a little piece of code that sets the eVar depending upon the content they view:

/* Set User Type */
userloc = location.href.toLowerCase();
if(userloc.indexOf("my.murdoch.edu.au/staff") != -1){
s.eVar25="Staff";
}else if(userloc.indexOf("murdoch.edu.au/staff/") != -1){
s.eVar25="Staff";
}else if(userloc.indexOf("murdoch.edu.au/students/") != -1){
s.eVar25="Student";
}else if(userloc.indexOf("my.murdoch.edu.au/students") != -1){
s.eVar25="Student";
}else{}

So now we have set the eVar25 (a conversion variable for User Type), we need to set s.prop25 to the same value and pass it on every request.

/* Set User Status*/
//Check existence of persisting prop25
s.prop25=s.getAndPersistValue(s.eVar25,'s_prop25',7);
if(!s.prop25&&!s.eVar25) {
s.prop25="Anonymous";
s.eVar25=s.prop25;
}
//Set persisting prop25 to value of the eVar25.
s.prop25=s.getAndPersistValue(s.eVar25,'s_prop25',7);

The getAndPersistValue plugin takes 3 parameters:

  1. The name of an eVar to get the value from
  2. The name of a cookie to store the value in
  3. The duration of time in days to store the value

So in the above code, we’re first putting the value of eVar25 into a cookie s_prop25 and storing it for 7 days.

Then, if the neither prop25 and eVar25 exist (in other words we’ve not identified them yet), then we set prop25 and eVar25 to “Anonymous”.

Then we set prop25 to the value of eVar25, store it in a cookie called s_prop25 and save it for 7 days.

Each time the s_code executes, it will now pass in prop25 at the same time it passed in things like keywords (prop5) and pagename and site section etc, enabling us to do correlations on it.

Additionally, as we have pathing enabled on prop25, we have Visits, Time Spent and Average Page Depth for each User Type.

Some additional valuable insights from traffic props

Another interesting thing to do with custom traffic props is to concatenate them with other traffic props and enable pathing on the resulting traffic prop.

For example, if you concatenate your Campaign codes with pageName, into a new traffic prop, then you can see what paths users take when they interact with the campaign, likewise you’d be able to see bounce rate by segment and so forth.

/* External Campaign Pathing */
if(!s.campaign){
s.prop26=s.pageName;
}else{
s.prop26=s.campaign+"|"+s.pageName;
}

The above will use another traffic prop and concatenate the campaign tracking code, if it exists, to the page name.  For more info on this one, check out my campaign pathing post.

Another insight you can gain is page scrolling, either at the unsegmented level, or by each segment.

And then there is form analysis and abandonment – all achieved through your custom traffic props – but I’m saving a post on form analysis for a later date, possibly the next one!

Final Thoughts

Despite traffic props being basic counters, there’s quite a few imaginative things that you can do with them to get further value and insights.  And customer segmentation is key to understanding your audience online and improving conversions.


Share this article