Wow, what a cool project.  Process improvement project like this one can be very satisfying and fulfilling to work on.

The first thing you should understand is that you can't improve a process if you don't have a starting set of metrics and data to start from.  What might these metrics be?

• Average call time/month
  
• Min call time/month
  
• Max call time/month
  
• 90 Percentile time/month (the time which 90% of the call fall at or below in length)
  
• Ranking of call complexity (how many call fall into easy, medium, or difficult to solve)
  
• Call center staff experience level (for example, less than 1 yr, 1-3 yr, 3 or more years)
  
• Categorization of call topics

This is just a small sample of the type of metrics you need to be able to measure and capture.  If your organization/group is not capturing this data today, then your starting point is to implement measurement processes and reports to begin capturing this. 

Once you can measure metrics like these you can analyze the data for certain repeating patterns and trends.  Only then will you be able to make recommendations about process changes that will have a sizable and meaningful impact.

If you group doesn't capture this type of data, then really you just floundering around trying to come up with suggestion for process improvement.

Are there other ways to come up with process improvement suggestions without this data.  Yes, but the techniques are inferior in comparison.  They tend to center around one time sampling techniques.  The goal is to get a large enough sample set that you can make reasonable conclusions about the information you receive.  You might:

• Interview call center staff and ask them what types of problem calls they encounter most often
  
• Watch and listen as call center staff take calls and identifying recurring themes
  
• Time how long the calls take to be resolved.
  

As you can see, these techniques try to obtain the same information as the reported data first listed, but the sampling size is going to be quite small and difficult to get extract accurate information from.

Hope this helps.