Assitance Required - Survey Result Analysis

#1
I have recently been appointed to a CRM change project team and one of my first tasks was to carry out a survey of attendees to the demo presentations of each potential CRM option.

The survey was designed to look at 3 separate topics: functionality, user interface, and administration. Each topic was given a separate section in the survey.

Respondents were firstly asked how important they felt the each area, e.g. functionality, was, overall, in the selection of a new CRM, on a scale of "Not Important at All" to "Very Important" (Total of 5 options).

Following this, respondents were given a set of different options, e.g. functionalities - for example, search functionality, and asked to rank how this related to their CRM use today on a scale of 1 - 5 (1 - most important; 2 - least important).

Subsequently, respondents were asked to rate how they thought the potential CRM fared on each of these options, based on the demo. They were given 6 options: N/A, Very Poorly, Poorly, Fair, Well, Very Well.

Having collected all the results, I am now at an impasse as to how to weight the different types of questions/answers in a matrix that will determine:

1. What was the most important area?
2. What were the most important options in each area?
3. How was each option deemed to have performed based on the demo?

I have some rudimentary statistics education from university (I studied Political Science), but like some advice on how best to approach my analysis.

All suggestions welcome! :)
 
#2
If this were a Likert scale, you could weight the responses and use some parametric statistics. However, with two-direction scaling (Not At All to Very Important) and with choices such as N/A, Likert scaling isn't really appropriate. You are probably best off to use categorical methods. For example, you could sum categories like "somewhat important" and "very important" and say something like "80% of respondents found x to be at least somewhat important". The "most important" area would be the one endorsed by the largest number of respondents. Similar analyses for the other questions. Chi-square tests might be helpful in comparing distributions. The data don't really lend themselves to more sophisticated analysis.