## Survey Project

Hello and thanks in advance for the help.

I have been tasked with solving the woes of the company, finding the smoking gun if you will.

there is a large survey that is conducted by my company to gauge customer satisfaction. within the survey there are lots of indicators to help breakout certain customer types and therefore get the scores given based on segments. I am trying to find the important segments, and i know little to nothing about statistics. At one time there was a CART analysis done and this seemed to get the job done in finding the smoking gun, but i am unable to get the basic programs, and analysis packs to provide a CART analysis on an ongoing basis.

With that said, this is what i do have and what i was thinking of doing, I ask for any insight you can give, and everything will be considered, just remember that i know little to nothing about statistics and the subsequent formulas.

One breakout that we have is a large category, 4 options, i can give the base size of this option as well we two of the important metrics used for measuring the survey. The metrics are resolution: was the issue resolved? the answers are yes, no, and don't know. the metric is calculated as follows (count of yes answers) /[(base size)-(count of don't know answers)]. The second is first call resolultion: was the issue resolved on the first call? the metrics calculation is (count of number of calls = 1 AND issue resolved = yes) / (Base size)

here is a readout of what the data would look like.

category | base size | resolution | first call resolution
category 1 | 100 | 90% | 60%
category 2 | 500 | 88% | 59%
category 3 | 300 | 60% | 50%
category 4 | 600 | 89% | 63%

We can see from this breakout that categories 1 and 4 are doing rather well, while category 1 does not have a large base size, it isnt that important to the overall survey, category 4 also did well and has a large base size, so this category is rather important. Category 3 did the worst, and also has rather significant base size so in this breakout i can see that category 3 is my smoking gun, and needs to be looked into further.

Again this is just a for instance, I have many more categories to look through with the same fields to consider, but a lot of the categories are rather small.

I want to know if there is a way to quickly point out a categories importance regardless of base size, knowing that a small base size with a very high score is not statistically viable compared to others with higher base sizes.

I hope i explained myself correctly, and am in the right forums. I have been pondering over this for many weeks and i am just coming up with nothing.

As i said before i believe a CART analysis would be the ticket, but i would need to know how to provide something like that without the proper tools like SQL analysis pack.