Since you are trying to estimate Cp for a number of errors in each month and your sample size is variable, you better use the U-chart (defects per variable unit). Using this, you can estimate Cp from the "average U" (mean) in a given period, just remember, your process must be "statistically in control" for Cp to be of a real value. Better yet, why not considering categorizing and tallying errors at the same time in the U-chart, so you4ll have better info about the nature of your erros and start analizying based on priorities?

Good luck

Guillermo Ochoa