Hi Armando,
Given the customer # you have in the example, it's only a 5 digit number. This means it will be more than likely to hit false positives due to it not being terribly unique.
Have said that, it may be possible to improve accuracy by creating a compound rule with a custom Data Identifier, Creating custom data identifiers.
Rather than creating exclusions, generally it would be better to start with defining the data pattern that you would need to match, and then begin to add optional validators for data identifiers. It might even be possible to use the existing validators.
You can take it one step further by creating custom script validators using the built-in DLP DLP scripting language. This is helpful if there is a specific algorithm used to generate the number.