How Many Customers Should I Survey?
The number of customers to include in a survey is a balance between the goals of the survey and your budget.
As a general rule, you should survey the smallest number of customers you can to meet your survey goals. If you can't get enough survey responses within your budget (or at all), then you should reconsider the survey goals.
For example, if your only goal is to measure a particular metric with a reasonable degree of precision, then 400 customers gives a 5-point margin of error and is usually not too difficult or expensive. But if you want to be able to measure the metric in four different subsets of the survey (for example, to calculate regional scores) you may want to target 1,600 total responses with 400 in each region in order to get the precision you want.
If your survey goal is to provide coaching and training for front-lien employees, you should consider how often you want to provide coaching opportunities. You will also want to make sure the survey method will give enough feedback to be useful--a telephone interview is preferable to an online survey for coaching. A goal of one customer interview per front-line employee per week might be a good compromise between cost and benefit.
It's generally not possible to collect enough feedback to precisely measure survey metrics at an extremely granular level no matter your methodology. For example, trying to get 100 survey responses per customer service representative per month is almost always unrealistic (and would not be economical even if you could do it).