Sample Size Matters

What sample size should I use?

Enter the response rate you expect to achieve and a “limit of error” (ie. how much variation from this hoped-for response is acceptable)

As you can immediately see, statistical probability does not absolve you of making any decisions in this process – you need to enter an expected percentage response rate, and by how much you can live with a result that differs from this expectation by a plus or minus percentage.

This calculation shows the minimum sample size you need to fulfil your stated objectives.

The lower the expected response rate and the smaller the range of error acceptable, the larger will be the recommended test sample size.

Where you have the luxury of a high expected test response rate and are not too worried by a fairly wide possible variation in this rate, you will be able to predict acceptable future performance quite accurately using a smaller test sample size.

Base Data
Direct Marketing & Data Specialists


You might expect a response rate of 1.8 percent to your test mailing but be able to accept a rate say 0.3 percentage points either side of this,
ie. between 1.5% and 2.1%
on your rollout – maybe because of cost savings that can be achieved on rollout volumes for example…

.. of course, we are normally only interested in how much below our expectation of response our rollout result is likely to be.