How the calculation works
- Observed CSAT p = positive responses ÷ survey responses (n)
- z = 1.96 for a 95% interval
- Centre = (p + z² ÷ 2n) ÷ (1 + z² ÷ n)
- Half-width = z ÷ (1 + z² ÷ n) × √(p(1 − p) ÷ n + z² ÷ 4n²)
- Likely range = centre ± half-width, kept between 0% and 100%
What the range tells you
The range shows how far the score could move from random sampling alone. A small number of responses gives a wide range, so a change between two periods may be noise rather than a real shift. The calculator always uses a 95% interval.
Assumptions and limits
- The interval covers random sampling error only. Non-response and selection bias can be much larger and are not included.
- It does not tell you whether respondents represent all customers. Report the response rate and segment mix next to CSAT.
- Each response is counted as either positive or not positive.