Customer Satisfaction Score, or CSAT, summarises how satisfied respondents say they are with a defined experience, product or relationship. It is often collected after a particular interaction, such as a support conversation, but the question’s wording determines what is being rated.
A score can look reassuring while leaving the underlying experience unclear. Someone may appreciate a courteous support agent yet remain unhappy that the product required a third contact. Defining the object of satisfaction is therefore part of the measurement, not a minor detail in the survey introduction.
State the question and scoring convention
A common approach uses a five-point scale and reports the percentage selecting the top two categories. If 80 of 100 valid respondents select “satisfied” or “very satisfied”, that top-two-box CSAT is 80%. Reporting only the highest category, or reporting a mean, produces a different measure.
Qualtrics’ explanation of CSAT describes the common top-two-box calculation. There is no single score that can be interpreted without knowing the question, scale, included responses and calculation.
Use balanced response options and a clear object: satisfaction with today’s delivery, a particular support interaction or the product overall. Avoid combining several experiences in one question if the team needs to distinguish them later.
Choose timing that fits the experience
A prompt immediately after a chat can capture the reaction to that chat, but the customer may not yet know whether the proposed fix worked. A later question about resolution can therefore answer something different. Select the timing according to the decision rather than automatically treating faster collection as more accurate.
Keep the trigger consistent across comparisons and record any change. If a team begins asking only after tickets are marked resolved, an apparent improvement may partly reflect who now receives the survey.
Check who responds and whose experience remains absent. Non-response bias is possible, but it is not safe to assume that only the delighted and furious answer every survey. Investigate participation using relevant information where available.
Use the score to direct investigation
A short open follow-up can reveal what informed the rating. Compare the comments with operational evidence such as repeat contact or unresolved issues, while preserving differences in what each source measures. A satisfaction rating does not directly measure effort, successful resolution or actual loyalty.
For a fictional delivery service, low ratings may concern late arrival, damaged goods or uncertainty about collection instructions. The same aggregate CSAT could conceal very different actions for the team. Examine the distribution and context before selecting a remedy.
Report the sample size and suitable uncertainty, and compare like with like. Customer Effort Score may help investigate difficulty, while Net Promoter Score asks a different question about recommendation. The aim is to choose evidence that explains the customer’s experience, rather than collecting every available score after every interaction.