Customer Survey Benchmarks
Surveys collect opinions of people. Opinions are subjective by nature. In addition, mostly to segment opinions, objective socio-economic criteria like income or gender are collected. By asking people to evaluate their opinions on certain companies, products or services on a numeric scale, one can compare opinions. This is one way to benchmark performances.
But different groups of people, e.g. users of different products or customers of different companies, have differing experiences of what is good and what is bad. Hence, customer survey benchmarks can only give an indication of the potential performance.
Quality of Survey Design is Key
On the one hand, the quality of surveys strongly depends on the design of the questionnaire. The more precise the questions, the better the answers can be used. However, there is no scheme to ensure, that respondants all understand the same question the same way. One only can reduce sources for errors.
Sample Size with Influence on Quality
On the other hand, the quality of surveys depends on the sample of respondants.
Respondants need to fulfill certain screening criteria, before they are admitted to participate in the survey. If respondants fail to meet the screening criteria, the are excluded. A typical screener of a CGG study is, whether the respondants are customers of a certain company.
The size of the sample is the other important factor. There is no uniform sample size that meets all potential requirements. There is also no 100% representative sample size, except the universe (population). But with logical assumption concerning the distribution of results (the standard is normally distributed), standard deviation, confidence interval and estimate error a fair guess of the sample size can be derived.
Our CGG-institutes normally strive to fulfill a sample size of 1.000 in population surveys. Given much smaller universes, sample sizes can be correspondingly smaller.
Our surveys reflect the opinions of users or customers of comparable products, services or companies. Carried out on the same scale, they allow to benchmark the performance of the research objects and give an indication, which choice could be best for other users.