Designing a customer survey

Many surveys and questionnaires fail to deliver the information that is required. Often much data is collected which in the end is very difficult to analyse effectively. In a survey conducted by a high profile basketball club, one question asked was:

How many games did you attend last season?

This seems simple and straightforward enough? Surely it is just a case of filling in a number? Indeed many survey respondents did state a specific number. However, others responded such as "Almost all", "Most", "Half", "All but two" etc. If a substantial number of survey respondents answer this question in this way then summarising the information becomes far more time consuming.

Perhaps the question should have been asked as follows:

How many games did you attend last season? (cross only one box)

It is now possible to summarise answers to the question easily. Moreover it is now possible to place customers into discreet categories and this information can be very useful when looking at answers to other questions.

So are there some simple rules to follow when constructing surveys? The answer is "yes"!

Rule 1	Determine the essential information that the organisation needs. This is a starting point for developing survey questions. Example: The organisation wants to better understand what new customers are looking for when they join the organisation. One of many possible questions that may be asked is: Which of the following represents the main reason you joined this organisation? (Cross only one box) Increase fitness and health Participate in competitive sport Learn new skills Develop new social contacts Other (please state)
Rule 2	Collect demographic data about survey respondents Age, sex, marital status and place of residence are all forms of demographic data that we are used to completing on forms, and with good reason. It is very important to understand the different views of males and females, or young and old, or employed and not employed. Knowing and understanding such differences helps to improve services for particular groups of people. See more information on Demographics
Rule 3	Collect DISCREET Data Surveys should collect data that can be simply analysed and summarised. For example, consider a question such as: What is your age? Such a question may be answered in ways that increase the difficulty of analysis: Possible answers: 14 years 2 months, 25½, over 55. These answers are examples of continuous data. For ease of data analysis such a question should be accompanied by a range of possible answers: What is your age?   Less than 15 years   15 - 24 years   25 - 34 years   35 - 44 years   45 - 54 years   Over 55 years The survey respondent is now required to tick the age category that is appropriate. This is an example of collecting discreet data.
Rule 4	Define terms Analysis of results may be rendered worthless unless terms are clearly defined for the survey respondent. For example, there may be a need to know how the needs of new customers differ from existing customers. In such a case there must be a definition of a "new customer" and "existing customer". Is a new customer someone who signed up last week or is it anyone who signed up less than one year ago? Definitions are clearly required here. Example: New member is a person who joined the organisation in the 12 months previous to completion of this survey. Existing member is a person who has been a member of this organisation for 12 months or more before completing this survey.
Rule 5	Determine your techniques for analysing and summarising answers. Supposing you have 300 survey respondents and each survey has 30 questions! It is extremely important that the survey is not constructed without a thought about dealing with the sheer volume of answers. In such a situation it is worthwhile constructing a database into which surveys are inputted, analysed and summarised. The database should be constructed at the same time as survey questions are being formulated. This reduces the risk of questions that are difficult to analyse.
Rule 6	Determine the number of respondents that would constitute a valid sample population. It may be possible to survey all members or all participants. However a very large number of participants may be too expensive and time consuming to survey. In fact, reliable results may be achieved by surveying only a small number of the total population provided certain rules are observed. The number of people surveyed is called the sample population. Where only a sample population will be surveyed it is best to consult someone with abilities in statistics. They will need to determine the size of the sample population and how it is selected from the total population.