March 2002 Meeting Recap

          Summary of March 27th Meeting: Developing a Reliable Survey Process

          - by Kim Rizzitano

          Dr. Fred Van Bennekom, Principal of Great Brook Consulting, gave us some valuable insights to surveying customers. When doing a survey, it’s important to understand what a survey is, and what you want to gain from a survey. Most surveys are done by looking at the population (the group we want to learn something about), and taking a random sample of that population (this helps eliminate some of the bias). Next you need to design and administer the instrument to the sample and generalize the results to the population.
          Instrument Validity + Administration Accuracy = Reliability

          There are 7 steps to designing the instrument (survey) process:

          1. Interview management. It is important to get their input about what they want to accomplish.
          2. Identify the questions to ask. This can be done with focus groups, etc.
          3. Draft the survey instrument. Remember to keep the survey relatively short and to the point. Construct questions in such a way that they can be analyzed. Watch the question formats and wording. Decide on what scale of use you will use for the questions.
          4. Review the survey by a project team. Verify that wording and questions accurately describe what you need the customer to answer. Insure that the answers will give the information they are seeking.
          5. Revise iterations. This can be done more than one time. It’s important to review the questions for accuracy and clarity.
          6. Conduct a pilot (or beta test) to a small group. This will catch vague questions and make sure your data is being answered correctly. Then go back to revise iterations.
          7. Redraft and finalize the instrument (survey).

          The first step, identifying the questions to ask, is very important. This takes some analysis by the survey team. The attributes of the service delivery (what do they need to understand and track), the attitudinal outcomes (perceptions of service or delivery), the demographics of the segments (is this by department, region of the country, etc) all need to be considered. The question format can skew the bias of the answers very easily. Then the validity of the survey results can be misleading. All of these things need to be taken into account.

          Draw a service blueprint; or a process flow diagram. You can do this by reviewing the complaint data from customers, conduct focus groups to get the data or perform interviews of customers to flush out the critical data.

          It is important to remember you are not looking for just their opinion on something, but examples. A good way to obtain these examples is to ask the customer to describe an event which was positive. Then describe an event that was negative. From this data, you can get the attributes you need to have in your survey.

          Attributes can be classified into several dimensions. Reliability to deliver on promises; responsiveness or being willing to help; assurance which inspires trust and confidence, having empathy and treating the customers as individuals, and tangibles which represent the services physically (tangibles are the least important).

          Market your survey. This is key and a step many miss. There should be a preannouncement letter or email from a senior. executive on why the survey is important and what they plan on doing with the survey results. The survey itself should have a brief introduction. There should be a set of instructions for the participant to follow (even if it seems obvious, include these in your survey to clarify any issues). The first question in a survey is key to grabbing the respondent and to get them thinking. Don’t solicit demographic information first. Information such as names, titles and geographic location should be at the end.

          Determine what rating scales to use. Interval rating scales work well. A scale of 1 to 5 where there are varying degrees of agree or disagree. These can be labeled above the number (fully anchored) or endpoint anchored (where 1 says Strongly Agree, 2-4 have nothing above them, and 5 says Strongly Disagree).

          Using Incentives as a way of getting participants to answer the survey was also discussed. There are many pros and cons to this. Pro is you may get more people answering the survey (higher return) whereas a con could also lead to bias in your random sample. Also having incentives to answer a survey adds to the cost of doing the survey.

          In conclusion, when doing a survey you want to insure that the data collected is good, reliable, and the accuracy can be validated. Management input is important. Drafting the questions and pilot testing them to insure accuracy is critical to the success of the survey. Editing and revising the survey for clarification is key to the success as well. Don’t forget to market your survey so the participants know it’s coming and what to expect.

          For more information about customer surveys, Dr. Fred Van Bennekom has written a book, Customer Surveying: A Guidebook for Service Managers. For additional information, please visit his web site at: http://www.greatbrook.com.



HDI-NE Home Page

Underwritten by GroupSoft Systems, Inc. HDI-NE Chapter is solely responsible for the content of this site.