Home > Resource Center > VUI-Design > Elements of Phone Survey Design

The Elements of Phone Survey Design

Phone surveys soliciting feedback on satisfaction with a service or a product are one of the most popular types of applications deployed at Angel.com. Whether it is a survey assessing customer satisfaction after a retail purchase (with the phone number printed on a receipt), a dining experience (with the phone printed on a stand-alone survey card), or a call-center interaction (with the human agent transferring the caller to an automated survey), capturing timely, contextually informed, and vivid feedback from a customer is often best done through the telephone.

But more often than not, clients who come to Angel.com wishing to deploy a phone survey are already soliciting feedback from their customers using a paper or a web questionnaire. What usually happens is the following: delighted by Site Builder's user-friendliness and feeling empowered by its intuitive interface, they dive right into a linear mapping of their paper or web survey questions onto Site Builder's question pages. The result is usually catastrophic: a survey that seemed perfectly structured and easy to fill on paper turns out to be nothing short of a harrowing nightmare to take over the phone. The questions are hard to understand, the flow from one question to the next seems to be totally arbitrary, and the demands the systems puts on the caller turn out to be unreasonable and overbearing.

Structure the Survey
Perhaps the most important first step in putting together a phone survey (whether from scratch or from an existing paper of web survey) is to sit down and carefully design its structure, keeping in mind that there are fundamental differences between a Voice User Interface (VUI) and a Graphical User Interface (GUI). See "Why VUI design is harder than GUI design".

There are at least two types of ways to structure your phone survey:

1. By Topic: if your survey solicits customer satisfaction on three departments within your company - sales, marketing, and technical support - then make your survey a three-part survey, with the first part soliciting feedback on sales, the second on marketing, and the third on technical support. Seek a structure that creates a small number of logical clusters and keeps your clusters conceptually on par with each other. If your survey about the three departments within your company asked five questions about each department, an inferior design would create five clusters of three questions rather than three clusters of five questions. This would sound more erratic, more repetitious, and as a result more tedious to complete. It would also tax the caller's patience and attention and result in more errors, thus needlessly extending the length of the call.
   
2. By answer type: sometimes you may need to build a survey that has no obvious, clear-cut object-oriented structuring. In such situations, try to group your questions by the type of answer you are soliciting from the caller. For example, if you are asking 10 questions, and four of them are yes/no questions, three are soliciting a number from 1 to 5, and three are open-ended questions (i.e., the answers are captured as recordings), then divide the survey into three parts (yes/no, 1-to-5, and open-ended).
   

Announce the Survey

1. The reason for giving the survey: give a one-sentence description explaining why you are asking the caller to take the survey and what you are trying to achieve. If the survey is a client satisfaction survey, then briefly explain that you are soliciting feedback from callers with in order to identify any problems that can be redressed. Use this part of the survey to persuade the caller into becoming a willing participant.
   
2. The length of the survey: when someone takes a paper or a web survey, one of the first things they do is assess the length of the survey - i.e., determine how many questions they are asked to answer. Unfortunately, callers do not have the ability to carry out such an assessment and are at the mercy of the system to tell them how many questions they will be asked. Be merciful on callers and tell them upfront how many questions they will be asked and how much time they should expect to spend completing the survey.
   
3. The structure of the survey: also helpful to callers is information on the structure of the survey, with a few words describing the purpose of each part. For example, "In this three-part survey, I will assess your satisfaction with our sales department, our marketing department, and our technical support."
   

Close the Survey

1. Inform the caller that the survey is complete: when filling out a paper or a web survey, a respondent can easily tell when they have answered the last question of the survey. Not so with phone surveys. Make sure that when the caller has answered the last question, you inform them that they have completed the survey.
   
2. Tell the caller what you will do with the answers: a one-sentence reminder of what was said in the opening greeting about the purpose of the survey should be given at the closing of the survey. For example, "Your answers have been saved and will be reviewed by our quality assurance department."
   

Position the Caller
In addition to marking the end points of your survey, make sure that you also mark the individual lower-level elements and sub-structures.

1. Announce beginning of sections: whether the section is a topic-centric or an answer-type-based grouping, preface each section with a one-sentence description of what it is about. In situations where the grouping is both topic-centric and answer-type based, announce both the topic and the answer type in the opening sentence to the section. For example, "First, I will ask you about your level of satisfaction with our sales department. Please answer each question by simply saying, 'yes' or 'no'."
   
2. Signal section transitions: it is also crucial to mark the end of a section and the beginning of the section that follows is. Often, a word or two will do the trick, especially in situations where the individual sections are short (less than 5 questions) and the survey taker can sustain in their mind the structure of the survey. For example, "Ok, next I would like to assess your satisfaction with our marketing department," would be a good transition from the first section that focused on the sales department to the next section focusing on marketing.
   
3. Explicitly mark end of long sections: in situations where the individual sections are long (more than 5 questions), and the individual questions themselves are long, it is helpful to explicitly signal to the caller the end of the section before transitioning to the next section. For example, "Great, this concludes the first section of the survey. Next, I would like to assess your satisfaction with our marketing department."
   
4. Announce length of sections: this is especially crucial when the number of questions comprising the section is greater than 5. But if you do announce the number of questions to follow, be sure to do so systematically for all sections.
   
5. Number the questions: usually this marker device is helpful for sections that have a large number of questions and where all the questions have the same answer type. For example, numbering is very helpful when soliciting a reaction to series of statements. For example, "Statement one: my immediate supervisor gives me regular feedback on my performance."
   

Help the Caller with How to Answer

1. At the beginning of each section: sections that solicit the same type of answers should open with instructions on how to answer the questions that are about to be asked: "I will now assess your level of satisfaction with our technical support. Please answer each statement by giving me a number between 1 and 5, 1 meaning that you totally disagree with the statement, and 5 meaning that you totally agree with it."
   
2. At no-match and no-input: always repeat the type of answer you expect from callers at a no-match or a no-input error before re-asking the question. See "Simple Error Recovery Strategies".