Designing IVR & online surveys: Guidelines for good practice
We have put together some guidelines for putting together good IVR surveys. All of these points are pretty important when it comes to maximising participation rates, so make this your checklist when you are designing your IVR survey. They last pointer is the most important: Always test to make sure it works in practice! For other pointers see also:
What are your survey objectives ?
Before you start, spend some time considering and clarifying your survey objectives. What are the most important issues? What key performance indicators do you wish to measure through the survey? This is the starting point. Time spent getting the survey right at this stage will always pay off later.
Design for interpretation
Once you have clarified your objectives, design the survey so that you can interpret the results easily. Will it give you the information you want? If not, think again. There is likely to be a better way of wording the questions and responses.
How long is your survey ?
Remember you are prevailing upon your customers' time and trading on their good will when you ask for their feedback. Anything over 3 minutes may be too long, though this will depend on your customers and how happy they are to participate. Always explain at the start how long the questionnaire will take to complete.
Remember the media
If it is a telephone survey you need to bear in mind that the questions will be answered using a phone key pad. Instructions will clearly differ for web based surveys.
Avoid jargon
Do not use words or terms that your customers will not understand properly or that will alienate them! Give clear instructions. Emphasise key words for easy comprehension. For example: "Press 1 if you strongly disagree with the statement made, through to 5 if you strongly agree with the statement."
Be consistent
Don’t ask for a rating between 1 and 4 for one question and for a rating between 1 and 6 for the next. It is important to make everything as easy as possible for the users of the survey. Also, keep the same end of the scale indicating satisfaction. For example, when asking callers to agree or disagree with a statement (1 disagree - 5 agree) do not ask questions like: "The person I talked to was polite", then "The information I got was unclear". In the first question 5 would be a good result, but in the second 5 would be a bad response. This makes it difficult for callers to remember how they should score each question.
Keep instructions simple
If you want to ask a series of questions that have a range of possible responses, the clearest question structure is where a statement is first made and the caller is then asked to say how much they agree or disagree with this statement. This means that one set of instructions can cover all the questions.
Confirm all responses
The customer needs to know that they have pressed the right button, ie. you need to build in recorded confirmation of what they have pressed,
e.g. “1: strongly agree” / “2: agree” / “3: neither agree nor disagree” / “4: disagree” / “5: strongly disagree”
Group similar types of question together
For ease of use, keep all the questions of a similar type together. The caller does not want to go from range type questions to yes/no and back to range, it is too confusing.
Be positive
Try not to ask negative questions e.g. "the person I talked to was not polite". Negatively worded questions give a negative impression and can be leading.
Allow open comments
Always allow callers to leave comments, as this helps to satisfy those callers who have a issue they are eager to express a view on, but which isn't covered by your survey.
Do you need to identify the customer?
You may need to find a way of identifying the caller, e.g. for participation in a prize draw as an incentive to take part, or in order to contact them in case of particularly alarming feedback. This can be done independently of the survey: in which case you need to ensure there is a link to this information. If you do not need to identify the caller, it might help to reassure them that their feedback will be anonymous - if indeed it is - or at least treated in confidence and not passed on to outside parties.
Top & Tail
A good introduction and ending are important. Make the transfer from any previous call as logical and as smooth as possible. At the end, always thank the caller for participating and indicate that they can hang up, with a phrase such as "Thank you for taking the time to complete this survey today. The results will be used to help us improve the service we provide. Goodbye."
Not a Sales Call
It may be important, depending on the context of the survey, to state clearly at the beginning of the survey that this is not a sales call, that participation is voluntary and that you are NOT trying to sell the caller something, purely seeking their input.
AND.....TEST !
It is easy to get a rough idea of how long and how clear the questionnaire will be just by speaking all the questions. Remember to include all the recordings you will need, ie. an introduction and a thank you & goodbye, and remember also to allow time for the customer to respond and the response to be confirmed. Also remember that the clearest written instructions may sound odd or awkward when spoken. You will not know until you try it out !
Survey Design & Methodology Resources
Association for Survey Computing: UK membership body for anyone involved in survey design & analysis. Their site lists events and resources on survey technology.
Resources for Methods in Evaluation and Social Research: this site lists FREE resources for methods in evaluation and social research, primarily US sites.