Voice IVR Survey: Best Practices to Improve Completion Rates
Organizations face many challenges when implementing Voice IVR surveys (and so many hurdles that they haven’t even considered), especially when seeking to survey large audiences.
From working with organizations in more than 50 countries, we’ve learned a lot about using IVR as a data collection tool. We’re now ready to share what we believe to be the most effective ways to build successful IVR surveys. If you want to increase your engagement rates, read on.
The Golden Rule: TEST, TEST, TEST
One measure of success of an IVR campaign is cost-effectiveness. Financial resources are involved in terms of scale, which makes pilot tests essential for finding creative ways to maximize the impact of your program.
Getting feedback from real end-users will ultimately improve your succeeding content. Test your campaigns to a few people before scaling and assess whether you are getting an acceptable response rate. If the results are not satisfactory in the first test, then make necessary changes in your content or approach. Don’t expect anywhere near a 100% interaction rate. In most cases, 50% to 70% is very good.
The key is to improve the content and test again until you get results you are happy with.
Know your target audience
Getting a high IVR response rate is about knowing your target audience. You should be able to customize your messages in light of what you know about them. The following are some best practices we’ve gathered for increasing response rates.
Assess their literacy level
In poorer countries, literacy level is low so people who are unable to read will not be able to understand SMS message. In TabangKO (“My Help”) program by Mercy Corps Philippines, it was found out that Voice calls had ten times the response rate of SMS (48.2% to 4.5%), bypassing lower literacy levels.
Although using IVR would cost more compared to sending SMS, illiterates will not be able to comprehend the messages that you end up spending for something that is not effective. With voice calls, illiterates can listen and understand pre-recorded audios, and may even find these calls entertaining as receiving an IVR message is something new and interesting to them (especially if it is delivered in an engaging format such as ‘soap operas‘ or uses storytelling). Hence, if your target audience is the poor and illiterates–IVR is a better survey tool.
Are you part of an NGO or research program at a university? Make sure to check out our dedicated IVR guide for emerging and developing countries.
Use their preferred language
In countries or regions that have multiple languages, it is best to let people select their local language to communicate. Using their preferred language or dialect will get your message across more effectively. The engageSPARK platform has the ability to support multiple languages in a single campaign. This way you don’t have to create a different engagement for different languages for the same content.
Give clear and specific instructions
Provide as much clarity as possible when giving instructions. Many users, especially in developing countries, never experienced an IVR so educating is key. When asking users to give a spoken response, make sure to add clear instructions on how they should answer. For example, “Please record your response after the beep sound. When you are done speaking, please press the 9 button on your phone’s keypad”.
Here is an example of a script that can be confusing to first-time IVR users: “If you would like to register, press 1. If no, press 2.” Users who never tried going through an IVR call might not have any clue that they need to press a button to answer. To make it clear you can say, “Press the 1 button on your phone’s keypad to register. If you can’t go, press the 2 button on your keypad.” Adding a few more words to explain can bring out clarity.
Using the right terms is another factor to consider for your messages to be interpreted correctly. People in different places may use different words. For example, people in the United States understands what a ‘keypress’ means. However, in other countries, it might make more sense if you use words like ‘key button’ when instructing them how to answer IVR questions.
Find out if the pace of the speaker is too fast or enough to be understood. Keep your questions simple. If you want to get a better feedback, asking open-ended questions is good. Also, set some expectations—provide a short introduction on what the call is all about and how long it will take.
Consider the call duration
The length of the call is important. Ideally, calls should be between 1 to 5 minutes long. Anything longer than 5 minutes is probably too long. For example, having 10 questions or more in a single survey call might be too long and time-consuming. You don’t want users to hang-up halfway on the survey. Instead, launch a separate survey call for the rest of the questions.
Figure out a good timing for making calls
Identify a good time of the day where people are most likely available to answer the call. Do an assessment beforehand to know the best period or time frame or frequency to call before the starting the actual campaign.
If you’re sending out 30-45 second messages with really useful content, then you could even do it daily. A short call that keeps them engaged and gives them useful, actionable information that they can actually use that day could be very powerful. If the calls will be longer, like 3 – 5 minutes, with one or more interactions on each call then we recommend 1-2 calls per week.
Set up automated call retries
Don’t forget to schedule automated call retries to increase the likelihood people will answer the IVR call. In case users do not answer during the first call attempt – let the system automatically call them back based on your call retry rules. For example, they are busy, or their phone is off or the battery was drained, or they are experiencing signal or network issues—the system will call again once they are available to pick up your automated call. This can significantly improve your engagement rate.
Analyze reports, reach out to test users, and make modifications
Lastly, be data-informed. Analyzing the results of your survey is crucial in order to come up with good decisions for the next steps of your program. Delve into the detailed reports of your IVR survey and get actionable insights on how to improve the content, to scale and create a real impact.
Figure out what questions in your IVR survey with most invalid responses. This will help identify what questions or audio recordings that users have difficulty with. Or better yet call the test users and get their live feedback—ask them if they clearly understood the questions and learn how they would have asked it themselves. This way you’ll get a better idea on how to formulate clear-cut questions. For the people who failed to answer or press keys, ask “Why did you not speak or press the key, what happened, and how can we make the instructions clearer?”.
By gaining a full understanding of the individual responses, you can make necessary changes in your audio script or call flow, be able to provide clear instructions and get the information you need from the survey.
Unilever recently compared SMS and Voice (IVR) for data gathering. See the results in this blog post.