IVR Surveys: Collect Feedback by Phone (and When Not To)

IVR Surveys: Collect Feedback by Phone (and When Not To)

IVR survey: outbound voice calls from a central IVR node reach customers, who answer by phone keypad with ratings to give customer feedback

An IVR survey is the right tool when you need feedback from people a written form leaves out: customers who can’t comfortably read, who live outside the cities, or who answer better in their own language. It is the wrong tool when your questionnaire is long, complex or full of open-ended questions.

The right method depends on who your customers are and what you need to learn from them. If many of your customers in Ghana would struggle with a text message or a web form, a short outbound IVR survey after a delivery, a support call or a branch visit gives them a way to answer. VoiceConnect, Arkesel’s cloud contact centre and IVR platform for businesses in Ghana, runs these calls.

What Is an IVR Survey?

An interactive voice response survey, or IVR survey, is an automated phone survey. A recorded voice asks each question and the respondent answers by pressing a key on the phone’s keypad, so no agent needs to be on the line. It runs on the same technology as the phone menus described in our guide to what IVR is and how it improves customer service.

IVR surveys run in two directions:

  • Outbound. You call a list of people, such as the customers you served last week, and the survey plays when they answer.
  • Inbound (post-call). In some contact centres, a customer who has called in is asked to answer a short survey on the same call, once the conversation ends.

VoiceConnect surveys are outbound calls to a contact list you import. They cannot run automatically at the end of an inbound support call. To hear from customers after a support call or a delivery, you call them with an outbound survey soon after the interaction, and the step-by-step section below shows how.

An IVR survey is also different from IVR self-service menus, where a caller dials in to finish a task such as checking a balance. In a survey, your business collects information and the customer supplies it.

When Voice Surveys Win: The Inclusion Case

The strongest reason to choose voice is reach. A written or SMS survey leaves out anyone who struggles to read, and in Ghana that is a large share of the population.

According to the Ghana Statistical Service’s 2021 Population and Housing Census, 30.2% of people aged 6 and older in Ghana are not literate in any language, and the share is higher among females (34.4%) than males (25.9%). A text questionnaire cannot reach those customers, however well you write it.

The gap is wider outside the cities. The same census found that 80.6% of the urban population aged 6 and older is literate in at least one language, compared with 55.2% of the rural population. By region, literacy among people aged 6 and older ranges from 87.9% in Greater Accra to 32.8% in Savannah Region. If many of your customers live in regions where fewer people read, a written survey will mostly hear from the ones who do.

Voice closes that gap on three fronts:

  • Literacy. A spoken question needs no reading. Respondents listen and press a key.
  • Reach. A voice survey works on any phone, including feature phones, and the respondent needs no internet connection or mobile data to answer it.
  • Language. You record each question in the language you need and upload the recording, so customers can answer in English, Twi, Ewe or whichever language they are most comfortable in.

Do IVR Surveys Work? Evidence from Ghana and Elsewhere

Yes, when the survey is short and aimed at the right people. Field studies in Ghana, Uganda, Ethiopia and Sweden show voice surveys getting answers, and the Ethiopian study shows how many people drop out between picking up and finishing.

In Ghana, a national IVR survey of mobile phone users, published in PLOS ONE, was fielded over 27 days in February and March 2017 and produced 9,469 completed interviews and 3,547 partial ones. Its response rate was 31% and its cooperation rate was 81%, meaning most eligible people who were reached went on to complete the interview. That survey dialled random mobile numbers. A survey of your own customers calls people who have just dealt with your business, so your results will differ. Treat these figures as evidence that the method works in Ghana, not as a forecast.

In northern Uganda, as reported by engageSpark, citing the Natural Resources Institute, University of Greenwich, 74% of enrolled women (116 of 156) completed an IVR survey of 88 questions, with an average call length of 23 minutes. Those women had enrolled in a research study, so this is a completion rate among willing participants rather than a cold-call response rate. It does show that a long voice survey can be finished when respondents have agreed to take part.

In Sweden, a randomised trial with 80 patients from a child and adolescent psychiatric outpatient unit, published on PubMed Central, found that 65% of the IVR group gave complete answers against 38% of the postal survey group. The trial covered one small clinic, so read it as an indication rather than a rule.

In Addis Ababa, Ethiopia, a study of health-facility clients phoned 2415 people with an automated IVR survey: 92.6% picked up, 65.6% answered at least one question and 42.9% completed the full survey. Nearly everyone answered the phone, but fewer than half finished. That drop-off is the main thing to design against when you write your questions.

How to Collect Customer Feedback After a Service Interaction

Choose the interaction you want feedback on, call the customers you served with three to five keypad questions in their language, and act on the pattern in their answers. If you run a business in Ghana, here is how to set up custom IVR surveys for customers on VoiceConnect.

  1. Pick the moment. Choose one interaction to measure, such as a delivery, a closed complaint, a support call or a branch visit. Asking about one recent experience means the customer knows exactly what they are rating.
  2. Build the contact list. Export the customers you served in that period, with their phone numbers, from your order system or CRM. VoiceConnect imports contacts from CSV or Excel, and Do-Not-Call lists are enforced on every campaign automatically, so customers who have opted out are not called.
  3. Write three to five questions. Give each question one idea and a keypad scale. The next section has guidance and a sample script.
  4. Record the prompts and build the flow. Record each question as an audio file in every language you plan to offer and upload it for that prompt. Then lay out the survey in the visual IVR builder, using keypad menus to capture the answers, or have Arkesel’s team set the flow up with you during onboarding. Our guide to designing an IVR call flow explains how to structure menus and branches.
  5. Send the survey while the experience is fresh. Run it as an outbound campaign to the list you imported, soon after the interaction it asks about, so customers still remember what they are rating. The survey does not start on its own when a support call ends, so build each list from the customers you served recently and send the campaign to that list.
  6. Read the results. The dashboard shows per-question response counts, so you can see how many customers chose each answer to each question. You can also download a CSV export of responses to sort and filter in a spreadsheet. Use the counts to decide which part of the service to fix first.
  7. Act on what you learn. Fix the cause behind the low scores, then run the same questions again after the change to see whether the scores move. If you want to call back individual customers who rated you poorly, check that the CSV export shows who gave each answer before you promise anyone a callback. If the complaints point to how calls are handled, call centre quality monitoring helps you find out where.

To see the survey, campaign and IVR builder features in one place, explore VoiceConnect, which is available in Ghana.

How to Write IVR Survey Questions People Will Finish

Keep a customer feedback IVR survey to three to five questions. In the Addis Ababa study, the full survey of 15 rating-scale questions took 8.1 minutes on average, and fewer than half of the people called completed it.

Use these rules when you write the script:

  • One idea per question. “How satisfied were you with the speed and the politeness of our driver?” asks two things and leaves the customer unsure which one to rate. Split it, or pick the one that matters more.
  • Put the scale inside the prompt. Tell the customer which key means what, such as “Press 1 for satisfied, 2 for not satisfied.” Keep the keys in the same order for every question so nobody has to relearn them.
  • Offer choices instead of asking for numbers. In the pre-test for the 2017 Ghana survey, a multiple choice format improved call continuation and data quality compared with asking respondents to enter a specific number. Ask for an age band, not an age.
  • Keep the introduction short and open in a local language. The same pre-test found that a brief greeting in Twi that named who was calling, followed by a choice of survey language, kept more people on the call. A shorter introduction also improved call continuation compared with a longer one.
  • Leave open-ended questions out. Keypad answers suit ratings and choices. If you need a customer’s comments in their own words, ask for them in a written channel or in a follow-up call from an agent.

Here is an example script for a delivery business. It is a sample to adapt, not a tested template.

  1. Greeting, recorded in Twi with an English option: “Hello, this is [your business name] calling about your delivery yesterday. For English, press 1. For Twi, press 2.”
  2. “Did your order arrive on the day we promised? Press 1 for yes, or 2 for no.”
  3. “How satisfied were you with the delivery? Press 1 for very satisfied, 2 for satisfied, or 3 for not satisfied.”
  4. “Would you order from us again? Press 1 for yes, 2 for not sure, or 3 for no.”
  5. Close: “Thank you for your time. Goodbye.”

Use Cases That Fit Voice Surveys Best

Voice surveys earn their place when reach and speed matter more than depth:

  • Customer satisfaction (CSAT). Call the customers you served with a short IVR customer satisfaction survey soon after a support call, a delivery or a branch visit, using the steps above.
  • Healthcare follow-ups. Reach patients after discharge, run symptom check-ins or confirm they are taking their medication, including patients who can’t read a form.
  • Market research. Collect responses across regions and income levels without excluding people who use feature phones or cannot read a questionnaire. Check who answered against who you meant to reach, as the next section explains.
  • Programme monitoring. NGOs that need to hear from beneficiaries in remote areas can read our guide to running voice surveys for NGO programmes in Ghana.

When a Written, SMS or Web Survey Wins Instead

Voice is not the answer for every survey. A written, SMS or web survey is the better tool when:

  • The questionnaire is long or complex. A respondent cannot re-read a spoken question, so long or complex wording is easy to lose. Keep voice surveys short, and send detailed questionnaires to a screen where respondents can re-read.
  • You need open-text or qualitative depth. Keypad replies suit ratings and multiple choice. When you need respondents to explain in their own words, a written channel captures it better.
  • Respondent convenience matters. A web or SMS survey lets people answer on their own time, pause and go back. A call asks for their attention right now.

Voice surveys also miss people. The 2017 Ghana survey found that younger, urban, better educated and male respondents were overrepresented, with fewer women, rural and older residents completing it than in household surveys. That study dialled random numbers rather than calling a customer list, so the gaps in your own results will look different. In Addis Ababa, clients of private facilities were less likely to take part and complete the survey than clients of public health centres. Compare who answered with who you served, and notice which groups stay quiet.

An SMS survey suits a quick, single-question check with people you know can read, and they can reply when it suits them.

Voice vs SMS vs Web vs Written: A Quick Comparison

Voice (IVR)SMSWeb / OnlineWritten / Postal
Best forLow-literacy, rural and multilingual reachFast, single-question checksDetailed, self-paced surveysFormal, document-heavy surveys
Reach / inclusionAny phone, no data, no reading neededWide, but needs readingNeeds a smartphone and dataNeeds literacy and a postal address
Question typesRatings, multiple choice (keypad)Short text, ratingsAny, including open textAny, including open text
Length toleranceLow, so keep it shortVery lowHighHigh
Response behaviourAsks for an answer nowEasy to read later, or to ignoreSelf-paced, drop-off mid-formSlow, often set aside

How to Run a Voice Survey on VoiceConnect

If voice fits your goal, VoiceConnect runs the whole survey, from the call flow to the results dashboard and CSV export. It is available in Ghana, with numbers and connection supplied in Ghana; see VoiceConnect in Ghana for local setup.

Build the survey flow. You build the flow by dragging steps into place, use keypad menus to capture answers, and change a menu later without a developer. Record and upload your prompts as described in step 4, and if Arkesel’s team sets the flow up with you during onboarding, you can still edit it yourself afterwards.

Because there is no agent on the line, a survey runs as an automated campaign rather than a staffed outbound dialling run. VoiceConnect is part of a wider cloud contact centre in Ghana, so surveys sit alongside inbound and outbound calling, IVR menus and agent tools when you need them.

Frequently Asked Questions

What is a post-call IVR survey?

A post-call IVR survey asks a customer for feedback by phone shortly after an interaction with your business, such as a support call or a delivery, while the experience is still fresh. In some contact centres the survey plays on the same inbound call once the conversation ends. VoiceConnect works differently: its surveys are outbound only for now, so you send an outbound survey call to a list of the customers you served soon after their interaction.

How many questions should an IVR survey have?

For customer feedback, keep it to three to five keypad questions. In the Addis Ababa study, fewer than half of the people called completed a 15-question survey, so keep only the questions whose answers you will act on.

When should you use a voice survey instead of an online survey?

Choose voice when you need to reach respondents who can’t comfortably read, who live in rural or low-connectivity areas, or who answer better in a local language. Choose an online survey when your questionnaire is long, needs open-text answers, or targets people you know can read and get online.

Are voice surveys good for low-literacy respondents?

Yes. The questions are spoken and the answers are keypad presses, so respondents don’t need to read. That matters in Ghana, where the 2021 census found that three in ten people aged 6 and older are not literate in any language. Check who completes your survey, though, because voice surveys still under-reach some groups.

What are the disadvantages of voice surveys?

They suit short surveys, not long or open-ended ones, many people who pick up do not finish, and some groups answer less often than others, as covered in the section on when other methods win.

How do you start an automated voice survey?

Follow the seven steps in the customer feedback section above, starting with the interaction you want feedback on. If you are in Ghana, you can explore VoiceConnect or talk to our team to set one up.

Related Articles

Choose the Method That Reaches Everyone

The best feedback method is the one your customers can actually use. When a written survey would leave out the people you most need to hear from, an IVR survey reaches them on any phone and in their own language. When you need depth and self-paced answers, a written, web or SMS survey is the better choice.

If you run a business in Ghana and want to start collecting customer feedback by phone, talk to our team about setting up your first survey flow and current pricing. Plan options are listed on the VoiceConnect page, and the pricing page lists rates for Arkesel’s other products.

Scroll to Top