%202.png)
How 60 Decibels tested WhatsApp against AI voice-call surveys in Kenya and El Salvador, with no incentives, and came back for six more studies.
How 60 Decibels tested WhatsApp against AI voice-call surveys in Kenya and El Salvador, with no incentives, and came back for six more studies.
In short
60 Decibels, an impact measurement company, piloted WhatsApp surveys with Yazi on two live client projects in Kenya and El Salvador in August and September 2026, with no incentives. Both pilots came in close to Yazi's average response rate of about 22%, against the under 5% 60 Decibels had generally seen from AI voice-call surveys. It then bought a six-study bundle.
60 Decibels measures social impact by asking the people a product or programme is meant to serve how it has changed their lives. Its research operations team runs studies for companies and funders around the world, often with customers who are hard to reach by email.
The team had been testing AI voice-call surveys, and response rates were generally under 5%. After reading Yazi's monthly product updates, 60 Decibels got back in touch in June 2026 and agreed to pilot WhatsApp on two live client projects.
AI voice-call surveys were reaching fewer than 1 in 20 of the people invited.
60 Decibels doesn't usually pay participants on its own projects, so the channel had to work on its own.
The pilots had to run in Kenya and, in Spanish, in El Salvador.
The team wanted results from live client projects to set against its own voice-call benchmark.
Yazi ran two pilot studies on WhatsApp for 60 Decibels clients: new users of an agricultural advice service in Kenya, and customers of a lender in El Salvador, in Spanish. Each pilot combined structured questions with AI follow-ups on open answers, and neither offered an incentive.
The same contact lists and questions 60 Decibels would normally take to the phone.
Illustrative screen from the Spanish-language pilot format. Not a real participant.
Out of every 100 people invited
60 Decibels' AI voice-call benchmark, against the two WhatsApp pilots
Both pilots came in very close to Yazi's average response rate of about 22%, with no incentives. In Kenya, Research Operations Manager Niranjan Suresh reported the interviews were done in two business days. When Niranjan showed the results to 60 Decibels' leadership, the answer was to do more.
“This one was sold as an online survey, and I told them let's do WhatsApp, because WhatsApp is way better.”
“We didn't use incentives, because we typically don't on our own projects, and it still worked.”
How the work grew
| AI voice-call survey | Yazi on WhatsApp | |
|---|---|---|
| Response rate | Generally under 5% | Close to 22% in both pilots |
| Incentives | Not used | Not needed |
| Participant experience | Answer a call when it comes | Reply in WhatsApp when it suits them |
| Open answers | Spoken, then transcribed | Text or voice, with AI follow-up |
| Fieldwork speed | Paced by call attempts | Kenya pilot done in two business days |
Evidence from its own client projects that WhatsApp reaches more people than its voice-call benchmark, without incentives. It now recommends WhatsApp to clients and has six studies lined up.
Customers in Kenya and El Salvador answered in a chat app they already use, in their own language, at a time that suited them.
With AI voice surveys we've generally gotten response rates of under 5%, so we didn't know what to expect. You mentioned 22% response rates are average for you, and these two pilots are actually very close to that average. I'm pleasantly surprised.
60 Decibels judged WhatsApp the way it judges impact: by asking real customers and comparing the result with its own benchmark. On two live projects, without incentives, response came in close to 22%, against the under 5% it had seen from AI voice calls. That evidence now shapes which channel its clients are offered.
In 60 Decibels' pilots, WhatsApp surveys came in close to 22% response in Kenya and El Salvador, with no incentives. 60 Decibels had generally seen under 5% from AI voice-call surveys. About 22% is also Yazi's average response rate.
No. 60 Decibels doesn't usually pay participants on its own projects, so neither pilot offered an incentive.
In Kenya, with new users of an agricultural advice service, and in El Salvador, with customers of a lender, in Spanish.
60 Decibels' leadership signed off more WhatsApp work, an upcoming multi-country project moved from online to WhatsApp, and in October 2026 60 Decibels bought a six-study bundle.
Want to test WhatsApp against your current channel?
Book a 30-minute demo →