Use Case: Survey and Feedback Collection
An AI agent runs phone surveys, NPS calls and post-service follow-ups: it asks a few questions, handles clarifications, and captures the answers so you can review them in one place.
The problem
- Email and text surveys are easy to ignore.
- Human interviewers are costly, and bigger samples need bigger teams.
- Open-ended feedback is hard to collect at scale.
How it works in Taalk
Survey list → AI calls → Asks 3–5 questions → Captures answers → Results in Call Analysis or Post-Call Analysis
To collect answers from every call, add your survey questions as Call Analysis questions on the agent (sent with the post-call webhook). To see them grouped and counted in Taalk, use the Post-Call Analysis add-on.
Survey types
NPS
"On a scale of 0 to 10, how likely are you to recommend us to a friend?" Then: "What's the main reason for your score?" Promoters score 9–10, passives 7–8, detractors 0–6.
Customer satisfaction
"How satisfied were you with your recent visit?" Follow up on low scores: "What could we improve?"
Post-service follow-up
"I'm calling about your appointment yesterday." Was the issue resolved? How was the technician? Anything else we should know?
Market research
"We're exploring a new service and would value your input." Interest, willingness to pay, priorities.
Sample script
Put the persona and goal in Objectives, and the questions in Script and logic. *Pending Caller Feedback* (from the Snippets bar) makes the agent wait for the answer.
Hi {name}, this is {agentName} with a quick follow-up from [Company] about
your recent visit. I have three short questions; it takes under two minutes.
Is now a good time?
*Pending Caller Feedback*
First, on a scale of 0 to 10, how likely are you to recommend us to a friend
or colleague?
*Pending Caller Feedback*
Thank you. How satisfied were you overall: very satisfied, satisfied,
neutral or dissatisfied?
*Pending Caller Feedback*
Last one: is there anything we could do better next time?
*Pending Caller Feedback*
That's really helpful. Thank you for your time, {name}. Have a great day!
Use your own columns (for example {Taalk_VisitDate}) for visit details. See Script Variables.
Tips
- Keep it short. 3–5 questions.
- Ask one open question. "Can you tell me more?" gives richer feedback.
- Call soon after the experience, within a day or two.
- Say how long it takes up front.
- Offer a text instead. A playbook with an NPS Survey step collects the score by text from people who don't pick up.
What to measure
| Metric | Where |
|---|---|
| Completed surveys | Your Call Analysis answers, or Post-Call Analysis outcomes |
| Human answer rate | Conversion card on the Dashboard |
| Text responses | Messaging Performance (Reply Rate) on Campaign › Reporting |
Features used
| Feature | Where | Learn more |
|---|---|---|
| Objectives and script, Snippets | Agent Configuration › Objectives and script | Objectives and Scripts |
| Call Analysis questions | Agent Configuration › Post-call and automations | Event Webhooks |
| Post-Call Analysis | Platform › Marketplace ($25.00 per month) | Post-Call Analysis |
| NPS Survey step | Outreach › Playbooks | Playbook Steps Reference |
| Campaign report | Campaign › Reporting | Campaign Report |