Use case · outbound voice
Voice NPS and CSAT surveys after a contact
The CallScribe voice AI agent calls the client after a contact, purchase or visit and runs a short survey: it asks for an NPS or CSAT score and one open-ended "why." The answers don't stay just a number — the platform transcribes the conversation, extracts sentiment and topics from the free-form comment, and builds a loyalty-trend dashboard. The survey runs by voice, not by email, so people actually respond.
The problem
Almost no one completes written NPS and CSAT surveys: the link in an SMS or email gets buried, a few percent respond, and the score arrives with no explanation — a "7" and that's it. You can't tell from that sample what exactly to fix, and it's skewed toward the happiest or the angriest.
Even when there is an open-ended comment, no one reviews it systematically: hundreds of answers sit as text, get read selectively, and no one counts the topics. So the company knows its "average score" but not the reasons — what delights customers and what pushes them toward churn.
And calling everyone with live operators is expensive and slow: it's a separate workload, the script drifts from person to person, and the results still have to be tallied by hand. The survey turns into a one-off campaign instead of a constant thermometer of loyalty.
How the platform solves it
- 1
A short survey script
You set up the voice agent's script: a greeting, the score question (NPS 0–10 or CSAT), and one open-ended "why." The score is a required, validated field, so the dialogue always reaches a number.
- 2
The agent calls after a contact
The outbound AI agent calls shortly after a contact, purchase or visit, asks the questions in a natural voice in the right language, and lets the client speak freely (barge-in). The conversation is short — a minute or two.
- 3
Score and open-ended comment
The agent captures the numeric score as a field and records the client's free-form answer. The call recording flows straight into the platform's analysis pipeline.
- 4
Sentiment and topics from the answers
The platform transcribes the conversation and tags it: it computes sentiment, extracts topics and keywords from the open-ended comment, and AI tags ("complaint about delays," "praises the service") are applied by meaning automatically.
- 5
A loyalty-trend dashboard
NPS/CSAT, sentiment and top topics go onto dashboards: trends by week and month, breakdowns by service and operator, period comparison — you see what's rising and what's slipping.
- 6
Alerts and escalating negatives
An automation rule catches low scores and negative sentiment and notifies the owner by email or webhook, and the survey outcome can be written to your CRM — you reach the client before they leave.
The result
A voice survey gives you both the number and the reason: clients respond more readily than to an email, and the platform turns the conversations into sentiment, topics and a loyalty dashboard — with no manual review.
- Voice calling instead of email — people respond more readily, and the score comes with a live explanation.
- Sentiment, topics and keywords are pulled automatically from open-ended answers, not just an average score.
- A trend dashboard for NPS/CSAT by week, service and operator, with period comparison.
- Low scores and negatives are caught by a rule and sent to the owner immediately — you can win the client back in time.
Key facts
- What it does
- A voice AI agent runs a short NPS/CSAT survey after a contact: a score plus one open-ended question.
- What's extracted
- The numeric score as a field, plus sentiment, topics and keywords from the open-ended answer — tagged automatically.
- Analytics
- A trend dashboard for NPS/CSAT and sentiment by week, service and operator, with period comparison.
- Handling negatives
- An automation rule catches low scores and notifies the owner via email or webhook; the survey outcome can be written to your CRM.
- Format
- A voice conversation in the right language (BCP-47), short (a minute or two); the client can interrupt (barge-in).
FAQ
Frequently asked questions
Why is a voice survey better than a written one?
A few percent respond to written NPS/CSAT surveys, and the score arrives with no explanation. The voice agent calls shortly after the contact, asks the questions in a natural voice, and gets not just a number but a live comment — and the platform turns it into data right away. The score is a required, validated field, so the dialogue always reaches a number.
What happens to the client's free-form answer?
The call recording enters the analysis pipeline: the platform transcribes the conversation, computes sentiment, extracts topics and keywords from the open-ended comment, and AI tags by meaning ("complaint about delays," "praises the service") are applied automatically. Hundreds of answers become countable topics instead of a pile of text.
How will I see the loyalty trend?
The numeric score, sentiment and top topics go onto dashboards: NPS/CSAT trends by week and month, breakdowns by service and operator, period comparison and normalization. You see exactly what's rising and what's slipping, and why.
Can I react to unhappy clients immediately?
Yes. An automation rule catches low scores and negative sentiment and notifies the owner by email or a signed webhook, and the survey outcome can be written to your CRM with the crmPush action. That way you reach an unhappy client before they leave.
Ready to start?
Turn every conversation into data, knowledge and action
Start by analyzing your conversations — no risk, no bots required. The platform turns your archive into data and a knowledge base, and voice agents plug in when you're ready.
Free plan, no card required.