Voice cancellation surveys
Capture the cancellation story, not just a reason code
Let customers answer by voice or text, ask follow-up questions grounded in their words, and keep the complete raw conversation alongside subscription context.
What this workflow actually does
A multiple-choice cancellation survey is easy to chart but often loses the explanation. Voice makes it easier to describe a pricing mismatch, missing workflow, reliability problem, or change in circumstances without forcing the customer to type a paragraph.
Paydirt transcribes the answer and treats that text as the customer’s response. The next question can clarify what “too expensive” means, which feature failed, or what outcome the customer expected. Text always remains available for people who do not want to record.
The Slack destination receives the exact accepted wording and an input marker. A summary can help scanning, but it is appended only after the raw Q/A, so the original evidence stays visible.
A strong fit for
- ✓ Open-ended cancellation research
- ✓ Mobile contexts where speaking is easier than typing
- ✓ Follow-up questions that depend on the previous answer
- ✓ Teams reviewing a small, high-signal stream in Slack
Not designed to replace
- — Silent environments where users cannot record
- — Quantitative research that requires fixed answer scales
- — Emergency support or time-sensitive account recovery
- — Recording without clear consent and a text alternative
From trigger to exact answers
01
Explain recording before capture
Show clear consent copy and keep the text path available. Microphone permission is requested only for voice.
02
Record a bounded answer
The SDK records to a protected temporary file and enforces a maximum duration rather than recording indefinitely.
03
Transcribe and follow up
The local recording is removed after transcription succeeds or fails; the accepted transcript drives the next question.
04
Preserve the evidence
Each turn is saved, while Slack gets one completed conversation with Audio or Text shown beside each answer.
Voice permission is explicit
The host app adds a purpose string that matches the actual feedback use. Paydirt presents the recording disclosure and text fallback in the form.
<key>NSMicrophoneUsageDescription</key>
<string>Record voice feedback about your experience</string>
// Text feedback remains available when the person does not record.
Paydirt.presentForm(formId: "CANCELLATION_FORM_ID")What reaches Slack
- • The exact transcript accepted as the customer’s answer
- • The question that produced each answer
- • Audio or Text as the input mode—not the audio file in Slack
- • Subscription and placement metadata when supplied
- • Optional summary below the entire raw conversation
Know the limits
- • Speech transcription can contain errors, so the UI should make the accepted text clear
- • Voice requires microphone permission and connectivity; text is the fallback
- • Free-form audio may include personal or sensitive information
- • Voice improves depth, but it does not make a small sample statistically representative
Questions developers ask
Does Slack receive the raw recording?
No. Slack receives the accepted answer text and an `I: Audio` marker. The local temporary recording is deleted after the transcription attempt.
Can a customer answer without microphone access?
Yes. Text is a first-class fallback, and Slack marks that answer as `I: Text`.
Is the AI summary the source of truth?
No. The exact Q/A transcript is primary. If enabled, the summary is added only after the raw conversation.
Ship the smallest useful feedback loop
Install one form at a real product moment, send the completed raw conversation to one Slack channel, and verify it in the app before expanding.
Start setup