🇬🇧 EN

📞 Phone Call

Places an outbound phone call using Twilio. Supports three modes: speak a message, listen for a voice response, or run a questionnaire over the phone.

Category: Communicate · Type identifier: phone_call

Overview

The Phone Call node dials an outbound call through Twilio and drives the conversation according to one of three modes. Each mode is suited to a different interaction pattern:

Speak — the node delivers a text-to-speech message to the recipient. You can optionally wait for a DTMF keypress (digits pressed on the dial pad) to collect a simple confirmation or menu selection before hanging up.
Listen — the node speaks a message and then records and transcribes the recipient's verbal response. The transcription is returned as output for use in later steps.
Questionnaire — the node runs an existing Flusso questionnaire flow over the phone call. Questions are spoken in sequence and the recipient's answers (DTMF or voice) are collected and returned as structured output.

Prerequisite: A Twilio account must be connected under Settings → Twilio before this node can place calls. The Twilio phone number must have voice capabilities enabled and must be verified for outbound calls to the destination country.

Configuration

Field	Status	Description
Mode	Required	`speak` — deliver a TTS message, optionally collecting a DTMF keypress. `listen` — speak a message then record and transcribe the recipient's response. `questionnaire` — run a Flusso questionnaire flow over the call.
To Number	Required	The E.164-formatted phone number to call, e.g. `+12025551234`. Supports `{{ variable }}` references.
Message	Required for speak / listen	The text to speak to the recipient when the call is answered. Supports `{{ variable }}` references. Not used in questionnaire mode.
Voice	Optional	The voice to use for text-to-speech. Select an ElevenLabs voice configured under Settings, or leave blank to use the default Twilio TTS voice.
DTMF Enabled	Optional	When enabled in `speak` mode, the call waits for the recipient to press digits on their dial pad after the message is delivered. Default: disabled.
DTMF Timeout	Optional	How long to wait for a keypress after the message ends, in seconds. If no key is pressed within this time, the call ends. Default: 5 seconds.
Max Digits	Optional	The maximum number of digits to collect in DTMF mode. Collection stops as soon as this many digits are received or the timeout is reached. Default: 1.
Record Max Length	Optional	Maximum recording length in seconds for `listen` mode. The recording stops automatically when the recipient stops speaking or this limit is reached. Default: 30 seconds.
Transcribe	Optional	When enabled in `listen` mode, the recorded audio is automatically transcribed to text and returned in `output.transcript`. Default: enabled.
Questionnaire ID	Required for questionnaire mode	The ID of the Flusso questionnaire to run over the phone call. The questionnaire must be published and configured for phone delivery.

Output Data

Output fields depend on the selected mode and the outcome of the call.

Field	Type	Description
`status`	string	Final call status. One of: `completed`, `no_answer`, `busy`, `refused`, `failed`, `timeout`.
`duration`	number	Call duration in seconds from answer to hangup. `0` if the call was not answered.
`transcript`	string	Text transcription of the recipient's voice response. Present in `listen` mode when Transcribe is enabled.
`recording_url`	string	URL of the recorded audio file. Present in `listen` mode.
`ended_by`	string	Who ended the call: `caller` (Flusso), `recipient`, or `timeout`.
`dtmf`	string	The digit(s) pressed by the recipient. Present in `speak` mode when DTMF is enabled.
`response_text`	string	Alias for `transcript` in `listen` mode. Provided for convenience in template expressions.
`questionnaire_answers`	object	Key-value map of question IDs to the recipient's answers. Present in `questionnaire` mode.
`call_outcome`	string	A structured outcome label set by the questionnaire flow or by a downstream Logic Gate, e.g. `confirmed`, `declined`.

// Reference call outputs in later steps {{ call_step.output.status }} {{ call_step.output.duration }} {{ call_step.output.transcript }} {{ call_step.output.dtmf }} {{ call_step.output.questionnaire_answers.q1 }}

Example Usage

Simple appointment reminder (speak)

A Schedule Trigger fires one hour before each appointment pulled from a calendar integration.
A Phone Call node in speak mode calls the patient. The Message field reads: "Hi {{ appointment.output.patient_name }}, this is a reminder that your appointment is at {{ appointment.output.time }}. See you soon."
A Logic Gate checks call_step.output.status. If no_answer, a follow-up SMS is sent via an HTTP Request node.

Collect confirmation with DTMF

A Phone Call node in speak mode with DTMF Enabled delivers: "Press 1 to confirm your delivery, or press 2 to reschedule." Max Digits is set to 1.
A Switch node routes on call_step.output.dtmf: value 1 → confirm branch, value 2 → reschedule branch.

Record customer feedback (listen)

A Phone Call node in listen mode delivers: "Please leave a brief message describing your experience." Record Max Length is set to 60 seconds and Transcribe is enabled.
An Agent node receives {{ call_step.output.transcript }} and classifies the sentiment, writing the result to a Google Sheet.

Run a survey (questionnaire)

A pre-built questionnaire is created in Questionnaires with several questions for a post-service NPS survey.
A Phone Call node in questionnaire mode is configured with the questionnaire ID and the customer's phone number.
After the call, questionnaire_answers and call_outcome are passed to a Transform node to shape the data before writing it to a CRM via HTTP Request.

Tips & Notes

Phone numbers must be in E.164 format (+ followed by country code and number). Use a Transform node to normalise numbers that arrive in other formats.
Test calls with a verified Twilio test number before enabling in production to avoid charges.
In listen mode, transcription accuracy depends on call audio quality. Background noise or poor signal may reduce accuracy. Consider prompting the recipient to speak clearly in the message.
DTMF collection ends as soon as Max Digits is reached — set it to 1 for simple yes/no or menu prompts to avoid waiting for the timeout.
The call_outcome field is only populated when the questionnaire flow or a downstream step explicitly sets it. It is null for basic speak and listen calls.

Related Nodes

AI Phone Call — free-form AI-driven phone conversation using an Agent node.
SMS — send a text message instead of placing a voice call.
Switch — route the workflow based on DTMF input or call status.
Logic Gate — branch on whether the call was answered or completed successfully.