Channels
Voice: keypad input, voicemail detection and caller ID
How callers enter numbers privately on the keypad, what the agent does when an answering machine picks up, and how your business name can appear on outbound calls.
On this page
Three details decide how well phone calls work in practice: a private way for callers to enter account numbers and codes, a sensible response when an answering machine picks up an outbound call, and a name customers recognise when you call them. Keypad input, voicemail and answering machine detection, and branded caller ID cover these, and each is configured during implementation for the lines and campaigns that need it.
The three at a glance
| Feature | What it does for your customers | How it works | What you decide |
|---|---|---|---|
| Keypad input | Callers type account numbers, reference codes and PINs rather than saying them aloud | The agent listens for keypad tones alongside speech | Which questions accept keypad input |
| Voicemail and answering machine detection | Outbound calls are not spent talking to a recorded greeting | The agent listens for the signatures of a recorded greeting | Whether the agent leaves an approved message, schedules another attempt, or both |
| Branded caller ID | Outbound calls show your business name, so customers recognise them | Your caller identity is registered with your carrier, where the carrier supports it | The name shown |
Keypad input: entering numbers privately
Some things are better typed than spoken. A policyholder in an open-plan office may not want to read out a policy number, and a customer on a noisy street may find a long reference easier to key in. Keypad input lets the caller type digits on their phone keypad while the agent listens, and they can still say the number instead if they prefer.
Keypad input is configured during implementation for the lines and questions that need it. A typical prompt is: "Please enter your policy number on your keypad, followed by the hash key."
Card numbers typed during a call are redacted, as are card numbers spoken. The agent never takes card details: where a payment is needed, it sends a payment link from your payment provider, so card data stays with that provider. See Personal information redaction.
The technical detail
Keypad presses are sent as DTMF (dual-tone multi-frequency, the tones a phone keypad produces). Telonic receives them as RFC 4733 telephone events by default: each key press travels as a labelled event alongside the call audio, rather than as sound, so it is not affected by background noise or audio compression. SIP INFO is also supported, where each key press is sent as a signalling message on the SIP connection (SIP is the standard protocol phone systems use to set up and connect calls). Your telephony team and ours agree the method your phone system uses when the line is connected.
| Item | Telonic's approach |
|---|---|
| Default method | RFC 4733 telephone events |
| Also supported | SIP INFO |
| Input alongside speech | Yes: the caller can type or say the answer |
| Card numbers | Redacted when typed or spoken |
Voicemail and answering machine detection
When the agent places an outbound call and a recorded greeting answers instead of a person, the agent recognises it and does what you have chosen for that kind of call. It can leave an approved message, end the call and schedule another attempt within your calling hours, or do both. Voicemail and answering machine detection is configured during implementation for the outbound calls and campaigns that use it.
Messages left on voicemail use wording your team approves in advance. Because you cannot be sure who will hear a voicemail, we recommend keeping these messages free of personal and account details: the organisation's name, the reason in general terms and how to get back in touch. The next attempt follows the retry rules set for that campaign. See Voice: inbound and outbound calls.
Branded caller ID: showing your name
Outbound calls from the agent leave through your own SIP trunk (the virtual phone line between your phone system and Telonic) and show your own number. Where your carrier supports it, your business name is registered with the carrier so it can appear on the recipient's phone as well. Customers are more likely to answer a call they recognise than one they might mistake for spam.
Branded caller ID is configured during implementation, with your carrier. Whether the name appears also depends on the recipient's network and handset, so your number stays the dependable identifier. See Telephony and carriers.
In practice
A motor insurer in Dubai runs renewal reminder calls to policyholders whose cover ends next month.
- The insurer's carrier supports branded caller ID, so the reminder calls show the insurer's name as well as its number.
- The agent calls Saeed at 11am. His voicemail answers. The agent recognises the greeting, leaves the insurer's approved message, and schedules another attempt for the next day at 5pm, within the insurer's calling hours.
- The next day, Saeed answers. The agent says it is an AI agent calling for the insurer about his renewal, and asks for his consent to record.
- To confirm who he is, the agent asks him to enter his date of birth on his keypad, one of the identity checks the insurer has set.
- Saeed wants to renew. The agent confirms his renewal terms from the policy system and sends him a payment link by SMS. His card details go to the payment provider, never through the call.
- The outcome, the renewal status and both call attempts are in Saeed's record.
What your team controls
- Which questions accept keypad input, and the method your phone system uses to send it.
- What the agent does when it reaches voicemail: the approved message, another attempt, or both.
- The retry rules and calling hours for each campaign.
- The business name shown on outbound calls, where your carrier supports it.