OUTBOUND • CALLING • SIP • GUIDE

What is an autodialer and how does it work?

A technical guide to dialing modes, SIP call flow, pacing, agent availability, contact governance and production testing.

AUTODIALER • FUNDAMENTALS

What an autodialer does

An autodialer is a telephone application that selects numbers from an approved contact list and places outbound calls without requiring an agent to dial each number manually. It records the result of every attempt and applies a configured workflow when a call is answered, rejected, busy, unanswered or invalid.

The workflow depends on the purpose of the campaign. An answered call may be presented to an available agent, connected after an agent previews the contact, routed to an IVR, or handled by an approved prerecorded-message process. The dialer may also collect DTMF input, schedule a callback or suppress further attempts after an opt-out.

An autodialer is not a complete call-center system by itself. A production deployment normally combines campaign data, eligibility rules, dialing logic, SIP call control, media handling, agent state, reporting and operational safeguards.

OUTBOUND • CALL • FLOW

How an outbound call moves through the system

The application should make each dialing decision from current campaign and agent state rather than simply walking through a spreadsheet of numbers. A typical SIP-based call follows this sequence:

Select an eligible contact

The campaign service chooses a record only after checking consent or other lawful authority, suppression lists, local time, retry limits and campaign status.

Reserve dialing capacity

The pacing service checks available agents, concurrent-call limits, carrier call-per-second limits and the number of calls already in progress.

Place the call

The call-control service sends an outbound SIP request through an IP-PBX, SIP server or approved trunk and tracks provisional and final SIP responses.

Classify the answer

After answer, the media workflow may evaluate greeting audio, collect DTMF or wait for an agent. Answering-machine detection is probabilistic and requires an uncertain-result path.

Apply the campaign workflow

A live answer can be connected to a reserved agent, passed to an IVR or handled by another permitted workflow. If no agent is available, the system follows the configured abandonment policy.

Store the disposition

The dialer records the attempt outcome, timestamps, selected route and retry decision without placing unnecessary personal or credential data in operational logs.

DIALING • MODES

Preview, progressive, power, predictive and agentless dialing

Dialer names are sometimes used inconsistently. The practical difference is when the call is placed, whether an agent is reserved and how the dialing rate is calculated.

AGENT CONTROL

Preview dialer

An agent reviews the contact record and starts the call. It offers the most preparation time but provides the least automation.

ONE AGENT • ONE CALL

Progressive dialer

The system starts one call when an agent becomes available. The contact is connected to that reserved agent after answer.

FIXED PACING

Power dialer

The dialer places calls at a configured ratio or cadence. It can increase agent occupancy, but pacing must account for real answer rates and agent availability.

DYNAMIC PACING

Predictive dialer

The system adjusts dialing volume from measured call duration, answer probability and agent state. It needs strict abandonment and capacity controls.

NO LIVE AGENT

Agentless dialer

The call uses an automated message, IVR, survey or notification workflow. Consent, disclosure and opt-out requirements can differ from live-agent calls.

RELATED WORKFLOW

Scheduled callback

A callback request can be attached to any dialing mode. It is a scheduling and routing workflow rather than a separate type of dialer.

SYSTEM • ARCHITECTURE

Core components of a production autodialer

Separating campaign, telephony and reporting responsibilities makes behavior easier to test and prevents a slow database or reporting task from blocking the SIP call loop.

CAMPAIGNS

List and campaign service

Imports contacts, validates fields, assigns campaigns, schedules attempts and preserves the source and status of each record.

ELIGIBILITY

Consent and suppression controls

Applies do-not-call, opt-out, jurisdiction, time-zone, retry and purpose-specific rules before a number reaches the dial queue.

CALL CONTROL

SIP session manager

Creates calls, interprets SIP responses, manages routes and caller identity, enforces timeouts and releases every failed or completed session.

MEDIA

Audio, DTMF and answer handling

Negotiates codecs, plays prompts, collects digits and applies measured human-or-machine classification when the workflow needs it.

WORKFORCE

Agent state and routing

Tracks ready, reserved, connected and wrap-up states so answered calls are presented only to agents who can receive them.

OPERATIONS

Disposition and reporting

Records normalized outcomes, retry reasons, route performance, abandonment and capacity data for operations and audits.

PACING • CAPACITY • AGENTS

Dialing speed must follow real system capacity

A dialing ratio is not a fixed performance promise. The safe rate changes with answer probability, average talk time, agent availability, carrier limits and the purpose of the campaign. A system that dials faster than its agents or trunks can handle creates failed calls, dead air and abandonment.

Useful pacing inputs include ready-agent count, reserved calls, calls awaiting answer, recent answer rate, average connection time, average handling time, trunk concurrency, call-per-second limits and media-worker capacity. Predictive models should use bounded rates and a conservative fallback when their measurements become stale.

  • Reserve an agent before connection when the selected mode requires one.
  • Enforce separate call-per-second and concurrent-call limits for every route.
  • Slow or pause a campaign when answer, abandonment or failure rates exceed policy.
  • Keep retry intervals and maximum attempts explicit for each disposition.
  • Measure capacity with realistic ringing, answer and talk-time distributions.
  • Define overload behavior before a database, SIP trunk or media worker fails.

CONTACT • GOVERNANCE

Contact lists and operating rules belong in the call path

Governance checks should run before every attempt, not only when a list is imported. A contact may opt out, a campaign may be paused, a calling window may close or the legal basis for a particular purpose may change after the record entered the system.

Rules vary by jurisdiction, audience, call purpose and whether the call uses a live agent, an automated message or artificial voice. Dialer features can enforce configured controls, but they do not determine whether a campaign is lawful. The operator should obtain qualified advice and maintain evidence for the rules applied to each campaign.

  • Record the source, scope and date of consent or other lawful authority.
  • Screen internal and applicable external suppression lists before each call.
  • Use the recipient's local time and the permitted calling window.
  • Transmit an approved caller identity and provide required disclosures.
  • Honor opt-out requests across active lists and future imports.
  • Define agent availability, abandonment and prerecorded-message behavior.

For examples of current regulatory guidance, review the FTC Telemarketing Sales Rule guide for the United States and the ICO electronic and telephone marketing guidance for the United Kingdom. These references are starting points, not a substitute for campaign-specific legal review.

SIP • DEVELOPMENT

Building an autodialer with a SIP server SDK

A SIP-based dialer normally runs as a server application. The campaign service sends an eligible contact to the call-control layer, which creates the outbound SIP session and receives asynchronous call events. Media, agent routing and database work should remain outside the signaling callback so that one slow task cannot block other calls.

The implementation needs explicit state for queued, dialing, ringing, answered, connected, completed and failed attempts. SIP response codes should map to meaningful dispositions rather than a single generic failure. For example, rejection, busy, timeout, invalid destination and route failure can require different retry decisions.

VaxVoIP SIP Server SDK can provide the SIP call-control layer for a Windows-based autodialer application. The surrounding application remains responsible for campaign rules, contact eligibility, pacing, media workflow, agent integration and reporting.

EVENT MODEL

Keep callbacks short

Update call state and queue longer work instead of performing database, analytics or external API calls in the SIP event path.

IDENTITY

Control routes and caller ID

Associate every campaign with an approved trunk, presentation identity and destination format.

FAILURE

Make cleanup deterministic

Release sessions, agent reservations and media resources after every terminal event, including timeouts and route failures.

OBSERVABILITY

Use normalized dispositions

Separate carrier responses, recipient outcomes, workflow results and agent outcomes so reports remain meaningful.

PREDEPLOYMENT • TESTING

Test the complete campaign path before production

Use controlled numbers and a non-production trunk first. Test the system as a campaign, not just as a function that can place one successful call.

  • Validate imports, duplicate handling, consent data and suppression updates.
  • Test busy, rejected, unanswered, invalid and network-failure SIP outcomes.
  • Verify human, machine and uncertain answer classifications with real samples.
  • Confirm agent reservation, transfer, wrap-up and unavailable-agent behavior.
  • Exercise DTMF, prompt interruption, silence, voicemail and disconnect handling.
  • Load-test call-per-second, concurrency, database and media-worker limits.
  • Verify retries, local calling windows, caller identity and opt-out suppression.
  • Reconcile campaign totals against trunk records and normalized dispositions.

AUTODIALER • FAQ

Frequently asked questions

What does an autodialer do?

It selects eligible numbers, places outbound calls and applies a configured workflow to each result. The workflow may connect an answer to an agent, run an IVR, play an approved message or record a disposition.

What is the difference between a power dialer and a predictive dialer?

A power dialer uses a configured cadence or ratio. A predictive dialer continuously changes its rate from measured answer probability, call duration and agent availability. Predictive pacing requires tighter abandonment controls.

Does an autodialer require an IP-PBX?

Not necessarily. It needs a call-control path to a carrier or SIP service. That path may be an IP-PBX, SIP server, hosted platform or an application connected to an approved SIP trunk.

How does answering-machine detection work?

It evaluates early call audio and classifies patterns such as speech duration, silence and greeting cadence. The result is an estimate, so the workflow needs thresholds and a safe path for uncertain classifications.

How many simultaneous calls can an autodialer place?

The safe limit depends on carrier call-per-second and concurrency limits, server and media resources, bandwidth, agent capacity and campaign policy. Establish it through realistic load testing.

Can an autodialer collect keypad input?

Yes. A media workflow can play a prompt and collect DTMF digits when the platform and endpoint use a compatible DTMF method. Timeouts, invalid input and opt-out behavior must also be specified.

Does autodialer software make a campaign compliant?

No. Software can enforce configured consent, suppression, timing, identity, opt-out and abandonment controls. The campaign operator remains responsible for determining and meeting the rules that apply to its jurisdiction, audience and purpose.

What should be tested before an autodialer goes live?

Test list eligibility, SIP outcomes, answer classification, agent routing, DTMF, retries, caller identity, opt-out, time zones, trunk limits, concurrent calls, failures and reporting with controlled test numbers.