The auto dialer mechanism, explained — from list upload to logged disposition
How auto dialers work: the call lifecycle from list to disposition
Every auto dialer — power, predictive, preview, single-line or multi-line — runs the same six-stage loop: a record comes off the list, a pacing engine decides when to dial it, a carrier places the call, the system classifies what answered, a live answer is bridged to an agent, and the outcome is logged as a disposition that decides the number’s next step. The modes differ only at stage two; the rest is shared machinery. This page walks the lifecycle, opens up predictive pacing technology, covers power and preview mechanics, and ends with a five-step setup. For the taxonomy of types, see the auto dialer guide and dialing-modes comparison in the related links.
The call lifecycle: six stages every dialer runs
| Stage | What happens | What can go wrong |
|---|---|---|
| 1. List | The campaign pulls the next eligible record from your list | Duplicates, DNC numbers, and dead records waste everything downstream |
| 2. Pacing engine | Mode logic decides how many calls to launch, and when | Predictive overshoot abandons calls; undershoot idles agents |
| 3. Carrier | The call is routed and placed on the phone network | Worn caller IDs get labeled or ignored before anyone hears a ring |
| 4. Connect detection | The system classifies the outcome: live voice, machine, busy, invalid | Misclassification hangs up on a person or feeds agents voicemail |
| 5. Agent bridge | Live answers are routed to an available agent | Connect delay before the agent joins burns goodwill |
| 6. Disposition | The outcome is logged and mapped to retry, callback, or retire | Sloppy wrap-up codes cycle dead numbers back into the list |
Stages 1–3: list, pacing, carrier. The dialer works through whatever you imported, in whatever shape — which is why list prep is setup step one. The pacing engine is where modes diverge: power waits for each call, multi-line runs several calls per agent, predictive paces the whole team statistically. The carrier stage then tests caller ID health: a worn number can be labeled or screened out before anyone decides whether to answer.
Stage 4: connect detection. The system classifies each resolved dial — Genesys Cloud’s outbound docs list live voice, answering machine, busy, no-answer, invalid number, and SIT tones marking a number uncallable (Genesys Cloud Resource Center, July 2026). Many predictive platforms add answering machine detection — Twilio’s engineering docs describe the predictive workflow as invoking the agent only when a human, not a machine, answers (Twilio Voice docs, July 2026). Detection is probabilistic — its error modes are covered in our answering machine detection guide. Enzo doesn’t use AMD: every answered call, in every Enzo mode, goes to a live agent.
Stages 5–6: agent bridge and disposition. A live answer is bridged to an agent — already on the line in power and preview, routed to whoever is free in predictive. The agent then logs a disposition, the stage most teams underrate, because dispositions keep the list alive. Genesys Cloud documents the mechanism: wrap-up code mappings translate each disposition into list actions — stop dialing every number for the contact, retire just the dialed number, or mark the right party reached (Genesys Cloud Resource Center, July 2026). Sloppy dispositions cycle dead numbers back around; clean mappings sharpen the list every session.
Predictive dialer technology: how the pacing engine thinks
Predictive dialer technology is the layer that earns the word “predictive”: instead of reacting to an agent becoming free, it launches calls ahead of availability, betting on when agents will finish — and the bet is statistical. Genesys Cloud describes its predictive mode as a stage-based algorithm fed by average handle time, after-call work, idle time, and historical contact rate (Genesys Cloud Resource Center, July 2026); NICE lists connect rates, talk time, agent breaks, and abandon rates, recalculated through the day (NICE glossary, July 2026); Convoso adds that predictive runs a variable call ratio and disconnects unsuccessful dials before an agent is involved, with some systems taking a programmed target drop rate (Convoso, July 2026).
All of that machinery manages one tension: agent occupancy against abandonment. Pace hot and agents stay busy but some people answer into silence; pace cold and abandons vanish but agents idle. There’s a legal boundary — U.S. telemarketing rules (47 CFR 64.1200) cap abandoned calls at 3% of telemarketing calls answered live by a person, per campaign over each 30-day period — and the rest of the legal picture lives in the TCPA autodialer rules guide. Statistics need volume, too: Genesys recommends roughly 15 agents for predictive accuracy, Five9 positions predictive for 10+ active agents — vendor-specific guidance, not an industry rule. For the pacing tradeoff in full, see the predictive dialer pillar.
How a power dialer works — and why the term needs a vendor check
The power dialer mechanic most sales teams mean is sequential: dial one number, wait for the outcome, talk if someone answers, then dial the next record when the call wraps. The agent is on the line for every dial, so a live answer meets a human with no pause — the mode trades throughput for a clean connect on every call.
Before you compare products, though: “power dialer” is not a standardized term. Five9 defines power dialing as a user-configured fixed ratio of calls per available agent, adjusted on abandon rates; Genesys likewise uses “power” for multi-number dialing per idle agent, while Genesys and NICE reserve “progressive” for strict one-call-per-agent pacing — and other vendors brand single-line products as power dialers (vendor docs, July 2026). Same word, different machines. Enzo’s map is plain: power dialing on Enzo is agent-paced, single-line or multi-line, with a live agent on every answered call.
How a preview dialer works
Preview mode inverts the automation: the system presents the contact’s record first, and the agent reviews it and decides to call or skip — Genesys Cloud’s canonical definition (July 2026); Five9’s version attaches CRM history. The dial then follows the same lifecycle as every other mode. It is the slowest mode by design, and that’s the point: preview fits lists where thirty seconds of context changes the conversation — past clients, referrals, expireds, callbacks. The ideal is preview as a campaign setting rather than a separate product, so cold and warm lists don’t need two vendors.
How to set up an auto dialer in 5 steps
The mode changes the pacing settings, not the sequence — these five steps work on any platform.
1. Prepare and scrub the list
Export to CSV, dedupe, and map names, numbers, and context fields to real columns. Scrub against the National DNC Registry and state lists through a third-party service before upload — Enzo’s internal DNC lists live at the campaign level, don’t carry across campaigns, and Enzo does not scrub registries, so this step stays yours.
2. Provision caller IDs
Volume dialing concentrates wear on the numbers you dial from. Provision enough caller IDs for the planned volume, matched to the regions you call, and decide who watches reputation. On Enzo this is managed: 35 caller IDs per seat on Starter, 100 on Standard, monitored and rotated or swapped when reputation dips.
3. Build the campaign
Create the campaign, attach the list and caller ID group, and pick the mode the list deserves — predictive for a big team on a long cold list, power or preview for warmer, higher-stakes records. Set calling windows that respect time zones and state calling-hours rules, and start pacing conservatively.
4. Load scripts and dispositions
Put the script where the call happens, and define the disposition set before the first dial: which outcomes retry, which schedule callbacks, which retire the number or contact. This stage decides whether week two dials a cleaner list than week one.
5. Launch small and tune
Run a short first session and watch: abandon rate if you’re running predictive, connect outcomes, agent wait time. Adjust pacing, lines, and windows from evidence rather than defaults, then scale up.
On Enzo, the $250 white-glove buildout runs this sequence with you — carrier registration, list import, campaign and caller ID group setup, and live training — waived on annual plans, with the full ladder at /pricing: $99/seat/month billed annually, $120 month-to-month, no seat minimum.
Where Enzo fits in the machinery
Enzo is cloud-based, with the carrier side provisioned and managed by Enzo — no hardware, agents work from a browser. The pacing engine runs power, predictive, and preview modes in single-line and multi-line form: 5 lines per agent on Starter, up to 14 on Standard, pooled, so two 14-line agents dial as one 28-line session. Every mode connects a live agent to every answered call — no AMD, no prerecorded voice, no AI voice. All minutes are included; dispositions flow to your CRM through native two-way Follow Up Boss sync, with GoHighLevel, Salesforce, HubSpot, and roughly 6,000 other tools connected via Zapier and webhooks. Book a free discovery call to see the lifecycle running — 20 minutes, and if Enzo isn’t the right fit, we’ll say so.
Mechanism facts are drawn from the Genesys Cloud Resource Center (dialing modes, wrap-up codes), Five9’s public dialer-system pages, NICE’s glossaries, Convoso’s predictive-vs-power guide, Twilio’s AMD docs, and 47 CFR 64.1200, as of July 2026. Vendor behavior and terminology change — verify with each vendor before you buy.