Predictive vs Progressive vs Preview Dialer

Outbound dialers automate calling so agents spend more time in live conversations. Without a dialer, agents manually dial, wait for ring, handle busy/voicemail, and log outcomes. With a dialer, the system paces calls, filters non-human answers via AMD (Answering Machine Detection), and routes live humans to agents. The three modes differ in who controls pacing.

Predictive dialers place *multiple calls per available agent*, using answer-rate algorithms to predict when an agent will be free. The pacing engine samples: average handle time (AHT), answer-seeking ratio (ASR), wrap-up time, and current agent count, then sets a dial ratio (calls per free agent). Example: 5 agents ready, dial ratio 3.0 → 15 calls launched. It recalculates every few seconds. When a live person answers, the call is quickly connected to the next free agent. This maximizes talk time (45–55 min/hour vs 20–30 manual) but risks abandonment if more people answer than agents are ready.

Progressive dialers dial *one number per free agent*, automatically, as soon as an agent becomes available. No over-dialing. If the call goes to voicemail or no answer, the agent waits briefly and the next call is placed for that same agent. No call is ever placed without a ready agent, so abandoned calls are zero. This is steadier, lower risk, and more predictable than predictive.

Preview dialers show the agent the contact's details and CRM history *before* dialing. The agent reviews context and decides when to place the call. This is slowest but gives maximum preparation — ideal for high-value B2B, account management or complex sales where context matters more than speed. The dialer does not auto-dial; agent initiates.

Dial ratio (calls per free agent) is central. Higher ratio → higher utilization but higher abandonment risk (answered call with no agent within ~2 seconds, per TCPA/FTC 3% 30-day rolling cap in US; India TRAI has its own abandoned-call handling but no single 3% stated in current TCCCPR — verify). Progressive has dial ratio 1.0 always. Predictive varies: at 12% ASR, ~8 dials needed per answer; at high ASR, lower ratio suffices. Predictive is *statistically unreliable below 7–10 simultaneous agents* — small pools cause volatile abandonment (per Five9: never use predictive with <15 agents consistently).

| Mode | How It Dials | Pacing Control | Abandonment | Agent Utilization | List Quality Need | Best For | | Predictive | Multiple per free agent (algorithmic) | System | Higher if misconfigured (must be capped) | 45–55 min/hour | Large, clean lists | High-volume telesales, lead gen | | Progressive | One per free agent (auto) | System (1:1) | Zero | 30–40 min/hour | Medium | Collections, appointments, compliance-sensitive | | Preview | Agent-initiated after review | Agent | Near zero | 20–30 min/hour | Small, high-value curated | High-value B2B, complex deals | | Power (variant) | One per free agent, continuous | System | Low | 35–45 | Medium | Steady outbound (often grouped with progressive) |

Campaign: 1,000 leads, 10 agents, ASR 15%, AHT 3 min. - **Predictive**: Dial ratio ~2.5 → ~25 calls in flight, ~3–4 live answers per minute, agents 50 min/hour talk, but 2–3% abandons if ASR spikes. Needs ≥10 agents, large list. - **Progressive**: 10 calls in flight (one per agent), 1–2 live answers per minute, agents 35 min/hour talk, 0 abandons, steady. Works with 5–10 agents. - **Preview**: Agents review 30 seconds, then dial one at a time, 0.5 live answers per minute, agents 22 min/hour talk, 0 abandons, but each conversation is high-context (e.g., B2B deal). Same list, different time-to-completion: predictive ~4 hours, progressive ~7 hours, preview ~14 hours.

Predictive needs large lists (clean, DNC-scrubbed) and consistent agent pool (≥7–10, ideally 15+ per Five9). Small lists or <5 agents → progressive/preview outperform. Voicemail/AMD: predictive benefits from AMD screening; progressive/preview handle voicemail per-agent. List recycling and DNC suppression are required for all modes in India (TRAI).

Predictive maximizes conversations per hour but can feel rushed (back-to-back, little preview). Progressive feels steady (one call at a time, predictable). Preview feels consultative (full context, agent-controlled). Choose based on whether your team values speed, control or preparation.

- **Do NOT use predictive** with <7 agents, small/curated lists, or where abandonment must be zero (e.g., sensitive outreach). - **Do NOT use preview** for high-volume cold outreach (too slow). - **Do NOT use predictive** if you cannot monitor abandonment in real time or enforce pacing caps. - All modes require DNC scrubbing, calling-hour windows and consent handling per TRAI.

All modes must respect TRAI TCCCPR, DND scrubbing and provider rate limits. Predictive abandonment is regulated in some jurisdictions (e.g., 3% in US TCPA); for India, verify current TRAI abandoned-call handling with your provider — do not state a specific India percentage unless verified against current TRAI direction.