Preview, power, predictive: which dialing mode when.

Lede

Outbound calling is a trade between two kinds of waiting. Either the agent waits for the phone, dialing and listening to rings and voicemail greetings, or the customer waits for the agent. Every dialing mode is a different answer to who should wait, and for how long.

Our dialer has five: manual, preview, progressive, power and predictive. They share one set of rules. Whatever the mode, every number is checked against your do-not-call list and the platform's before it is dialed, and every contact is called only inside the calling window, 08:00 to 21:00, in every time zone the contact could be in. What changes from mode to mode is who presses the button and how many lines ring per agent. Nothing in this note is legal advice; talk to counsel about your own campaigns.

Manual and click-to-call

The simplest mode is the one every agent already knows. The agent types a number into the dialpad, or clicks one anywhere in the app, and the call is placed from the agent's browser. No campaign, no pacing. One agent, one call.

Manual dialing still runs the checks. An emergency number is refused outright. A number on a do-not-call list, or a contact outside calling hours, is blocked or flagged with a warning, depending on how your admin has set the policy. An agent who overrides a warning gives a reason, and the override is recorded. Calls placed through the API are stricter: they are always blocked.

Manual is the right mode for callbacks, follow-ups and anything an agent decides to do on their own. It is on every plan.

Preview

Preview is a campaign mode where a person still decides. The dialer reserves an available agent, picks the next lead and shows the contact before anything rings. The agent reads, then clicks Call. Or skips: a skipped lead goes back into the list for a later try, and the skip doesn't count as an attempt.

A campaign can also set a preview timeout. A countdown then dials on its own once the agent has had long enough to read.

Preview is slow on purpose. Use it when the first ten seconds of a call matter more than the number of calls per hour: large accounts, sensitive conversations, calls where the agent needs to know what happened last time before saying hello.

Progressive

Progressive removes the click. As soon as an agent is available, the dialer places one call for them. One line per available agent, never more. The agent hears the ring, the customer answers, and the agent is already there.

Because progressive never dials more calls than there are free agents, it never abandons a call. Nobody picks up to silence. That makes it a safe default for most outbound work, and it is what the dialer falls back to whenever predictive can't run safely. Preview and progressive are both on every plan.

The persistent line

Power and predictive dialing work differently. They dial customers before an agent is assigned, then connect whoever answers to an agent who is free at that moment. For that to happen without a gap, the agent's audio has to be connected already.

So agents who join a power or predictive campaign open a persistent line. Their browser stays connected, parked, between calls. When a customer answers, the dialer bridges the customer straight onto the line. Nothing rings on the agent's side; the next conversation simply starts. The workspace shows the state of the line: "Line connected", or "Line disconnected" with a way to reconnect.

A parked line uses carrier minutes even when it is quiet, so the dialer hangs it up when it isn't needed: after two minutes on break, two minutes after the campaign pauses or completes, when the agent leaves the campaign, or after ten minutes without a call. Line minutes count toward your account's carrier-spend ceiling. Agents on a persistent line aren't offered inbound calls either. In this version an agent is on the campaign or on the queues, not both.

Power dialing places a fixed number of lines per available agent, which you set on the campaign. It suits long lists with low answer rates and a team that would rather spend its hours talking than listening to rings.

Predictive does the arithmetic for you. It watches how often calls are answered, how long calls and wrap-up take and which agents are about to be free, and places enough calls to keep them busy without leaving answered calls unattended. It needs at least five agents with connected lines. With fewer, the campaign runs as progressive until more join.

Machines and messages

On power and predictive campaigns, answering-machine detection listens to the start of each answered call. If a person picked up, nothing changes. If a machine did, the agent is released back to the campaign and the dialer does what the campaign says: hang up, leave the agent on the call anyway, or wait for the beep and play a prerecorded voicemail message.

That last option comes with a rule. The dialer plays a prerecorded message only to contacts who gave express written consent, or when your company has recorded an attestation in the app. Otherwise a detected machine is simply hung up on, and the agent's own "Drop VM" button is disabled with the reason written on it.

Three percent

When predictive guesses high, a customer answers and no agent is free. That is an abandoned call, and US telemarketing rules cap abandoned calls at three percent of answered calls, measured over thirty days.

The pacing controller aims for 2.5 percent. Once a campaign has enough answered calls to measure, a thirty-day rate of 2.8 percent stops the predicting: the campaign runs as progressive until the rate falls back below 2.3 percent. A spike within the last hour brakes it too. And a customer who waits more than two seconds for an agent after answering counts as abandoned, not connected.

When no agent is free, the customer is never left in silence. They hear a short message naming your company and a number to call back, and they can press 9 to join your do-not-call list. That is why a power or predictive campaign can't start until you've entered a seller name and a callback number. Every campaign dial also rings for at least twenty seconds before it gives up.

Which plan, which mode

The modes map onto the plans like this:

• Starter: manual and click-to-call, preview and progressive campaigns.
• Growth: everything in Starter, plus power dialing, answering-machine detection and the agent's voicemail drop.
• Enterprise: everything in Growth, plus predictive dialing and automatic voicemail drop.

Predictive is in beta. The campaign form labels it that way, and it becomes available on Enterprise accounts once the beta is switched on.

If you are choosing, a short guide:

• Calls that need reading first, or a floor of one or two agents: preview.
• A small team that wants no idle dialing and no abandonment risk: progressive.
• Long lists, low answer rates and a steady team: power.
• Five or more agents on lines, lists big enough to keep them busy, and someone who watches the abandonment rate: predictive.

Whichever you pick, the rules come along. A campaign stops dialing on its own when every remaining lead is outside its calling window, when no agent is available, or when your account reaches its usage allowance, and it starts again by itself when the reason clears.