// GuideDialing
Predictive, power, and preview dialing
Published
Every outbound dialer has to decide when to place the next call. Preview, power, and predictive dialing answer that question in different ways, and the answer sets how long agents wait between conversations, how often someone answers to silence, and how much an agent knows about the person before they pick up. This Impact Dialing guide explains each mode, compares them side by side, and covers which one fits which kind of list.
Preview dialing
In preview mode the dialer shows the agent the next record first, with the contact's name, history, and notes, and the call starts only when the agent clicks to dial or a short countdown runs out. There's one call per agent, and the agent is already on the line when it connects, so a preview dialer never leaves a person answering to silence.
The cost is agent time. The agent waits through every ring, busy signal, and voicemail greeting, so talk time per hour is the lowest of the three modes. Preview suits small or high-value lists where research before the call pays off, such as business accounts, callbacks, and renewals, and programs that want a person starting every call.
Power dialing
A power dialer, which some vendors call progressive, places a call as soon as an agent becomes free, at a fixed number of lines per available agent that a supervisor sets. At one line per agent the dialer only calls for an agent who's already waiting, so abandoned calls are rare, and the agent skips the manual dialing and record lookup that preview requires.
Raising the ratio above one cuts idle time further, because the dialer works through unanswered numbers in parallel. It also brings back the risk that two people answer for one free agent, and the second one becomes an abandoned call. Vendors use these terms loosely, so check what a "power" or "progressive" setting does in your dialer before relying on it. Power dialing fits small and mid-size teams and lists with moderate answer rates, where predictable pacing is worth more than the last few minutes of idle time.
Predictive dialing
A predictive dialer places calls before any agent is free. It estimates when each agent on a call will finish, using the current answer rate, average handle time, and how long each live call has run, and dials enough numbers that an answer arrives as an agent becomes available. On large lists with low answer rates it keeps agents talking for more of each hour than either of the other modes.
The estimate is a statistical bet, so it sometimes loses. When more people answer than there are free agents, the extras are abandoned calls, and for telemarketing those are capped at 3 percent of live answers per campaign in each 30-day period, as abandonment rate and the 3 percent rule explains. The pacing math also needs enough agents logged in to average across, which is why predictive mode tends to swing between idle agents and dropped calls on very small teams. The article on optimizing predictive dialer settings covers the dial ratio and answering machine detection settings that control this trade-off.
How the three modes compare
| Preview | Power | Predictive | |
|---|---|---|---|
| When a call is placed | When the agent clicks or a countdown ends | When an agent becomes free | Before an agent is free, timed to an expected finish |
| Lines per agent | One | One, or a fixed ratio the supervisor sets | Varies with answer rate and handle time |
| Abandoned calls | None | Rare at one line per agent, possible above it | Expected, and capped for telemarketing |
| Agent sees the record | Before dialing | Usually as the call connects | As the call connects |
| Agent idle time | Highest | Moderate | Lowest on large lists |
| Fits | Small or high-value lists, callbacks | Small and mid-size teams | Large lists, low answer rates, larger teams |
Choosing a mode for your list
Start with preview when each contact is worth some research or the list is small enough that agents would work through it quickly anyway. Use power dialing at one line per agent when you want simple staffing and almost no abandoned calls, which makes it a sensible default for a new team or a list with a decent answer rate. Move to predictive when the list is large, answer rates are low, and enough agents are logged in at once for the pacing to settle.
Most dialers let a campaign switch modes, so a mode can follow the list. One workable pattern is to run predictive through the peak calling hours and drop to power or preview for callbacks, small segments, and the last stretch of a list. Whatever the mode, track contact rate, idle time, and abandonment with the formulas in the call center metrics guide, so the comparison rests on your own numbers.
Is a predictive dialer an autodialer?
Under federal law, usually not anymore. In Facebook v. Duguid (2021), the Supreme Court read the TCPA's autodialer definition to cover only equipment that uses a random or sequential number generator to store or produce the numbers it calls, and dialers that work from a list of specific numbers generally don't. The FCC's rule text uses the same definition (47 CFR 64.1200(f)(2)).
That narrowed one source of risk without removing the others. The abandonment limits apply to any telemarketing call a person answers, whatever the mode, and calls that use a prerecorded or artificial voice, including AI-generated voices, still need consent under the TCPA. Some states also define automated dialing more broadly. Florida's telemarketing law, for example, requires prior express written consent for sales calls that use "an automated system for the selection and dialing of telephone numbers." The TCPA and Telemarketing Sales Rule basics guide covers consent in more detail, and none of this is legal advice.