How to choose a dialer for an outbound floor

The four dialing modes

Vendors do not use these names consistently, so ask each one to describe exactly what its modes do. In broad terms:

• Preview: the setter sees the record first and decides when to dial. It is the slowest mode, but every call starts with a person ready, which suits callbacks, warm leads and sensitive follow ups.
• Progressive: the dialer places a call only when a setter is available, usually one line per setter. The person who answers rarely waits, and speed is moderate.
• Power: the term usually means dialing a fixed number of lines per available setter, a ratio you set. It is faster than progressive, but the ratio does not adjust on its own when answer rates change.
• Predictive: the dialer dials ahead of availability and adjusts its ratio continuously based on answer rates, talk time and staffing. It produces the most conversations per hour and carries the most abandonment risk.

Match the mode to the campaign

You may want more than one mode on the same floor: predictive on large cold lists with plenty of seats, preview or progressive for callbacks, small teams and high value records. A dialer that lets you choose the mode per campaign is more useful than one locked into a single style. Small floors should also check whether predictive pacing behaves sensibly with only a few setters logged in.

Compliance controls come first

A dialer that books more appointments while putting the floor at legal risk is not a better dialer. Before comparing speed, check how it handles the rules that apply to automated outbound calling.

• Abandon rate: does it enforce a cap per campaign, measured the way the FTC and FCC measure it, and slow down before the limit?
• Abandoned call message: can it play a compliant identification message within two seconds, with an automated opt out?
• Ring time: does it let unanswered calls ring at least 15 seconds or four rings?
• Calling hours: does it block calls based on the called person's local time, not your office clock?
• Do not call: can it suppress your internal list and registry scrubs automatically, and record opt outs taken on calls?
• Records: can you export call level data that shows what happened, months later?

Know what the AI is doing

Many dialers now advertise AI features, and the term covers very different things, from transcribing calls to speaking with the person who answers. If a product uses AI generated voices on calls, the FCC said in 2024 that such voices count as artificial voices under the TCPA, so the consent rules for artificial or prerecorded voice calls apply. Ask the vendor exactly what the AI does on a live call and what it does behind the scenes, and confirm your consent practices with counsel.

Carriers and caller ID

Connect rates depend as much on carriers and caller ID as on the dialer. Ask who carries the calls, what STIR/SHAKEN attestation level your calls will receive, and whether you can bring your own numbers and carrier. Find out how numbers are assigned to campaigns, whether callbacks to those numbers reach you, and what tools exist to check how your numbers display on the major carriers. A dialer that treats carriers and caller ID as someone else's problem leaves you to debug connect rates alone.

Data, lists and integrations

Setters can only call what gets loaded. Check how lists are imported, how duplicates are handled, whether each record can carry its source and consent details, and whether do not call and litigator scrubs can run before a list goes live. Look at how dispositions, callbacks and booked appointments flow to your CRM or calendar, and whether setter notes travel with the appointment to the closer.

The setter's screen and the manager's view

Setters live in the dialer for hours a day, so watch one use it. Is the script visible? Is the record on screen before the person says hello? How many clicks does it take to book an appointment or set a callback? Small frictions repeated across thousands of calls become a real cost.

Managers need reporting that answers floor questions without a spreadsheet: dials, contacts, conversations, sets and abandon rate by campaign, setter, list and hour. Live monitoring, call recording and coaching from recordings matter for quality. Check the recording consent rules in the states you call, since some require the consent of every party.

Run a real pilot

Demos run on perfect data. Before you move the floor, run a pilot on a few seats with real lists and real campaigns, and compare it against your current setup over the same period. Track contact rate, conversations per setter hour, set rate, abandon rate and how often the dialer or audio failed. Ask what support looks like when something breaks during calling hours, and who you actually talk to.

Also ask who shapes the roadmap. Tools built without operator input tend to ship features nobody uses while the gaps everybody feels stay open.

If you want to pilot an AI layer on top of a predictive dialer, designed with operators who run outbound setting floors, Telgrow's page for call center owners is where early access is opening to a limited number of floors.

Questions