A predictive dialer is the most aggressive dialing tool you can legally run. It places several calls per available agent, uses pacing math to guess when someone will be free, and accepts that some answered calls will have no agent to take them. With 30 agents working an aged consumer list, that tradeoff can double talk time per hour. With four agents and a warm inbound list, the same tradeoff mostly produces abandoned calls, angry prospects and regulatory exposure.

This post is for ops and call-center leads building a shortlist. It covers the three ways to get a predictive dialer (a hosted outbound vendor, a contact-center suite, or an open-source stack like Vicidial on Asterisk), what each one costs you in money and staff time, how to run a trial that tells you something, and when a progressive dialer or plain click-to-call is the better choice. We make Ready, and Ready does not include a predictive dialer. We say where it fits near the end, and the comparison stands without it.

What a predictive dialer actually does

The dialer watches how long agents spend on calls and in wrap-up, along with your list's answer rate. It then dials ahead. If the answer rate is roughly 1 in 5 and the system expects two agents to be free in eight seconds, it might launch eight to ten calls now. Answering-machine detection filters out some voicemails. Live answers go to whichever agent is free.

When the guess is wrong, a live person answers and no agent is available. That is an abandoned call. In the US, the FTC's Telemarketing Sales Rule caps abandonment for telemarketing calls at 3% of answered calls per campaign over a 30-day period. It also requires connecting to a live agent within two seconds of the greeting, or playing a recorded message that identifies the caller. TCPA consent rules for calls to cell phones apply on top of that. Our post on whether auto dialers are legal covers those rules in detail. Treat this section as context, not legal advice.

If the terms are blurry, see predictive dialer vs auto dialer. The short version:

  • Click-to-call: the agent clicks a number and the call is placed. One call at a time, with the agent listening.
  • Progressive dialer: the system dials the next contact automatically when the agent is ready. Still one line per agent, so no abandoned calls.
  • Predictive dialer: the system dials multiple lines per agent based on statistics. This mode creates abandoned calls and the compliance work that comes with them.

The three ways to get a predictive dialer

1. Hosted outbound dialer vendors

These are companies whose main product is outbound calling for sales and collections. Convoso is a common name on these shortlists, and several smaller vendors serve insurance, solar, debt and real estate teams. You get a web agent interface, list management, pacing controls, abandonment reporting, caller ID management and a carrier connection, all running on their infrastructure.

Strengths: you can be live within days, someone else handles the telephony, and abandonment tracking and DNC handling are usually built in. Weaknesses: pricing is often per seat per month plus per-minute usage, contracts may have minimums, and your data and call recordings sit in their system. Ask how easy an export is before you sign.

2. Contact-center suites

Five9, Genesys Cloud, NICE CXone and Talkdesk sell full contact-center platforms. Inbound routing (ACD/IVR), workforce management, QA, omnichannel and outbound dialing including predictive mode all come in one product. Predictive is one feature among many, and you pay for the platform.

Strengths: depth, reporting, and a single system if you also run an inbound support floor. Weaknesses: longer sales cycles, an implementation project, and a price that rarely makes sense for a team that only dials outbound. If outbound calling is your whole business, a suite is usually more product than you need.

3. Open source: Vicidial on Asterisk

Vicidial is the open-source predictive dialer most people mean when they say "open source." It runs on Asterisk, the open-source PBX, and is commonly installed from ViciBox, a packaged Linux distribution. The license costs nothing. It supports predictive (adaptive) dialing, inbound queues, recordings, agent and manager screens, and extensive reporting. It has been in production at call centers for many years.

Asterisk on its own is a telephony toolkit, not a dialer. If you "build on Asterisk" without Vicidial, you are writing the pacing algorithm, the agent UI, list handling and abandonment tracking yourself. FreeSWITCH is the other common open-source telephony engine for custom builds. Both are capable, and both take real engineering to turn into a compliant predictive dialer.

Strengths: no per-seat license, full control, your data on your servers, and a large community. Weaknesses: you run everything. That means servers, SIP trunks, caller ID reputation, security patches, scaling the database as call logs grow, and making sure your configuration keeps abandonment under the limit. Many teams end up paying a hosted-Vicidial provider or a consultant, which turns "free" into a services bill.

Side-by-side comparison

Hosted outbound vendorContact-center suiteVicidial (self-hosted)Custom on Asterisk/FreeSWITCH
Predictive modeYes, core featureYes, one moduleYes, built inOnly if you build it
License modelPer seat + usage, oftenPer seat, often contractFree (GPL)Free engine
Time to first callDaysWeeks to monthsDays to weeks with a skilled adminMonths
Who runs telephonyVendorVendorYouYou
Abandonment reportingBuilt inBuilt inBuilt in, you verify configYou build it
Best fitOutbound-only teams, ~10+ agentsLarge blended inbound/outbound floorsTeams with a Linux/VoIP admin on staffProducts where the dialer is the product

We're not quoting vendor prices here. They change, they're negotiated, and many vendors don't publish them. Get written quotes, and make sure each one includes per-minute usage, number costs and any minimums.

Build vs buy: the real cost of "free"

The Vicidial license is free, but running it is not. Here is a worked example with stated assumptions. Swap in your own numbers.

Example assumptions: 15 agents, self-hosted Vicidial, one server cluster, SIP trunks billed per minute by your carrier, and one person (in-house or contracted) spending part of their week on administration.

The cost lines you'll actually see:

  1. Servers: cloud or colocated hardware sized for concurrent calls and recording storage.
  2. SIP trunking and numbers: per-minute carrier charges plus number rental. You pay these with any option. A vendor just bundles them.
  3. Admin time: this is the line people underestimate. If your admin spends 8 hours a week on patching, trunk issues, caller ID rotation, reports and agent problems, that is about 400 hours a year. Multiply by your loaded hourly cost to see the real figure.
  4. Compliance configuration: DNC lists, calling-hour windows by time zone, abandonment monitoring and recording-consent prompts. These are all possible in Vicidial, and they are your responsibility to set up correctly and keep working.
  5. Downtime cost: if the dialer goes down at 10 AM on a Tuesday, 15 agents sit idle until your admin fixes it.

The usual pattern: self-hosting wins when you already employ someone who knows Linux and VoIP and your seat count is high enough that per-seat licenses would cost more than that person's time. It loses when "the admin" is a sales manager learning Asterisk from forums. A full custom build on Asterisk or FreeSWITCH only makes sense if you're a software company shipping a dialer as your own product.

How to run a trial that tells you something

A vendor demo uses a clean list and ideal pacing. Your trial should use your own data:

  • Use your real list and your real caller ID. Answer rates on your aged leads are what drive the pacing math, so the vendor's numbers won't predict yours.
  • Watch the abandonment rate every day, not just the average. Ask how the system enforces the 3% cap automatically, whether it slows pacing as you approach it, and whether the safe-harbor recorded message is configured.
  • Measure agent talk time per hour, not dials per hour. Dials are vanity. Talk time and conversions are what you pay agents for.
  • Check caller ID health. Ask how the vendor handles STIR/SHAKEN attestation and spam labeling, and what happens when a number gets flagged. Rotating numbers to dodge flags is a reputation problem, not a fix.
  • Test answering-machine detection accuracy. Aggressive AMD hangs up on real people. Listen to a sample of AMD-classified calls.
  • Export everything once. Pull recordings, call logs and dispositions out during the trial. If that's painful now, leaving later will be worse.
  • Scrub before you dial. Run the trial list against DNC and known TCPA-litigator lists first. Statutory TCPA damages are often cited at $500 to $1,500 per violation, so a few hundred bad numbers can cost more than the dialer.

Our checklist on dialer software for call centers has more vendor questions, and cloud-based dialer software covers hosting tradeoffs.

When a progressive dialer (or less) is the better choice

Predictive pacing depends on statistics, and statistics need volume. A common rule of thumb is that predictive dialing starts to pay off at roughly 10 or more agents working the same campaign. Below that, the system has too few agents to pace against, so it either dials conservatively and gains little, or dials aggressively and abandons calls. Treat that number as an approximation, not a law.

A progressive dialer, or plain click-to-call, is usually the better choice when:

  • Leads are warm or inbound. A person who filled out a form two minutes ago should get a rep immediately, not a two-second silence and a recorded message.
  • You have fewer than about 10 agents. You get most of the efficiency gain with no abandoned calls.
  • Calls are consultative. Long, high-value conversations gain little from faster dialing.
  • Your compliance team wants a simpler story. One call per agent, every call connected to a person, no abandonment to report.

Where Ready fits (and where it doesn't)

Ready is not a predictive dialer, and it doesn't auto-dial at all. The Ready Dialer is click-to-call. A rep dials from the pad in the Ready dashboard or inside GoHighLevel. It includes consented call recording with transcripts, inbound calls that ring the open pad, and a Missed call card in the Inbox thread for each call you miss. There is no auto dialing, no queue dialing and no voicemail drop. If you need predictive pacing for a 40-seat floor, use one of the options above.

Where Ready fits is the warm-lead workflow that predictive dialers handle badly: text a consented lead right away, then call them by hand. Billing is $0.025 per minute, and each started minute is billed as a full minute. Only answered time is billed, so ring-outs and busy lines cost nothing. A call answered by voicemail counts as answered.

Worked example: a rep makes 100 manual dials in a day. Assume 20 are answered (people plus voicemails), each lasting 3 started minutes. That's 20 × 3 × $0.025 = $1.50 in calling. The 80 unanswered dials are free. If that rep also texts the 20 leads who opted in on your form, using a one-segment message each, the cost is 20 × ($0.0075 + $0.0045) on the Standard rate, under a dollar. The SMS side runs on registered 10DLC routes with STOP handling and quiet hours built in. Developers can see the messages API docs.

Ready's all-in prices per segment include the carrier fee but not screening: Scrub lists and Number validation, the two switches in Settings under Compliance, each add $0.003 per segment when switched on, shown as screening on your bill, and both are off by default. Unless you turn the opt-out footer off, the first text to each contact, and one again from time to time, also gets "Reply STOP to opt out" and your business name if the message does not already have them, so a text close to the segment limit can bill one more segment.

If you buy a predictive dialer elsewhere, Ready's standalone TCPA and DNC litigator scrub is $0.005 per contact. That reduces risk. It doesn't make you lawsuit-proof, and compliance remains the caller's responsibility.

The practical takeaway

  • 10+ agents dialing aged consumer lists: shortlist two hosted outbound vendors and one contact-center suite, and trial with your real list.
  • Strong Linux/VoIP admin in-house: Vicidial is a serious, proven option. Budget for admin time, not the license.
  • Building a dialer product: Asterisk or FreeSWITCH, with months of engineering planned in.
  • Small team or warm leads: skip predictive. Use a progressive dialer, or click-to-call paired with instant texts.

If the last case describes you, try Ready. The first 20 one-to-one texts are free, and to send any text you need a card and a number ($2 a month, the first month charged when you get it).