Predictive Dialer & Call Abandonment: The 3% Rule
The FTC lets telemarketers abandon up to 3% of live-answered calls, not zero. Here's the safe harbor's four rules, the B2B exemption, and where TCPA applies.
Is it right for you?
Your checks are saved on this device only — they reset if you clear browser data or switch devices.
What actually counts as an "abandoned" call
An outbound call is "abandoned" under the FTC's Telemarketing Sales Rule (TSR) the moment a person answers it and the telemarketer fails to connect them to a live sales rep within two seconds of their completed greeting, the dead air a consumer hears right after saying "hello" [FTC, Complying with the Telemarketing Sales Rule, accessed 2026-09-09]. That two-second window is not a suggestion; the rule treats a slower connection, or a prerecorded pitch that starts talking instead of a human, as the same violation. The reason this shows up at all is almost always the same piece of technology: a predictive dialer, which places more calls than there are available agents on the assumption that some percentage will not be answered, so an agent is ready the instant a real person picks up. When the math is off, or a burst of answers arrives at once, the excess connects to a live person and finds no agent free, that call gets abandoned.
The rule does not ban abandoned calls outright, and it does not require zero. It sets a ceiling, three percent of all calls answered by a live person, and gives telemarketers a defined safe harbor for staying under it. Most of the compliance confusion in this space comes from teams not knowing which of two very different questions they're answering: Is this dialer legal to use on cell phones (a Telephone Consumer Protection Act / FCC question about consent), or is this campaign staying under the abandonment cap (an FTC Telemarketing Sales Rule question about call center operations)? They are separate laws, enforced by separate agencies, and a dialer can clear one test while failing the other.
Why only one dialer mode can actually abandon a call
Vendors use "power," "progressive," and "predictive" loosely, but the mechanical difference between the three is what determines whether abandonment math applies at all. A power-mode dialer places one call per available agent, in sequence, the instant that agent finishes the previous call; it cannot get ahead of the team, so it produces close to zero abandoned calls simply by how it operates. A progressive-mode dialer works the same way but adds automated answering-machine detection, dialing the moment an agent frees up instead of waiting on a fixed sequence, still one call per free agent. A predictive-mode dialer is the outlier: it dials multiple numbers per agent, ahead of when they will actually be free, using pacing algorithms built on the campaign's own answer-rate history to guess how many extra calls to place. Vendor-reported throughput ranges illustrate the gap: roughly 80-120 calls per agent per hour in power mode versus 150-300 in predictive mode [MightyCall; JustCall; Dialpad, checked 2026-09-09, cross-referenced across vendor comparison pages, directional industry figures, not an official standard]. That extra throughput is exactly what creates abandonment risk, and exactly why the FTC's rule targets predictive dialing by name in its own guidance.
The tradeoff has a rough breakeven point. Below roughly 8-15 concurrent outbound agents, the statistics that make predictive pacing profitable don't have enough volume to work with, so a team gets the compliance exposure of predictive dialing without the throughput gain that's supposed to justify it. Teams evaluating outbound-heavy platforms for that higher-volume case: Five9 built its dialer engine and pacing algorithms specifically for large blended outbound operations, while smaller sales teams making a few dozen dials a day are usually better served sticking with a power- or progressive-mode setup, the kind offered as an add-on tier on tools like Aircall or bundled into JustCall's Sales Dialer. Matching the dialer mode to actual team size, not to whichever mode sounds most advanced, is the first compliance decision, before any safe harbor paperwork. None of this pacing math touches a separate lever outbound teams pay for on top of it, putting a verified name on the call itself so more of the calls that do connect actually get picked up.
The 3% safe harbor, in full
The TSR's abandoned-call prohibition has a built-in exception, and it only protects a telemarketer who satisfies all four of its conditions, not just the headline 3% number. First, the technology has to ensure no more than three percent of all calls answered by a live person go unconnected to an agent, measured separately for each calling campaign, over each successive 30-day period the campaign runs (or the full campaign if it's shorter than 30 days). Second, an unanswered call has to ring for at least 15 seconds or four rings before the system hangs up, so a consumer gets a fair chance to pick up rather than facing dead air after two rings. Third, on the small percentage of calls where no agent is available within that 2-second window, the system has to play a recorded message stating the seller's name and a callback number, never a sales pitch. Fourth, and the piece most small teams overlook, the telemarketer has to keep records proving all three of the above actually held [FTC, Complying with the Telemarketing Sales Rule, accessed 2026-09-09].
That third element, per-campaign, per-30-day measurement, is where averaging goes wrong. A telemarketer running two simultaneous campaigns cannot offset a 6% abandonment rate on one against a 0% rate on the other and call the blended number compliant; the FTC evaluates each campaign against the 3% ceiling on its own, separately, campaign by campaign [FTC, Complying with the Telemarketing Sales Rule, accessed 2026-09-09]. A vendor dashboard that only reports a company-wide or account-wide abandonment average is not answering the question the safe harbor actually asks.
What getting it wrong has actually cost
The FTC can pursue civil penalties of up to $53,088 per violation, current as of the January 2025 inflation adjustment, and because each abandoned call is typically treated as a separate violation, the number scales with campaign volume fast [FTC, Complying with the Telemarketing Sales Rule, accessed 2026-09-09]. A real case shows the mechanics: in a 2005 settlement, the FTC found that telemarketer Braglia Marketing Group had abandoned calls to consumers (in addition to calling numbers on the Do Not Call Registry and not paying required access fees) while telemarketing timeshare properties on behalf of Flagship Resort and Atlantic Palace in Atlantic City. The settlement required BMG and its owners to pay more than $526,000 in civil penalties, reduced to $3,500 based on a demonstrated inability to pay, while the timeshare sellers who hired BMG paid a separate $500,000 civil penalty of their own for the same underlying conduct [FTC press release, "FTC Announces First Do Not Call Rule Settlements," Feb. 2005, accessed 2026-09-09]. Then-FTC Chairman Deborah Platt Majoras summed up the theory of liability that made the sellers, not just the telemarketer they hired, pay: "You cannot hire subcontractors to break the law for you and then walk away free of consequences." [FTC, Feb. 2005, accessed 2026-09-09]
“You cannot hire subcontractors to break the law for you and then walk away free of consequences.”
FTC Chairman Deborah Platt Majoras, on a 2005 settlement over abandoned calls and Do Not Call violations
That case is two decades old, but the liability structure it illustrates hasn't changed: a business that outsources its outbound dialing to a call center or BPO doesn't get to point at the vendor when the abandonment rate runs hot. The seller who benefits from the calls carries exposure alongside whoever placed them.
Who this doesn't apply to, and where it still might
A narrower exemption in the TSR (16 CFR 310.6(b)(7)) covers telephone calls between a telemarketer and any business made to induce that business to buy goods or services, and it exempts those calls from most of the rule, including Section 310.4(b), the section that houses the abandonment prohibition and the 3% safe harbor [eCFR, 16 CFR 310.6, current through the December 2024 amendment, accessed 2026-09-09]. In practice, that means a B2B outbound sales team predictive-dialing a list of business phone numbers to pitch software or services is not bound by the 3% abandonment cap the way a consumer-facing telemarketing campaign is. The exemption is not unlimited, though: it explicitly does not cover the TSR's deceptive-practices provisions (Section 310.3(a)(2) and (4)), and it carves out calls selling nondurable office or cleaning supplies, a narrow retail category that stays fully covered [eCFR, 16 CFR 310.6, accessed 2026-09-09]. A team relying on the B2B exemption should confirm the calls are genuinely business-to-business, not a consumer being called at a home office number, before treating the abandonment rule as inapplicable.
This is also the point where a lot of vendor content quietly conflates two different rules. The B2B exemption above is specific to the FTC's TSR and the abandonment/Do-Not-Call machinery. It says nothing about the separate consent requirements the Telephone Consumer Protection Act imposes on calling a cell phone, which is the subject of the next section, and which does not carve out businesses the same way.
The other law in the room: TCPA, ATDS, and state mini-TCPA statutes
The Telephone Consumer Protection Act (TCPA), enforced by the FCC, not the FTC, is a separate question from abandonment rate: it governs whether a device counts as an "automatic telephone dialing system" (ATDS) that requires the called party's prior express consent, particularly on a cell phone. For over a decade, several federal appeals courts had held that any dialer pulling from a stored contact list, including most predictive dialers, qualified as an ATDS. That changed in April 2021, when the Supreme Court ruled unanimously in *Facebook, Inc. v. Duguid* that a device only meets the ATDS definition if it has the capacity to use a random or sequential number generator to store or produce the numbers it calls [multiple law firm summaries of the ruling, cross-referenced, accessed 2026-09-09]. Most modern predictive dialers work from a curated calling list, a CRM export or a purchased lead file, not randomly generated digits, which took a large share of them outside the federal ATDS definition and, with it, outside the TCPA's strictest prior-consent requirement. That ruling has no effect on the FTC's separate 3% abandonment rule; a dialer can be entirely outside ATDS scope under Duguid and still owe a full accounting under the TSR's safe harbor.
The Duguid ruling only narrowed the federal definition, and several states did not follow along. Florida's Telephone Solicitation Act (FTSA), enacted in 2021 specifically in response to Duguid, originally defined an "automated system" far more broadly than the federal ATDS standard, covering any automated system for selecting or dialing numbers regardless of random or sequential generation, and it triggered a wave of class-action litigation as a result [Davis Wright Tremaine; McGuireWoods; Goodwin, cross-referenced, checked 2026-09-09]. Florida narrowed the FTSA's definition in a May 2023 amendment to track more closely with the federal ATDS standard, and a 2024 Florida circuit court ruling (Leigue v. Everglades College, an 11th Judicial Circuit trial-level decision, not yet affirmed on appeal) further held that the law's statutory damages, the greater of $500 in statutory damages per violation or actual damages, trebled to $1,500 for willful violations, are capped per lawsuit rather than stacking per individual call, though other courts have not uniformly adopted that reading [Bradley Arant; Quarles, cross-referenced, checked 2026-09-09]. The practical takeaway for a business calling into Florida, or any state with its own mini-TCPA statute, is that clearing the federal Duguid bar and the FTC's 3% safe harbor does not automatically clear a state law that defines things differently; check the specific state before assuming the federal rules are the whole picture.
FAQ
Does the 3% limit apply to every outbound call I place, or just the ones a person answers?
Just the ones a person answers. A ring-out, a voicemail pickup, or a dead number never enters the calculation at all [ftc.gov, Sept. 2026].
If our predictive dialer vendor says their platform is 'TCPA compliant,' does that mean we're covered here too?
Not automatically. TCPA compliance is about whether the dialer needs the called party's prior consent (the ATDS/Duguid question), a completely different legal question from the FTC's abandonment safe harbor, which is a campaign-level operations requirement with its own 3%, per-30-day, per-campaign math and its own recordkeeping duty. A platform can satisfy one and still leave you exposed on the other if nobody is pulling the actual campaign reports.
Are the two lower-throughput dialer modes exposed to this rule too?
Structurally, no, since each only dials the next number once an agent has actually opened up, so dead-air connects are rare by construction, but that only holds when a team genuinely operates in one of those two modes rather than a predictive mode mislabeled as such in a vendor dashboard. Confirming which mode is genuinely active is worth an explicit check, not an assumption based on the plan name.
Does calling companies, not individual consumers, get us out of this entirely?
From the abandonment math specifically, generally yes, since the FTC's B2B exemption covers it, though it leaves deceptive-marketing claims fully in play, and it does nothing to the TCPA's own consent rules or a state law written to catch autodialers on its own terms, if either reaches the numbers being called.
What actually triggers an FTC abandonment-rate investigation?
A consumer complaint to the agency or a state attorney general is the usual trigger, and it often arrives bundled with a registered-number complaint, since a campaign careless enough to abandon calls at scale is frequently also dialing numbers that opted out. The 2005 BMG case folded both types of complaint into one settlement.
For that exact comparison, see our PCI Compliance for Call Centers guide.