The case

Smith v. ExamWorks, 2026 WL 2823260, decided in the District of Maryland on 21 September 2026, case number 8:21-cv-02746-PX. Filed back in October 2021, it was a two-count class action against GEICO and ExamWorks over non-emergency prerecorded calls placed by a GEICO vendor to schedule independent medical examinations.

The court refused to certify the class. The reason given most attention in the write-ups is adequacy: to represent a class you have to show you are in the same position as its members, and this plaintiff was not, because at some point during the litigation he destroyed the single most important piece of evidence. With the handset gone, the defence could not examine it to establish who had been messaging him, what he had been doing, or even that the calls had landed on that phone at all.

Where the consent language comes in

Along the way the court made some observations that would help callers if they stand. One is that consent can be conveyed by conduct, so handing a business your phone number can itself amount to prior express consent. The other is the reading being quoted everywhere: that the statutory phrase about a call made with prior express consent attaches to the call, not to the use of a prerecorded voice. On that reading a consumer does not have to agree to the technology, only to being contacted.

That is genuinely interesting, and it may matter later. It is also not what the case was decided on. The certification denial did not rest on it, which is the textbook definition of a remark a later court is free to ignore.

The lawyer who broke the story is blunt about it himself. His write-up describes the court’s other findings as likely driven by its irritation at the destruction of evidence. When the person arguing your side says the reasoning was coloured by the plaintiff’s behaviour, that is a signal about how much weight it will carry.

Why we are not changing anything, and nor should you

Start with the fact that the same district pointed the other way six months ago. A March 2026 decision in the District of Maryland suggested consent has to reference the regulated technology. Two orders, one courthouse, opposite directions, and no appellate ruling tidying it up. A split like that is a reason to sit still, not to move.

Then there is geography. A call center software operation places calls into every state, and the plaintiff picks the forum. A helpful reading in Maryland does nothing for you in a district that has never considered the question.

State law is the third problem, and the one people forget. Florida, Oklahoma and Washington run their own telemarketing statutes with their own consent standards and their own private rights of action. A federal court’s reading of the TCPA does not move any of them. Plenty of operators who are clean federally have been sued under a state statute instead.

And this is also single-source for now. One specialist blog has written it up, in two posts on the same day. That is not a criticism of the reporting, it is a statement about how much confirmation exists. Before rebuilding a consent flow on the strength of it, wait for other lawyers to read the order.

What is actually worth doing this week

The useful lesson in this case is not about consent at all. It is that the defence won because it could point at missing evidence, and the plaintiff lost because the record was gone.

That cuts both ways, and your side of it is the consent file. For every number you dial you want to be able to produce where it came from, what the person was looking at when they gave it, the exact wording they agreed to, and the timestamp. Not a flag in a column saying consent equals yes, but the underlying record.

Three things make that real in a dialling platform:

• Store the form as it appeared, not just the answer. Consent language changes. If you cannot show what the page said in March, you cannot prove what somebody agreed to in March.
• Keep the provenance with the record. Numbers arrive from web forms, imported lists, partner feeds and inbound calls, and those carry very different weight. Merge them into one field and you have thrown away the distinction a court will ask about.
• Retain long enough to survive the limitation period. A four year old claim needs a four year old record. Retention policies written for storage cost tend not to account for that.

None of that depends on how this consent question resolves. It is what turns a lawsuit into an early exit regardless of which way the reading goes.

Frequently asked questions

Does this mean giving out my number is now consent to robocalls?

No. One district court suggested it might be, in a case decided on other grounds, and another judge in the same district suggested otherwise earlier this year. Nothing has changed about what you are required to do.

Is the ruling binding anywhere?

District court decisions bind nobody, not even other judges in the same district. They carry persuasive weight, and a point the decision did not rest on carries less of it again.

Should we stop asking for technology-specific consent?

No. Asking somebody to agree to prerecorded or automated calls costs you a line of text and removes an argument. Dropping it to chase a favourable reading that might not survive is a poor trade.

What about informational calls like appointment scheduling?

That was the setting here, and informational calls have always had a different consent posture from marketing. The distinction is worth keeping clear in your campaign configuration, because it decides what a revocation covers.

How long should consent records be kept?

Long enough to cover the limitation period for every claim you could face, federal and state, which is longer than most retention defaults. Work backwards from the longest one that applies to you.

<p>The post The Robocall Consent Ruling Everyone Is Quoting Was Decided on a Phone in the Mississippi first appeared on ICTBroadcast.</p>

That is a change in kind, not degree. For five years the database has been a place where carriers file a description of how they fight illegal robocalls. The proposal turns it into a vetting file that decides whether you get to hand traffic to anyone.

What the proposal actually says

The Further Notice was adopted 22 July and released 23 July 2026, under WC Docket Nos. 24-213 and 17-97 and CG Docket No. 17-59. None of it is law yet. Read what follows as the direction of travel rather than a checklist you owe anybody in October.

The scope language is the part worth reading twice. The FCC proposes clarifying that the rules apply across the voice ecosystem regardless of how an entity describes its service, and it lists VoIP resellers, MVNOs, PBX providers, dialing platforms and call centers. If you have spent years confident that the database was your carrier’s problem, that sentence is aimed at you.

I want to be careful not to overstate this, because plenty of compliance blogs will. The proposal asks questions. It does not yet impose a filing duty on every call center in the country, and a lot of comment filings between now and November will argue about exactly where that line lands. But the Commission has been widening this net steadily since the gateway provider rules, and I would not bet on it narrowing.

Four certifications, and one with a clock on it

Filers would certify four things: that they have not submitted false, misleading or inaccurate information to the FCC, that they comply with the robocall, caller ID authentication and numbering rules, that they meet STIR/SHAKEN attestation requirements, and that they respond to traceback requests within 24 hours.

That last one is the operational item. Twenty four hours is not long if traceback requests land in a shared inbox that somebody checks when they get a minute. The Industry Traceback Group typically wants to know which customer originated a specific call at a specific second, and if your platform cannot answer that from a CDR in a few minutes, the clock is already against you.

Run a drill. Pick a call from last Tuesday out of your own logs and see how long it takes someone who is not you to produce the originating tenant, the source IP, the caller ID presented, and the consent record. Most teams find the answer is somewhere between twenty minutes and never.

The disclosures are about people, not technology

The proposed filing would carry principals’ contact details and citizenship, every affiliate and subsidiary, a US registered agent, prior regulatory actions, and identification of third party vendors. Any STIR/SHAKEN exemption you claim would have to name the specific rule that allows it and explain why it applies using facts about your own network.

Read that list again and notice what the FCC is building. It is a map of who is behind an entity, so that a provider removed under one name cannot reappear the following month under another. The database stops being a description of your mitigation and becomes a description of you.

For an honest operator this is mostly paperwork. For anyone who has ever quietly stood up a second company to carry traffic the first one could not, it is the end of that trick.

How a stale filing becomes blocked traffic

The enforcement side is where the proposal has teeth. Staff would be able to hold a filing in pending status, withhold publication, or reject it outright, particularly where it looks connected to a previously removed provider. Grounds for suspension or removal get codified: a deficient filing, inconsistencies with what the Commission can see elsewhere, failure to update on time, carrying illegal calls, mitigation that does not work in practice, and caller ID authentication violations. Serious cases get an expedited process with five days to cure or respond before removal.

Removal from the database is not a fine. It means downstream providers are required to stop accepting your traffic. You do not get to appeal your way back into service while your campaigns keep dialing. That asymmetry is the whole point, and it is why the existing obligations are worth more attention than they usually get: recertify by 1 March every year, and update the entry within ten business days of a change to your address, ownership, registered agent or officers.

The ten day rule catches more people than the annual one. Nobody forgets 1 March twice. Plenty of companies change a registered agent in June and update the database the following spring.

What this asks of your call center software

Strip out the filing mechanics and four capabilities decide whether you can live with rules like these. They are worth auditing whatever happens to this proposal.

Per-tenant attribution that survives a year. When a traceback arrives you need to name the customer, not the platform. On a multi-tenant deployment that means CDRs that carry the tenant identity, the campaign, and the caller ID presented, retained long enough to answer a request about a call from eight months ago.

Consent you can produce, not just claim. The certification about complying with the robocall rules is only as good as the record behind it. If consent lives in a CRM nobody exports and the dialer only holds a phone number, you have an assertion rather than evidence.

Controls that actually stop a campaign. Your mitigation description has to match what the platform enforces. Do not write that you throttle suspicious traffic if the only throttle is a person watching a dashboard. Pacing, abandonment and calling window controls that are configured at the tenant level are the difference between a description and a control.

Somebody who owns the traceback queue. This is a staffing answer, not a software one, but the software decides how painful it is. A named person, a monitored address, and a documented lookup that takes minutes.

Self-hosted platforms have a real advantage on the first two, and I say that as someone with an obvious interest. When the CDRs and the consent records sit in your own database you can answer a traceback without opening a vendor ticket. When they sit in a hosted platform you are waiting on somebody else’s support queue while a 24 hour clock runs. We made a similar argument about upstream provider vetting last year, and this proposal is the same pressure arriving one layer closer to you.

What I would do before March

Not much of this requires a decision from the FCC first.

Check whether your company or any affiliate has an entry in the database, and read what it says. A surprising number of entries were written by someone who left, describe a stack that has since been replaced, and have not been touched since the original filing. If the description does not match what your platform does today, that is an inconsistency under the proposed grounds for removal, and it is also just bad practice.

Then write down your affiliates. Every entity that touches the traffic, every reseller relationship, every white label tenant operating under their own brand. The proposal wants that list and most operators cannot produce it in an afternoon.

Last, if the scope language worries you, file a comment. The window is open until 9 October and the record is genuinely thin on what these rules cost a twenty seat call center as opposed to a carrier. Regulators write the rule they have evidence for. Nobody is going to make that argument on your behalf.

FAQ

Does my call center have to file in the Robocall Mitigation Database today?

Today the filing obligation sits with voice service providers and intermediate and gateway providers. If you buy origination from a carrier and do not hold numbering resources yourself, you are most likely not a filer. The proposal asks whether that line should move, and names call centers and dialing platforms while asking.

What happens if a provider is removed from the database?

Downstream providers are required to stop accepting its traffic. In practice that is an outage rather than a penalty, which is why filings and updates deserve more attention than their tedium suggests.

When are the comment deadlines?

Comments are due 9 October 2026 and reply comments 9 November 2026, in WC Docket Nos. 24-213 and 17-97 and CG Docket No. 17-59. The Further Notice was adopted 22 July and published in the Federal Register on 9 September 2026.

Is the 24 hour traceback response a new rule?

Responding to traceback has been expected for years. What the proposal adds is a certification, signed by you, that you do it within 24 hours. Certifying something you cannot demonstrate is a worse position than not certifying it.

Does STIR/SHAKEN exemption paperwork change?

Under the proposal, claiming an exemption would mean naming the rule that authorises it and explaining why it applies with facts about your own network. A generic exemption claim would not survive screening.

How does ICTBroadcast help with any of this?

It is not a compliance product and I would not sell it as one. What it gives you is the underlying evidence: per-tenant CDRs on your own server, recordings, calling window and pacing controls set per tenant, and a REST API to pull all of it out when someone asks. Our TCPA and STIR/SHAKEN checklist covers the rest of the routine. The filing is still yours to write.

If you are sizing up whether your current stack could answer a traceback inside a day, that is a good test to run this week rather than next March. Take ICTBroadcast for a look and try the lookup on your own data.

<p>The post The FCC’s Robocall Database Overhaul Names Dialing Platforms and Call Centers first appeared on ICTBroadcast.</p>

Attestation and reputation are two different things

A lot of teams treat STIR/SHAKEN as a spam-label fix. It isn’t one. The framework lets a carrier vouch that the caller ID hasn’t been spoofed. A-attestation means your provider knows you and confirmed you own the number. That’s real value, and you want it. But the phone that rings is a mobile handset running its own screening, fed by the carrier’s analytics engine, and that engine scores every number on behavior.

What the carrier model actually looks at

Every carrier keeps its own recipe, but the signals overlap. High call volume from one number in a short window reads like a spam run. A large share of very short calls, the kind that end in a second or two, looks like people declining a robocall. Complaint reports weigh heavily, and so do repeated hits against numbers that have gone quiet or landed on do-not-call lists. Roll those together and you get a rolling reputation score. Cross a threshold and the label flips.

Six habits that keep numbers clean

Reputation is earned by how you run campaigns, day after day. These are the levers that matter, and good call center software should make each one easy rather than manual.

Start with pacing. Keep abandonment low and close the silent gap between an answer and an agent, because that gap is the single clearest robocall tell. Scrub your lists against do-not-call and known-litigator data before the campaign runs, not after a complaint arrives. Add branded caller ID through Rich Call Data so your name and logo show on the screen, which lifts answer rates and shortens the guessing that leads to declines. Spread volume across numbers sensibly instead of burning one line. Track spam-label and complaint rates per number so you can pull a number before it drags the rest down. And respect calling windows and per-day contact caps by region.

None of this replaces STIR/SHAKEN. It sits alongside it. The standard gets you in the door with the carrier, and clean dialing keeps you welcome. For the mechanics of pacing and abandonment specifically, our guide on how predictive dialers work in 2026 goes deeper.

Recovering a flagged number

If a number is already labeled, rest it. Move active campaigns to healthier lines, stop the behavior that triggered the score, and let the rolling average recover, which usually takes weeks rather than days. Some carriers and analytics providers accept remediation or registration requests, and branded caller ID registration can help re-establish trust. What does not help is rotating to a fresh number and repeating the same pattern, because you’ll just train the model again.

FAQ

Does A-attestation guarantee my calls connect? No. It confirms the number is yours and unspoofed. The carrier’s reputation model still decides the label based on your calling behavior.

Why do short calls hurt me? A high ratio of one or two second calls looks like people rejecting a robocall, which is one of the strongest spam signals a carrier tracks.

Is branded caller ID the same as STIR/SHAKEN? No. STIR/SHAKEN authenticates the number. Branded caller ID, delivered through Rich Call Data, displays your name and logo. They work best together.

How long until a flagged number recovers? Reputation is a rolling average, so plan on weeks of clean, low-volume dialing rather than an overnight reset.

Should I just keep buying new numbers? No. If the dialing pattern doesn’t change, new numbers get flagged too. Fix the pattern first.

The short version: STIR/SHAKEN is table stakes, not a shield. Your numbers stay clean when your dialing looks like a business talking to people who want to hear from it.

<p>The post Your Calls Pass STIR/SHAKEN and Still Say Spam Likely. Here Is Why first appeared on ICTBroadcast.</p>

What Article 50 actually asks for

The obligation is short, which is exactly why it slips past people. If a person interacts directly with an AI system, they have to be told. On a phone call, that means audio, at the start, in words a caller understands without hiring a lawyer.

Two facts have to land. First, that the caller is a machine. Second, on whose behalf it is calling. A watermark buried in call metadata does not clear the bar. Neither does a clause in your terms of service, which nobody hears while a phone is ringing. My honest read is that the second half will catch more operators than the first, because resellers and white-label teams habitually name the platform instead of the client whose campaign is running.

Penalties reach 15 million euros or 3% of worldwide annual turnover, whichever is higher. There is a separate December 2026 deadline for certain content-marking duties, but the phone disclosure date did not move with it.

The uncomfortable number sitting behind the rule

A 2026 blind study found that 71% of callers could not reliably tell an AI voice agent from a human. That single figure is most of the justification for Article 50. Back when synthetic speech was obviously synthetic, disclosure was redundant. It stopped being obvious somewhere around the point where end-to-end response latency dropped into the 250 to 800 millisecond band and the pauses started sounding like someone thinking.

Volume moved too. Voice AI went from 6% of inbound contact center traffic in 2024 to 19% in 2026, with banking and telecom leading. Regulators do not write disclosure rules for fringe use cases.

Where this breaks in a real dialer setup

Picture a twelve-seat outbound team running EU campaigns. A bot qualifies, a human closes. The bot discloses and the human does not need to, which sounds simple until the same audio prompt gets reused on a campaign where a person is doing the dialing. Now you have a recording announcing an automated system on calls that are not automated. Wrong in the other direction, but still wrong.

Language is the second trap. Plain language means the caller’s language, not yours. One English prompt on a campaign dialing Belgium and Poland does not satisfy anything, and this is the failure I would expect to see most often in audits over the next year.

Placement is the third. “At the first interaction” is not a synonym for “somewhere in the first minute.” If your script opens with a two-sentence hook and discloses afterwards, the interaction already happened. Answering machine detection complicates this further, because a disclosure that fires before detection resolves ends up on voicemail, and a disclosure that fires after it may land several seconds into a live conversation. Pick the order deliberately rather than inheriting whatever the flow does by default.

The FCC is walking the same road, just slower

US operators are not exempt from this thinking. The FCC ruled back in February 2024 that calls using AI-generated voices count as “artificial” under the TCPA, which pulls them under prior express written consent. A final rule dealing specifically with AI-generated calls is expected around Q4 2026 or Q1 2027, though the Commission under Chairman Carr has signaled a lighter regulatory posture than the previous one.

The likely shape is familiar: an explicit AI identification requirement at call open, consent language that names AI directly, and a formal definition of an AI-generated call. That overlaps enough with Article 50 that building two separate implementations would be wasted effort. Build one, then relax it per region if the FCC lands somewhere softer.

The money is not theoretical. TCPA class actions in 2025 and 2026 have settled in the $5M to $20M range, and statutory damages sit at $500 to $1,500 per unauthorized call. If your team is still working through the basics here, our guide to autodialer laws and TCPA compliance covers the consent side that these AI rules attach to. Teams dialing across borders should also read the country-wise auto dialer regulations before assuming one script works everywhere.

Proving it happened, six months later

This is the part almost everyone underinvests in. A complaint lands long after the campaign has been archived, and “we configured it correctly” is not evidence of anything. What you need is a call record that can be read back by someone hostile.

If you only add one field, make it the offset. Whether the disclosure played at all is usually easy to establish from audio. When it played, relative to the first word of the conversation, is the thing that actually gets argued about. Keep these fields for as long as you keep call detail records generally, because the complaint window outlasts the campaign by a wide margin.

One more practical note: version your disclosure text. Campaign wording changes mid-flight more often than teams admit, and a log entry pointing at “the standard disclosure” is useless if the standard disclosure was edited twice in March.

Doing this in ICTBroadcast

ICTBroadcast integrates with Google Dialogflow for building conversational agents, so the disclosure does not need a separate compliance module. It belongs in the call flow, ahead of the handoff. A few things worth setting up deliberately:

• Place the disclosure as the first prompt in the IVR flow, before control passes to the Dialogflow agent, so it always precedes the first conversational turn.
• Keep one prompt per language and bind it to the campaign’s target country rather than to a global default.
• Run AI campaigns as their own campaigns, separate from plain voice broadcast lists, so the opening prompt cannot leak onto human-dialed calls.
• Name the client business in the prompt text, not the platform. If you resell, this needs to be a per-tenant value, not a shared one.
• Export call detail records with the campaign identifier attached so the disclosure version can be reconstructed later.

None of this is heavy engineering. It is an hour of prompt work and a decision about where the boundary between bot and agent sits.

A short checklist before your next EU campaign

• Does the disclosure play before the first substantive sentence, not after it?
• Does it name a business, and is that business the client rather than the dialer vendor?
• Is there a prompt in every language the campaign targets?
• Does answering machine detection run before or after the disclosure, and did you choose that on purpose?
• Can you pull a call from four months ago and show when the disclosure played?
• Is the AI campaign isolated from lists that humans dial?

Frequently asked questions

Does Article 50 apply if a human agent uses AI assistance during the call?

The trigger is direct interaction between a person and an AI system. A human agent reading AI-suggested text to a caller is not the caller interacting with the AI. A synthetic voice speaking to the caller is. The grey zone is real-time voice cloning of an agent, and I would disclose in that case rather than argue about it later.

Do we need consent as well as disclosure?

They are separate duties and both apply. Article 50 governs transparency. Consent for the call itself comes from other law, and in the US the TCPA requires prior express written consent for AI-voiced marketing calls regardless of what you announce at the start.

What if the caller hangs up during the disclosure?

That is a compliant outcome, not a failure. Log it as a disclosure played with an early disconnect. Teams sometimes try to shorten the notice to reduce these drops, which is the wrong optimisation to run.

Does this apply to inbound calls too?

Yes. The obligation follows the interaction, not the direction of dialing. An inbound AI receptionist handling EU callers needs the same spoken notice at the top of the call.

We only dial the US. Can we ignore this?

For now, mostly. But the FCC rule expected in late 2026 or early 2027 points in the same direction, and retrofitting disclosure into a live campaign under deadline is worse than building it now. Most teams we talk to are treating the EU wording as the template.

Is a disclosure needed on every call in a campaign, or once per contact?

Per call. There is no concept of a caller remembering last week’s notice, and building that assumption into your flow creates a record you cannot defend.

Related resources

• Autodialer laws and regulations: TCPA compliant auto dialer
• Country-wise regulations about auto dialers
• Press 1 campaigns for interactive voice broadcasting
• IVR features in ICTBroadcast
• Free Asterisk based auto dialer

Getting this configured

If you are running EU campaigns and want the disclosure wired into the flow properly rather than bolted on, take a look at what ICTBroadcast ships with, and at the edition comparison if you need the Dialogflow integration specifically. Questions about a particular campaign setup are best raised through our support portal, where they reach the engineering team directly.

<p>The post Your AI Caller Must Now Say It Is a Machine. What Call Center Software Has to Change first appeared on ICTBroadcast.</p>

That gap is usually explained as a training problem or a change management problem. The 2026 research points somewhere else, and it is a much more boring answer: the AI cannot see enough of the business to be useful.

The number that explains the other numbers

Bought is not the same as running, and the distance between them has a cause.

Puzzel’s State of Contact Centres 2026 research found that the average team runs 3.9 separate contact center technologies, and only 3% operate on a single unified platform. Three percent. That is the number that makes the adoption gap make sense.

Think about what four disconnected systems does to an AI feature. The dial outcome lives in the dialer. The customer history lives in the CRM. The recording lives in a third system, and the ticket lives in a fourth. Any task that needs two of those at the same time either fails or gets handed back to a person.

So the AI ends up confined to the things one system can do alone. Transcribe this recording. Summarise this ticket. Those are real conveniences and nobody minds them, but they are not what anyone approved the budget for, and they are not enough to change how a day runs.

Meanwhile the pressure to show something keeps building. Gartner found 91% of customer service leaders under executive pressure to implement AI. Pressure plus a fragmented data path produces exactly what the numbers show: fast purchasing, slow adoption.

Why a tool that is usually right gets ignored

There is a second failure mode that follows straight from the first, and it is the one that quietly kills pilots.

When a model can only see part of the picture, it guesses at the rest. It is right most of the time, because most cases are ordinary. Being right most of the time sounds fine until you look at what it does to the person using it.

A tool that is right most of the time gets checked every time. The agent reads the suggestion, then opens the CRM to confirm it, then acts. That is slower than just opening the CRM. So after a few weeks they stop reading the suggestion, and the feature is technically live and functionally dead. It is in the 88% and not in the 25%.

Trust here is not a soft factor. It is a straightforward calculation the agent is making about whether verification costs less than doing it manually. Fix the data path and the calculation flips.

The research keeps landing on agent workflow, not self-service

The order most teams buy in, and the order the returns actually show up in.

Verint’s 2026 research found the highest returns in agent workflow rather than customer self-service. That finding keeps recurring across different studies, and it keeps getting ignored, for a reason that is easy to understand.

Self-service is the part that looks like a headcount saving. It has a clean business case you can put on one slide. Agent workflow improvements look like helping people do their existing jobs slightly better, which is a much harder thing to write a number next to before you have done it.

But the self-service case has a measurement trap in it. Deflection is easy to count and easy to misread. A call the bot handled without a transfer counts as deflected, whether or not the customer called back an hour later, gave up, or emailed instead. Teams declare a win on the deflection number, then find total contact volume did not move, then blame the model and start shopping for a better one. That is how you end up with a fifth system to integrate.

The agent workflow route starts smaller. The AI drafts the summary, suggests the next action, and pulls the relevant history in before the agent asks. It needs less autonomy to be useful, and crucially it fails safely, because a person is reading everything it produces. But it needs the joined-up data that self-service also needs, which is why doing it first is the right order rather than a compromise.

Get the order right

Three steps, in this sequence.

First, put the call outcome and the customer record somewhere a tool can read them together. That is not a platform migration and it does not mean ripping out four systems. It means one place where a query can join what happened on the call to who the person is. If that place does not exist, nothing you buy next will work properly, no matter how good the model is.

Second, point AI at the agent’s work. Summaries, next-action suggestions, quality scoring across the whole call volume instead of the two calls a supervisor had time to listen to. These pay quickly and they generate the labelled outcome data you need for anything more ambitious.

Third, automate the calls that never needed a person, once you can prove from your own data which ones those are. By then you are automating a category you have measured, not a category a vendor deck told you exists.

Done in that order you finish with fewer moving parts than you started with. Done in the usual order you finish with more.

Three questions before the next purchase

These are worth asking any vendor, and worth asking about tools you already own.

Can it read the outcome? If the tool cannot see how the last call ended, it is guessing at the next one. Ask specifically how it gets the disposition, not whether it integrates.

Can it write back? A tool that only reads produces a report. A tool that writes back changes what happens on the next contact. This is the difference between an analytics product and an operational one, and the demo rarely makes it obvious.

Who owns the join? If the only place your call data and customer data come together is inside the vendor’s platform, you have rented your data path along with the model. Switching costs then grow every month, and the thing you would be switching away from is not the AI, it is the integration.

That last question is where self-hosting earns its keep. When the dialer runs on your own infrastructure, the call records, dispositions and recordings sit in a database you control, and you decide what reads them. Our write-up on open source cloud call center software covers what that looks like in practice.

Where ICTBroadcast fits

Worth being precise here, because vague AI claims are part of the problem this article is about.

ICTBroadcast integrates with Google Dialogflow. That means you build the conversational agent in Dialogflow, with its intents and flows, and ICTBroadcast handles the telephony side: placing the calls, managing the campaign, and recording what happened. The integration is the connection between the two, not a separate AI engine bundled into the dialer.

Why that matters for everything above: the campaign data, call outcomes and recordings stay in your ICTBroadcast installation on your own server. The conversational layer is a component you chose and can replace. Nobody else owns the join between what the agent said and what your business knows about the customer.

ICTBroadcast also ships a REST API by default, which is the practical mechanism for step one. If the outcome data needs to reach your CRM, or your CRM needs to feed the dial list, that is an API call rather than a support ticket. We covered the API in more detail in the SP Edition REST API post.

What to do this week

Count your systems. Just count them. Dialer, CRM, ticketing, recording, quality, reporting. If the number is around four, you are average, and the average is where the adoption gap lives.

Then pick one question your team asks often that needs two of those systems to answer, and time how long it takes to answer today. That number is your integration debt, expressed in a unit anyone can understand. It is also the thing your next AI purchase will run into.

Frequently asked questions

Why do most contact centers have AI but not use it daily?

Because the data the AI needs is split across separate systems. Research puts the average team on 3.9 different contact center technologies with only 3% on a unified platform, so anything requiring two systems at once either fails or gets escalated to a person. The tool then gets used only for single-system tasks like transcription.

Is agent-facing AI really a better first step than self-service?

Verint’s 2026 research found the highest returns in agent workflow rather than customer self-service, and that finding shows up repeatedly. Agent-facing tools need less autonomy to be useful, fail safely because a person reviews the output, and produce the outcome data that later automation depends on.

What is wrong with measuring call deflection?

Deflection counts calls the bot finished without a transfer. It does not count the customer who called back later, gave up, or switched to email. Check total contact volume over the same period instead, because that is the number that reflects whether work actually went away.

Does ICTBroadcast include an AI voice agent?

ICTBroadcast integrates with Google Dialogflow, so you build the conversational agent there and ICTBroadcast handles call placement, campaign management and recording. The AI capability comes from the Dialogflow side, and the integration connects it to your telephony.

What does it mean to own the join between call data and customer data?

It means the place where a call outcome gets matched to a customer record is infrastructure you control, rather than a vendor’s platform. If only the vendor can perform that join, replacing them means rebuilding the integration, which is usually a larger job than replacing the AI model itself.

Do we need one unified platform to get value from AI?

No, but you do need one place where the call outcome and the customer record can be read together. That can be achieved with an API and a shared datastore. The point is a readable join, not a single vendor.

Related resources

• Gartner’s $80 Billion Call Center AI Forecast: The Fine Print
• Open Source Cloud Call Center Software
• Self-Hosted Auto Dialer Software
• ICTBroadcast SP Edition: REST API Included by Default

Where to go next

If counting your systems produced an uncomfortable number, the first move is a readable join between call outcomes and customer records, not another platform. ICTBroadcast runs on your own infrastructure with a REST API included, so the campaign data stays somewhere you can query and connect. Tell us which systems you are trying to join and we will help you work out the shortest path. Open a ticket at service.ictinnovations.com.

<p>The post 88% of Call Centers Have AI. Only 25% Use It Daily. Here Is What Is In the Way first appeared on ICTBroadcast.</p>

Reply comments on the numbering proceeding closed on 7 July 2026, so this is live policy rather than speculation. The direction is clear enough to plan around even before the final text lands.

What the FCC is proposing

The core idea is that a voice service provider should know the upstream providers it accepts traffic from, in a specific and documented way. The proposal spells out what that looks like: confirm the upstream provider has a filing in the Robocall Mitigation Database, actually read that filing and its mitigation plan for completeness, confirm the provider holds a Service Provider Code token where it claims full or partial STIR/SHAKEN implementation, and check whether the provider appears on certain national security or enforcement related lists.

Read that list again and notice what it is not. It is not a technical standard. It is a due diligence obligation, closer to what a bank does before opening an account than anything in a SIP specification.

Why this reaches your call center software even though it targets carriers

Most outbound teams read “provider obligation” and assume it is somebody else’s paperwork. That assumption held for a few years. It is getting weaker.

Carriers that are told to vet upstream sources will pass that scrutiny down, because the cheapest way to satisfy a regulator is to tighten who you accept. Wholesale voice is a chain, and every link that gets squeezed squeezes the next one. In practice that means the questions on your next carrier onboarding form get longer, and the tolerance for vague answers gets shorter.

The other reason it reaches you is that the Commission has been willing to cut providers off entirely rather than fine them. Losing a route with no notice is an operational problem, not a legal one, and no compliance budget fixes a dialer that suddenly cannot complete calls. If your campaigns run through a single upstream, that is concentration risk you can measure today.

Teams running a self-hosted stack have a real advantage here, because the call detail records and campaign logs sit on hardware you control. Producing six months of evidence is a query rather than a support ticket. That is not a small thing when someone asks you to prove what you sent and when.

The four checks, and the part people skip

If you resell, aggregate, or hand traffic to anyone downstream of you, the four checks below are the ones the proposal describes. Even if you only buy termination and never sell it, running them on your own carrier is a reasonable way to judge whether that carrier is about to have a bad year.

The part people skip is the evidence. Checking the Robocall Mitigation Database takes two minutes and leaves no trace unless you deliberately keep one. A dated screenshot and a three line review note, filed somewhere you can find it in eighteen months, is the difference between having done the work and being able to show it. I would treat that filing habit as the actual deliverable here.

One honest caveat: none of this is a substitute for the consent and suppression work you already owe. Vetting your carrier does nothing for a campaign calling numbers it should not. Our guide to autodialer laws and regulations covers that side, and it has not become less important.

What to do in the next month

Start by writing down which upstream providers your traffic actually touches. A surprising number of operators cannot answer that quickly, especially where a reseller sits in the middle. You cannot vet a chain you have not drawn.

Then check your own filing. If you appear in the Robocall Mitigation Database, reread what you filed. Plans written two or three years ago often describe a system that no longer matches how the business runs, and an out of date plan reads worse than a modest one that is accurate.

After that, look at attestation levels on your own outbound. If your calls are going out with anything less than full attestation, find out why. Sometimes the answer is a number the carrier does not believe you own, which is fixable in an afternoon and quietly costs you answer rate every day it stays broken.

Finally, add a second termination route if you only have one. Not because your current carrier is doing anything wrong, but because enforcement in this area is fast and does not care about your campaign calendar.

Where this is heading

My read is that STIR/SHAKEN has stopped being a caller ID feature and become the paperwork trail for an accountability regime. The signature was never the point on its own. The point is that a signature ties a call to a provider, and a provider to a filing, and a filing to a name somebody can act against.

For legitimate outbound operations that is mostly good news, because the traffic it squeezes is the traffic that has been degrading answer rates for everyone. The cost is a bit more diligence and a lot more record keeping. That is a trade most serious teams should take, and the ones who set up the record keeping early will find the next rule change much less disruptive than the ones who wait for a carrier to demand it.

Frequently asked questions

Does this apply to me if I am not a carrier?

The obligations described in the proposal sit with voice service providers. If you only originate your own campaigns, you are not the direct target, but your carrier’s new diligence will show up as tougher onboarding and faster disconnection for accounts that look risky.

What is the Robocall Mitigation Database in plain terms?

It is the FCC’s public register where providers state how far they have implemented caller ID authentication and describe the steps they take to prevent illegal robocalls on their networks. Anyone can look up an entry, which is what makes it usable as a vetting tool.

What is a Service Provider Code token?

It is the credential a provider needs in order to sign calls under STIR/SHAKEN. Claiming implementation without holding one is exactly the mismatch the FCC wants downstream providers to catch, which is why confirming the token appears as a separate check.

Will full attestation stop my calls being labelled as spam?

Not by itself. Attestation tells the terminating carrier who vouched for the call, but analytics engines also weigh complaint rates, call duration patterns and how many numbers you rotate through. Good attestation with bad calling behaviour still gets labelled.

How long should I keep vetting records?

Longer than feels necessary. Investigations look backwards, and the useful answer is usually about a relationship that ended a year ago. Keeping dated evidence for the life of the relationship plus a couple of years is a sane default.

Related resources

• Autodialer laws and regulations for TCPA compliant calling
• Country wise regulations about auto dialers
• Custom caller ID with a personalised name
• Voice broadcasting with ICTBroadcast
• ICTBroadcast editions and pricing

Compliance gets easier when the records live on your own server instead of a vendor’s. ICTBroadcast is an Asterisk based auto dialer you host yourself, with campaign and call detail data you can query directly. Have a look at the available editions if you are weighing a move off a hosted platform.

<p>The post Your Upstream Carrier Just Became Your Problem: The FCC’s Next Robocall Move first appeared on ICTBroadcast.</p>

So nine out of ten calls still need a person. Where does $80 billion come from, then? That gap between the headline and the mechanism is worth understanding before you sign anything, because vendors will quote you the big number while selling you the narrow slice.

Where the Call Center Software Savings Actually Come From

Full automation gets the attention, but it’s the smallest contributor. The savings pile up in less glamorous places. Shorter handle times account for a big share: when an agent gets a live transcript, caller history, and a suggested answer on screen, a six minute call becomes a four minute call. Multiply that across a 50 seat floor and you’ve cut the equivalent of a dozen positions without automating a single conversation end to end.

After-call work is the second quiet winner. Agents have historically spent 30 to 60 seconds per call typing notes. Auto-generated summaries wipe most of that out. Then there’s QA: instead of a supervisor sampling 2% of calls, AI scoring covers every call, which means coaching happens where it’s needed rather than where the sample happened to land.

The mid-market reality check matters here. The contact center AI market hit roughly $4.89 billion this year, and enterprises like Cigna and Comcast are running voice AI at serious scale. But those firms have data teams tuning the systems weekly. If you run a 20 seat operation, your savings come from the assistive layer first, and I’d argue that’s the right place to start anyway.

What This Means for Outbound Campaigns

Most of the AI coverage focuses on inbound support, but outbound teams have their own version of this math. Predictive pacing is decades old and still delivers the single biggest efficiency gain in outbound calling: agents talk instead of waiting for someone to answer. Layer AI on top and the picture improves further. Answering machine detection gets sharper, so fewer agent seconds burn on voicemail. Campaign analytics tell you which lists, scripts, and time windows convert, so tomorrow’s dialing plan is better than today’s.

You don’t need a Fortune 500 budget for any of that. An Asterisk based auto dialer paired with sensible campaign design captures most of the assistive-layer gains at a fraction of the licensed-seat pricing the big CCaaS vendors charge. The honest answer is that the software matters less than the discipline: track connect rates, listen to failed calls, and adjust weekly.

The One-in-Ten Rule Should Shape Your Buying

Here’s the practical takeaway. If a vendor pitch centers on replacing agents, be skeptical: the research says that slice stays small through 2026. If the pitch centers on making each agent 20 to 30 percent more productive, the numbers back it up.

That framing also protects you from overbuying. Voice AI minutes, LLM API fees, and per-seat AI add-ons stack quickly. Start with what pays back fastest: automated summaries, full-coverage QA, smarter routing, and voice broadcasting for the messages that never needed a live agent in the first place. Appointment reminders, payment notices, and service alerts are automation that has worked for years, no LLM required.

Then pilot one AI voice agent use case with clear exit criteria. Measure containment and the 7 day recontact rate, not just deflection. A deflected call that comes back angry tomorrow saved you nothing.

FAQ

Will AI replace call center agents in 2026?

Not at any meaningful scale. Gartner expects only about one in ten agent interactions to be fully automated in 2026. The bigger effect is assistive: AI shortens calls, automates notes, and covers QA, which changes how many agents you need per thousand calls rather than eliminating the role.

Where does call center AI save the most money?

Handle time reduction, after-call work automation, and full-coverage QA scoring deliver the most reliable savings. Full call automation helps for narrow, high-volume intents like order status or appointment confirmation, but it’s the smallest slice of the total.

Can small call centers benefit from AI without enterprise budgets?

Yes. Open source call center software with predictive dialing, answering machine detection, and campaign analytics captures most of the productivity gain. Add AI summaries and QA scoring through affordable APIs rather than per-seat platform fees.

What should I measure in an AI voice agent pilot?

Containment rate, escalation quality, and the 7 day recontact rate. If callers the AI “resolved” call back within a week, the automation moved the work instead of removing it.

Is voice broadcasting still relevant when AI voice agents exist?

Very much so. One-way notifications like reminders and alerts don’t need conversational AI, and broadcasting them costs a fraction of an interactive AI call. Use the cheap tool where the job is simple.

Related Resources

• Free Asterisk Based Auto Dialer
• Voice Broadcasting
• TCPA Compliant Auto Dialer: Laws and Regulations
• ICTBroadcast Packages
• Multi User vs Multi Tenant Software

<p>The post Gartner Puts an $80 Billion Number on Call Center AI. Here’s the Fine Print first appeared on ICTBroadcast.</p>

Most call centers still treat outbound as sales and inbound as service. The 2026 data argues for a third lane: proactive service calls that reach customers before problems do. Let’s look at what the numbers say and how to run this play without hiring a single extra agent.

The 2026 Evidence for Proactive Outbound

The headline result came from a major telecom that used automated voice agents to welcome newly activated internet customers during their first 90 days. The calls checked whether setup went smoothly, answered common questions, and routed anyone with a real problem to a person. Cancellations in the first month dropped by more than half among customers who received the call.

That wasn’t the only proactive story at CCW this year. Vendors across the floor reported the same pattern: outbound service touches beat reactive support on both cost and retention. Gartner now projects that agentic systems will resolve 80% of common service issues autonomously by 2029, and the CCaaS market is forecast to more than double to $15.8 billion by the same year.

Here’s my honest read: the AI layer gets the headlines, but the mechanism doing the work is much older. A timely, relevant phone call in week one beats a perfect apology in month three. Automation just makes that call affordable at scale.

Why Welcome Calls Work

New customers churn for boring, fixable reasons. The router sat in the box for a week. The first invoice looked wrong. Nobody explained the feature they actually bought the service for. Each of those is a two-minute conversation if you catch it early, and a cancellation if you don’t.

A welcome call also resets the relationship. The only calls most customers ever receive are collections and upsells, so a genuine “did everything work?” call stands out. One regional ISP we know saw support tickets from new customers drop noticeably after adding a day-three check-in campaign, simply because problems surfaced on the call instead of festering into complaints.

How to Run Welcome Campaigns with Call Center Software

You don’t need an AI research team to copy the telecom playbook. Standard outbound campaign tools cover it today. The flow looks like this:

In ICTBroadcast, that maps to features you already have. Load new signups as a contact group, schedule a voice broadcast campaign with your welcome message, and attach a press-1 option so anyone with an issue lands on a live agent instead of a voicemail queue. If you want structured feedback, run it as an automated phone survey and export the responses.

Timing matters more than script polish. Day three catches setup problems while they’re fresh. A second touch around day 21, just before the first bill, catches invoice confusion. In most cases two short calls outperform one long one.

Where AI Voice Agents Fit

The telecom in the CCW case used a conversational AI agent rather than a recorded message, and that’s clearly where the industry is heading. Several providers rolled out AI voice agent features this month alone. ICTBroadcast already integrates with Google Dialogflow for conversational flows, and fuller AI voice agent capabilities are in development on our roadmap.

Should you wait for AI before starting? No. The churn reduction comes from the proactive contact itself, not from the voice being synthetic. An IVR welcome message with a press-1 escape hatch captures most of the value today, and you can swap in a conversational agent later without changing the campaign structure.

FAQ: Proactive Welcome Call Campaigns

What is a proactive welcome call?

An outbound call placed to a new customer shortly after signup, usually automated, that confirms everything works, answers early questions, and offers a path to a live agent. It’s a service call, not a sales call.

Do welcome calls really reduce churn?

Results shared at CCW 2026 showed customers who received a proactive welcome call were less than half as likely to cancel in the first 30 days. Your numbers will vary, but early-tenure churn is exactly the churn a timely call can prevent.

Can I run welcome campaigns without extra agents?

Yes. Automated voice broadcasting handles the call volume, and agents only get involved when a customer presses 1 to talk. A campaign covering thousands of new signups might generate only a handful of live conversations a day.

What should a welcome call say?

Keep it under 40 seconds: thank them, confirm the service is active, mention one useful feature, and offer press-1 for help. Skip the upsell entirely on the first touch. The goodwill is the product.

Does ICTBroadcast support this out of the box?

Yes. Contact groups, scheduled voice broadcast campaigns, press-1 agent transfer, answering machine detection, and phone surveys are all standard features across our editions.

Start Calling Customers Before They Call You

Retention is an outbound problem wearing an inbound disguise. If your call center software only reacts, you’re paying for churn you could have prevented with a 40-second call. ICTBroadcast gives you the campaign engine to run welcome calls, check-ins, and surveys on autopilot; see the packages page to pick an edition that fits your volume.

Related Resources

• Voice Broadcasting with ICTBroadcast
• Press-1 Campaigns for Interactive Voice Broadcasting
• Automated Phone Call Surveys and Polls
• Multi-User vs Multi-Tenant Call Center Software

<p>The post Proactive Welcome Calls Cut Churn in Half: What New 2026 Data Means for Call Center Software first appeared on ICTBroadcast.</p>

I run outbound campaigns for a living, so this is written from the operator’s chair. If you have ever watched a queue stall because the dialer guessed wrong, you already know the whole game is timing. Here is how modern predictive dialer software actually works, and where the rules bite.

What a predictive dialer really does

The core idea is simple. Instead of waiting for an agent to finish and then dialing one number, the system dials ahead. It looks at your live talk times, your drop rates, and how many agents are logged in, then it starts calls before anyone is free. When a human picks up, the call routes to whoever opens up next.

That is the difference between a predictive dialer and a power dialer. A power dialer fires one call per free agent. A predictive dialer fires more than that, betting that some calls will hit voicemail or ring out. Get the bet right and agents talk almost non-stop. Get it wrong and you either strand agents or drop live answers, which is where compliance comes in.

Inside the pacing algorithm

The pacing engine is the brain. It samples a moving picture of your campaign: average handle time, the share of dials that reach a human, wrap-up time, and current agent count. From those numbers it sets a dial ratio, the number of calls it launches per available agent.

The ratio is not fixed. If drops start creeping up, the engine backs off and dials fewer lines. If agents are sitting idle, it pushes harder. A good algorithm adjusts every few seconds, because the answer rate on a Tuesday morning list is nothing like a Friday afternoon one. This is the loop that keeps your call center software productive without breaking the rules.

Notice what falls out of the loop: no-answers, busy signals, and voicemail. Voicemail detection matters here. If the dialer connects an agent to a machine, you have wasted a seat and annoyed nobody but yourself. Modern detection screens those out and pushes only live humans to agents.

Why the whole industry moved to the cloud

On-prem dialers used to mean a rack of hardware, a telecom engineer, and a long wait for more capacity. Cloud predictive dialer software changed the math. You spin up agents in a browser, scale seats up for a campaign and back down when it ends, and pay for what you run.

The bigger win is that carrier features, caller-ID authentication, and analytics all live in one place and update centrally. You are not patching a box at 2am. For most teams building a modern call center software stack, the cloud model is now the default rather than the exception.

The compliance guardrails you cannot skip

Here is the part that separates a campaign that runs for years from one that gets a cease-and-desist. In the US, the TCPA framework expects predictive dialers to keep the abandonment rate under 3%, measured across a rolling 30-day window per campaign. An abandoned call is one where a person answers but no agent is free within about two seconds. Push your dial ratio too high and you blow past that cap fast.

So a serious dialer treats 3% as a ceiling the pacing engine respects on its own. It also scrubs do-not-call lists, honors local calling-hour windows, and signs your caller ID with STIR/SHAKEN so carriers can verify the number is really yours. Skip that last one and more of your calls get flagged as spam and never ring.

The rails are not there to slow you down. They are there so your numbers keep answering and your campaigns keep running. A blocked caller ID or a busted abandonment rate costs you far more reach than the tiny pacing headroom you gave up.

CRM and CTI: giving agents the full picture

A connected call is only half the job. The other half is what the agent sees when it lands. With CRM and CTI integration, the account record, past tickets, and call history pop on screen the instant the call connects. The agent knows who they are talking to before they say hello.

That context is what turns a dialed number into a real conversation. Screen pops cut handle time, warm transfers keep the customer from repeating themselves, and every disposition writes back to the record so the next touch is smarter. Pair that with call analytics and you can see which lists, agents, and scripts actually move the needle.

Predictive vs other dialer modes

Predictive is not always the right tool. For small lists or sensitive calls, a slower mode protects your relationships. Here is how the common modes stack up.

Where ICTBroadcast fits, and what is coming

ICTBroadcast runs predictive, power, preview, and progressive modes with voicemail detection, DNC handling, and call analytics built into the campaign, not bolted on afterward. You set the abandonment ceiling and the pacing engine holds it while it keeps agents busy. You can compare plans and seat counts on the packages page before you scale a campaign.

On the roadmap, AI features like an automated voice agent and live sentiment analysis are coming soon to ICTBroadcast. They are not shipping yet, so treat them as future additions rather than something you can switch on today. If you want a hand mapping your pacing and compliance setup, open a ticket at service.ictinnovations.com.

Frequently asked questions

How does a predictive dialer decide how many calls to place?

It reads live campaign data: average talk time, wrap-up time, how many dials reach a human, and how many agents are logged in. From that it sets a dial ratio and adjusts it every few seconds. If drops rise, it dials fewer lines. If agents idle, it dials more.

What is the 3% abandonment rule?

Under the TCPA framework, a predictive dialing campaign should keep abandoned calls under 3% of answered calls, measured across a rolling 30-day window. An abandoned call is one a person answers but no agent takes within about two seconds. A good dialer enforces that cap for you.

What does STIR/SHAKEN do for my calls?

STIR/SHAKEN signs your outbound caller ID so carriers can verify the number really belongs to you. Verified numbers are far less likely to get labeled as spam, which means more of your calls actually ring instead of getting blocked.

Is a cloud predictive dialer better than on-premise?

For most teams, yes. Cloud dialers scale seats up and down on demand, update carrier features and caller-ID authentication centrally, and drop the hardware maintenance. On-premise still suits a few tightly controlled environments, but the industry has largely moved to the cloud.

Do I always want predictive mode?

No. Predictive shines on big lists with lots of agents. For small lists, high-value accounts, or sensitive outreach, preview or power mode gives your agents more control and near-zero abandonment. Good call center software lets you switch modes per campaign.

Related resources

• ICTBroadcast Features
• Call Center Solution
• ICTBroadcast Packages and Pricing
• FCC Abandoned-Call Rule and Pacing

Want a predictive dialer with pacing and compliance built in? See what ICTBroadcast offers at ictbroadcast.com.

<p>The post How Predictive Dialers Work in 2026: Pacing, Abandonment, and Compliance first appeared on ICTBroadcast.</p>

What the 3% Rule Says Today

Right now, telemarketing rules cap abandoned calls at 3% of live answers, measured per campaign over a rolling 30-day window. An abandoned call is one where a person picks up but no agent is available within two seconds, so they get dead air or a recorded disclosure instead. The companion rule requires you to let the line ring for at least 15 seconds or four rings before you treat it as unanswered.

These two numbers shape everything about how aggressive your dialing can be. Push too hard and you blow past 3%. Play it too safe and your agents sit idle. That tension is exactly why predictive dialing exists.

Why Predictive Dialer Software Paces Around 3%

A predictive dialer places more calls than you have free agents, betting that some won’t answer. The algorithm watches your live answer rate, average handle time, and agent availability, then dials just ahead of the moment an agent frees up. When it guesses right, an agent is ready the instant someone says hello. When it guesses wrong, you get an abandoned call.

The 3% cap is the guardrail that keeps that betting honest. If you want to understand the mechanics in more depth, our explainer on what a predictive dialer is walks through the pacing loop step by step. The short version: your dial ratio, your agent count, and your answer rate all feed one equation, and 3% is the ceiling that equation must respect.

What the FNPRM Could Change

The FNPRM floats a few directions rather than a single fixed proposal. The ideas getting the most attention are:

• A tighter or per-call cap. Regulators are asking whether the rolling 30-day average lets bad actors hide bursts of dead air, and whether a shorter measurement window or lower percentage would protect consumers better.
• Rethinking the 15-second ring floor. With more people screening calls, the FCC wants comment on whether 15 seconds still matches how folks actually answer their phones.
• Clearer treatment of answering-machine detection. There’s interest in how AMD false positives inflate or hide abandonment, since a call wrongly tagged as a machine never reaches a person at all.

None of this is settled. But the direction of travel points toward more scrutiny of pacing, not less.

How to Tune Pacing and AMD Either Way

You don’t have to guess the outcome to get ahead of it. Whether the cap tightens or holds, disciplined pacing protects you. A few moves pay off in both scenarios:

• Run predictive mode only with enough agents (roughly eight or more) so the statistics hold; below that, switch to progressive or power dialing.
• Tune your AMD sensitivity carefully so real people aren’t misread as machines, then audit a sample of recordings weekly.
• Keep your rolling abandonment comfortably under target, not right at the line, so a bad hour doesn’t breach the cap.
• Log every disposition with timestamps so you can prove compliance if the measurement window shrinks.

ICTBroadcast is unified auto-dialer and call-center broadcasting software covering voice, SMS, fax, email, survey, and press-1 campaigns, with a predictive dialer, AMD, and multi-tenant white-label support. You can set pacing targets and AMD thresholds per campaign, which makes it straightforward to dial these settings tighter if the rules move. Our call center software overview shows how those controls sit alongside agent panels and reporting. AI-assisted pacing is on the roadmap and coming soon, but today the tuning is yours to set.

Compliance Isn’t Just the Cap

The abandonment percentage is one line in a longer checklist. Consent, calling windows, DNC scrubbing, and caller ID authentication all sit alongside it, and regulators tend to look at the whole picture. If you want a single place to review those obligations, our TCPA and STIR/SHAKEN outbound calling checklist lays them out in order so nothing slips.

Frequently Asked Questions

What counts as an abandoned call under the 3% rule?

An abandoned call is one where a live person answers but no agent connects within two seconds, so the caller hears silence or a recorded message. Machine answers and no-answers don’t count against the cap, which is why accurate answering-machine detection matters so much.

Is the FCC’s 2026 FNPRM already law?

No. A Further Notice of Proposed Rulemaking is a request for public comment, not a final rule. It signals which questions the FCC wants to explore. Any change would come later through a separate order, so you have time to prepare rather than react.

How many agents do I need for predictive dialing to stay under 3%?

As a rough floor, plan for eight or more active agents. Below that, the statistical model has too little data to pace accurately, and your abandonment rate gets volatile. Smaller teams usually get steadier results from progressive or power dialing.

Does answering-machine detection affect my abandonment rate?

Yes, in both directions. Overly aggressive AMD can tag real people as machines, which hides live answers and skews your numbers. Too-loose AMD passes machines to agents and wastes talk time. Tuning it well keeps your reported rate honest.

What should I change first if the cap gets tighter?

Start by lowering your dial ratio so your rolling abandonment sits comfortably below the new target, then tighten AMD sensitivity and review a sample of call recordings. Per-campaign pacing controls let you make these changes without rebuilding your setup.

Rules shift, but disciplined pacing is always the safe bet. If you want to see how per-campaign pacing and AMD controls work in practice, take ICTBroadcast for a spin and set your targets where you’re comfortable.

<p>The post The 3% Abandoned-Call Rule May Be Changing: What the FCC’s 2026 FNPRM Means for Predictive Dialer Pacing first appeared on ICTBroadcast.</p>