The rule everyone worries about is written for residential subscribers. The rule that actually reaches your calls keys on the type of number you dialled.
Read the regulations rather than the compliance blogs and the picture inverts. The national Do Not Call Registry, the 8am to 9pm window and the internal do-not-call list obligation are each expressed in the communications rules as applying to “any residential telephone subscriber”. A business line is not that.
Meanwhile the prohibition on automatic dialling systems and artificial or prerecorded voices is written to cover “any telephone number assigned to a paging service, cellular telephone service, specialized mobile radio service, or other radio common carrier service”. There is no residential qualifier anywhere in that paragraph. Your prospect’s mobile is covered whether or not the phone belongs to a company.
So the operative questions are not the ones usually asked. They are: what kind of number is this, and how am I dialling it.
Two regimes, and almost everyone conflates them
The trade commission’s rule is the more generous of the two. Its business-to-business exemption covers “telephone calls between a telemarketer and any business to induce the purchase of goods or services”, and because it exempts the call from the whole part, the Do Not Call machinery falls away with it. Two carve-outs survive, both about honesty rather than permission: the prohibitions on misrepresentation and on false statements made to induce payment.
But the exemption has a boundary that is stated in plain words, and it is the one most likely to catch a modern outbound team. The commission’s own compliance guidance says that calls to business lines “that solicit individual employees to buy products or services for their own use or make personal charitable contributions, also are not business-to-business solicitations and are not exempt”. Selling a seat licence to a company is B2B. Selling a personal subscription to someone who happens to be at work is not.
And there is a presumption that shifts the burden onto you. When the commission allowed wireless subscribers onto the registry, it reasoned that deciding whether a given wireless subscriber is residential “may be more fact-intensive” than for a wireline number, and resolved it by presuming that wireless subscribers who register are residential. Since a great many decision-makers’ working numbers are personal mobiles they registered years ago, the practical effect is that you are the one who has to be able to show the number is a business number.
The rule that was never in force
A large amount of compliance content, some of it still being sold, describes a one-to-one consent requirement as current law. It is not, and it never operated for a single day.
The rule was adopted in December 2023 with a compliance date in January 2025. Days before that date, a federal appellate court vacated the relevant part of the order, holding that the new consent restrictions “impermissibly conflict with the ordinary statutory meaning of ‘prior express consent’”. The agency then conformed its regulations and recorded the sequence in the Federal Register in a single decisive sentence: “Prior to the court’s mandate, the Commission had postponed the effective date of the revised rule and the revised rule had not gone into effect.”
We checked the live rule text rather than the commentary. The current definition of prior express consent contains no “one-to-one”, no “single seller” and no “logically and topically associated” language. It is the pre-2023 definition.
What did come into force, in April 2025, is a set of revocation rules, and they are worth knowing precisely because they are real. A person may revoke consent by any reasonable method; certain methods are reasonable per se, including replying with words such as stop, quit, end, revoke, opt out, cancel or unsubscribe; the request must be honoured within a reasonable time “not to exceed ten business days”; and a caller “may not designate an exclusive means to request revocation”. These govern robocalls, robotexts and registry-covered solicitations rather than manually dialled business calls, but the ten-day figure is the one to build your process around.
Authentication, attestation, and a myth worth killing
Carriers were required to implement the call authentication framework in their internet protocol networks by 30 June 2021. That much is in the regulations.
Attestation levels are not. The A, B and C labels everyone repeats appear nowhere in the Code of Federal Regulations. They come from an industry specification, and the clearest official description of them sits in the agency’s own 2020 order: full attestation where the provider can confirm both the identity of the subscriber and that the subscriber is entitled to the number, partial where it can confirm the subscriber but not the number, and gateway where it is merely the point of entry for a call that originated elsewhere.
Two consequences follow, and the second is the one that matters operationally.
Carriers are not required to block calls for lacking full attestation. There is no such rule. What does get blocked, at the provider level rather than the call level, is traffic from any voice provider whose filing is missing from the robocall mitigation database. Confusing the two produces a lot of wasted effort.
Your dialler probably cannot get full attestation, and the agency has said so. The 2020 order acknowledges that where an enterprise places outbound calls through a provider other than the one that assigned its numbers, the call “will not pass through the authentication service of the voice service provider that controls the numbering resource”. That is the ordinary configuration for a team using a third-party dialler or a communications platform. Expect partial attestation, and understand that this is a structural artefact of how your stack is wired rather than a judgement about your calls.
Being labelled “Spam Likely”, and the free way out
No regulation governs spam labelling. It is done by private analytics engines under contract to carriers, largely on the strength of user reports and calling patterns. That has one good consequence: because it is not a sanction, there is a remediation path, and it is free.
A single industry form submits your numbers to the three main analytics providers at once. Separately, the main trade body publishes a directory of per-carrier and per-analytics redress contacts, covering roughly seventeen carriers and five analytics and registry services, with direct links to each one’s review process. Neither charges a fee.
Two honest caveats. The registry form says in its own words that “registration of phone numbers does not guarantee redress”, and that it is not a substitute for monitoring your own reputation. And branded calling, where your company name renders on the handset, is a paid commercial product, not an entitlement.
The operational lesson. If connect rates fall without anything else changing, check the label before rewriting the script. Call your own outbound numbers from a handset running one of the common screening apps. Changing numbers without understanding why the old one was flagged simply restarts the cycle.
Note also what is not required: displaying your company name. The rule requires that caller identification carry a number, and that the number “permit any individual to make a do-not-call request during regular business hours”. A working callback line is the obligation. The name is a product you buy.
The state that does reach business calls
Federal law leaves B2B calling hours unregulated. Some states do not, and the most instructive is Florida, because the usual summary of it is backwards.
Two statutes operate in Florida. The consumer-facing one defines its subject as a call “to a consumer” and does not reach genuine business calls. The Telemarketing Act does reach them, and it contains a business-to-business exemption conditioned on three years of continuous trading under the same name and at least half of dollar volume coming from repeat sales to existing businesses.
The trap is in how the exemption is written. The provision opens: “The provisions of this part, except ss. 501.608 and 501.616(6) and (7), do not apply to”, and the excepted section is the one imposing the 8am to 8pm window and the three-calls-per-24-hours cap on the same subject. An exempt B2B caller in Florida is still bound by both.
We checked one state’s text directly and will not vouch for summaries of others. The lesson generalises anyway: the federal position is permissive for B2B, and the exposure sits in state law, which has to be read one state at a time rather than taken from a compliance vendor’s map.
Recording, where the answer is simple
Federal law permits recording with the consent of one party, which can be your own representative. Roughly eleven states require all parties to consent, and a handful more require it for some conversation types but not others. The compilations that list them, including the most careful one, hedge with words like “about” and “primarily”, which reflects real legal ambiguity rather than sloppiness.
Courts also disagree on whether the law of the recorder’s location or the recorded party’s location governs an interstate call, and you cannot reliably infer where a mobile user is standing from their area code.
So the per-state routing rule that vendors sell is fragile by construction. The robust answer is duller: announce recording on every call and capture the response. There is no jurisdiction where clear disclosure plus consent is insufficient.
Nobody neutral publishes connect rates
We looked specifically, because this is the number every plan is built on.
The most widely quoted average comes from a platform whose dataset is its own customers’ recorded calls. That is a self-selected population of the most heavily instrumented sales teams in the world, described as though it were a population estimate. The figure may well be right for companies like those. It is not a benchmark.
The most interesting number we found is not a conversion rate at all. In a 2011 experiment, fifty agents placed 6,264 calls over two weeks: 28 per cent were answered, 55 per cent were not, and 17 per cent were non-working numbers. It is residential real estate, it is fifteen years old, and it was run with a franchise partner that supplied the script, so treat it accordingly. But the one durable thing in it is the 17 per cent, and that is a statement about list decay rather than about persuasion or scripting.
Where a US B2B phone list comes from, and why it decays
We queried the data rather than reading about it, and the result is consistent across states.
Colorado’s full business entity dataset returns 34 fields: entity identifier and name, principal and mailing addresses, status, jurisdiction, entity type, registered agent name and addresses, formation date. No phone field. The same check against Oregon’s and New York’s published business datasets returns the same answer.
The federal contractor registry does hold point-of-contact phone numbers, but its own data dictionary places them in the restricted tier alongside email and fax, while names and postal addresses sit in the public tier. A commercial marketer cannot reach them. The statistical business register is confidential by statute.
This is structural rather than accidental. Registries exist so that a company can be served with legal process. That requires an address. A phone number serves no statutory purpose, so states never ask for one and therefore never hold one.
Which means every commercial US B2B phone list is assembled privately, by scraping, data co-operatives, contributed contact data, purchase and inference, then matched onto the public registry spine. The firmographics can be traced to a public record. The phone number cannot. That provenance gap is also the best explanation for a 17 per cent non-working rate: nobody in the chain has a source of truth to correct against.
What to do with this
Sort your list by number type before you sort it by anything else. Direct dial landlines, switchboards and mobiles carry different rules, and the only one that constrains you is the mobile, where auto-dialling and prerecorded voices are out regardless of business status.
Dial mobiles by hand, and reserve any automation for landlines. That single rule removes most of the federal exposure from a B2B programme, and it costs you less than the compliance tooling sold to manage the alternative.
Check your caller ID before you change your script. Call your own numbers from a screened handset, use the free remediation forms if you are labelled, and expect partial attestation if your dialler is not the carrier that owns your numbers.
Announce recording on every call, everywhere, and stop trying to route by state.
And treat every published connect rate as a sales document. The number worth tracking is your own non-working rate, because it tells you something true about your list that no vendor benchmark can.
The related pieces are what US law actually allows in cold email and running a cold email campaign.