Your numbers are showing as spam on customer handsets and your connect rate is paying for it. This is the full remediation runbook: who applies the labels, where to dispute them, and the number hygiene that stops them coming back.
How do I fix a Spam Likely label on my outbound caller ID?
Register the affected numbers once at freecallerregistry.com, which feeds all three carrier analytics engines (TNS for Verizon, First Orion for T-Mobile, Hiya for AT&T), then file disputes in each engine's own portal for numbers already labeled. In our remediation work labels clear in days to a couple of weeks, and they stay clear only if you also fix the calling behavior that triggered them.
It is not the carrier you think it is, and there are three of them.
Start with the uncomfortable part: nobody at Verizon looked at your number and decided you were a nuisance. Handset labels come from analytics engines the carriers contract, and each major US network uses a different one. TNS scores traffic for Verizon, and also for US Cellular, Spectrum and C Spire. First Orion scores for T-Mobile and Metro. Hiya scores for AT&T and powers the native call screening on Samsung devices. USTelecom's Industry Traceback Group publishes the canonical list of labeling and blocking points of contact; bookmark it, because the pairings and portals do shift over time.
Each engine keeps its own reputation database, its own scoring model and its own remediation queue. Clearing a number with TNS does nothing for how First Orion sees it, which is why the same DID can display clean on a Verizon handset and flagged on a T-Mobile one the same afternoon. The wording differs too: Verizon subscribers see "Potential Spam", T-Mobile shows "Scam Likely", and AT&T uses "Spam Risk" or "Fraud Risk" depending on severity. The exact string varies by network, handset and screening app, but operators search for "Spam Likely", so we will use it here for the whole family of labels.
The practical consequence is that remediation is a three-front exercise. Any fix that only talks to one engine, or worse, only to your own originating carrier, leaves most of the mobile market seeing exactly what it saw before. When a client tells us they are flagged, our first diagnostic question is always the same: flagged on whose handset?
The models score behavior, and they cannot see your consent records.
None of the engines publish their scoring models, but they are reasonably open about the inputs, and the inputs are behavioral. The engines watch what a number does, not what its owner intends. The signals that recur across engine documentation and across the floors we audit:
The Hiya figures come from Hiya's own write-up of its data sources, and its guidance on why calls get flagged lists the same behavioral factors: call patterns, quick hangups, complaint history and unregistered numbers. The other engines describe similar inputs. Notice what is absent: whether your calls are lawful, consented and useful. The models cannot see consent. A mortgage servicer calling its own book with documented opt-ins can profile identically to a scam operation if it dials hard, connects rarely and talks briefly. That is the most misunderstood point in this whole area, and it is why full STIR/SHAKEN attestation helps but does not immunize you. Attestation is one input among many, not a clean bill of health.
Our free Five9 Health Check is 17 questions across dialer settings, DID strategy and deliverability. It takes about ten minutes and tells you where your configuration is working against your connect rate.
One form, all three engines, no charge.
The Free Caller Registry is the one genuinely shared piece of infrastructure in this mess. TNS, First Orion and Hiya jointly operate a single submission form that distributes your registration to all three engines at once. You provide business details (name, address, website), a contact with a verifiable email, your call purpose and monthly volume, and up to 20 US numbers entered individually, with a file upload for larger batches. Each engine then acknowledges receipt separately and processes the registration against its own database.
It is free, which has not stopped a cottage industry of resellers charging for it. If a vendor quotes you a fee for "carrier registration" of US numbers, what you are usually buying is somebody else filling in this form.
Be clear-eyed about what registration does, because the registry itself is honest about it: registration establishes your numbers as belonging to an identifiable business, it does not guarantee removal of an existing label, and the site states plainly that it is not a substitute for reputation monitoring. Resubmitting the same numbers adds nothing. In our experience the engine acknowledgments arrive within a few business days, and First Orion's own registration page says numbers are registered "within days". Treat the registry as identity establishment, step one of four, and expect to do the dispute work separately for any number already carrying a label.
Per number, with evidence, on all three fronts.
Registration rarely removes an existing label on its own. That work happens in each engine's own queue, and the entry points are different for each. The table reflects the USTelecom points-of-contact list as of mid 2026; check the live page before filing, because portals get renamed and redirected. Case in point: calltransparency.com, long quoted as First Orion's dispute address, now redirects to First Orion's registration page.
| Engine | Scores for | Typical label | Where to go |
|---|---|---|---|
| TNS | Verizon, US Cellular, Spectrum, C Spire | Potential Spam | voicespamfeedback.com, reportarobocall.com/trf, or email communications@tnsi.com |
| First Orion | T-Mobile, Metro | Scam Likely | firstorion.com/register-your-number (calltransparency.com redirects here); T-Mobile also runs callreporting.t-mobile.com |
| Hiya | AT&T, Samsung native screening | Spam Risk / Fraud Risk | att.com/reviewmycalllabel, or Hiya's support portal at hiya.com/manageyourcallerid |
File disputes per number, not per company, and attach substance: who you are, what the calls are, the consent basis, realistic daily volumes, and a website that matches the email domain you filed from. Disputes submitted from a Gmail address on behalf of a company with no discoverable web presence go nowhere, and viewed from the engine's side of the desk, they probably should.
On timelines, nobody commits to one, and we would distrust anyone who did. Our experience across client remediations: straightforward mislabels on registered numbers clear in a few business days to two weeks per engine. Numbers with genuine complaint history take longer, and sometimes only step down in severity rather than clear outright. Silence for more than two weeks usually means the dispute was rejected, and the right move is to refile with more evidence, not to keep waiting.
Labels are usually discovered weeks late. They do not need to be.
Most teams learn about a label from a customer who mentions it in passing, by which time the damage is a month of missed connects. Monitoring is cheap by comparison, and there are four layers worth running.
The bluntest instrument still works best: keep a test handset (or eSIM profile) on each of the three major networks and ring them from every campaign DID on a weekly cycle. Ten minutes of work, and it shows you exactly what recipients see, label, name display and all.
Second, your own dialer telemetry. Per-DID answer rate is the leading indicator; a number that picks up a label shows an answer-rate cliff days before any human tells you. We flag any DID whose answer rate drops sharply against its own baseline and treat it as labeled until proven otherwise. If the whole campaign's answer rate has sagged at once, that is a different diagnosis, covered in our runbook for a dropped answer rate.
Third, blocking, as distinct from labeling, now announces itself. Under the FCC's Eighth Call Blocking Report and Order, adopted February 2025, terminating providers that block calls based on analytics must immediately return SIP code 603+ with redress contact information in the response, and the industry-wide compliance deadline of March 25, 2026 has now passed. Ask your carrier or CCaaS provider how 603+ responses surface in your logs. Rules here move quickly and enforcement practice is still settling, so verify against the current text and take qualified advice where real money rides on the answer.
Fourth, commercial caller ID reputation monitoring. These services poll engine data or run device farms across carriers and alert you when a label appears. They cost real money and the good ones earn it at scale; below a few dozen DIDs, the handset-and-telemetry approach covers most of the need.
Everything above is triage. This is the cure.
The honest answer on safe volume is that no engine publishes a threshold, so anyone quoting you an exact safe number of dials per DID per day is guessing. What we can tell you is how we run the floors we manage: volume spread across enough DIDs that no single number works absurdly hard, per-DID answer rate watched daily, and numbers rested when their answer rate decays rather than on a fixed rotation clock.
Rotation discipline matters more than rotation speed. Resting a tired number for a few weeks while its reputation recovers is hygiene. Churning through fresh number blocks every week is snowshoeing, and it is precisely one of the patterns the engines are built to catch: brand-new numbers with no history, high velocity and thin registration records look like a spam operation warming inventory, because that is what a spam operation looks like. The same caution applies double to local presence dialing, where the temptation to buy large pools of local DIDs collides head-on with reputation management.
Feed the answer-rate signal back into list strategy, because the engines are effectively scoring your list quality. Dead records, wrong numbers and exhausted leads all convert into short calls and unanswered dials, which convert into labels. Cleaning the list, capping attempts per record and honoring opt-outs promptly does more for caller ID reputation than any dispute filing. Keep identity consistent while you do it: every number registered, CNAM stable, new DIDs warmed up on low volume before they carry a campaign. The full structure, pool sizing, warm-up and retirement included, is in our DID strategy guide.
Verified name and logo on the handset, priced per call.
Branded calling is the industry's paid answer to the trust problem: instead of a bare number or a warning label, the recipient sees a verified business name, logo and call reason. The current standards-led effort is Branded Calling ID (BCID), run by the wireless industry body CTIA, which carries verified Rich Call Data inside the STIR/SHAKEN authentication chain, with enterprises vetted by authorized partners before anything displays. It stopped being a slideware concept a while ago: Verizon signed in September 2025 to bring BCID to its network, and the analytics engines sell their own branded display products on the networks they score.
Costs are per call and they are material at outbound scale. TransNexus, which operates in the BCID ecosystem as a signing agent, cites published market rates of 3 to 12 cents per branded call. Run the arithmetic on your daily dial volume before signing anything: at 5 cents, 50,000 branded dials a day is 2,500 dollars a day. For high-value, lower-volume calling (collections on large balances, mortgage retention, appointment confirmation) that math can work comfortably. For high-volume cold outbound it usually does not.
Two honest limits. First, branding does not exempt you from analytics. The engines keep scoring behavior, and a badly run campaign can still earn a label or a block regardless of the logo; the BCID ecosystem's vetting standards explicitly exclude non-compliant traffic, so branding is a privilege for clean operations, not a paint job for dirty ones. Second, coverage is per network and still expanding, so verify current carrier support at the moment you buy rather than trusting a vendor deck, or for that matter this article, dated even a few months earlier.
Remediation without behavior change is a subscription to disputes.
Here is the cycle we see constantly in audits: a team discovers labels, grinds through disputes or pays someone to, celebrates the clean handsets, changes nothing about how they dial, and is labeled again inside a month or two. The engines re-score continuously on live behavior. A cleared number that resumes the same velocity, the same short average talk time and the same weak answer rate walks straight back into the same classification, and second labels are harder to shift because now there is history on file.
The cadence that holds is the four steps run as a loop, not a project. Register every number once at the Free Caller Registry. Dispute actively labeled numbers with each engine, per number, with evidence. Fix the behavior that earned the label: volume spread, list quality, attempt caps, rest rules. Then monitor weekly so the next label is caught in days rather than weeks. Remediation is the visible part of the work. The larger part is running outbound in a way that does not look like spam to a model that cannot see your consent records, and that part never finishes.
Not the recipient's carrier directly. Each major US carrier contracts an analytics engine: TNS scores for Verizon (plus US Cellular, Spectrum and C Spire), First Orion for T-Mobile and Metro, and Hiya for AT&T and Samsung's native screening. Each engine keeps a separate reputation database, so a number can display clean on one network and flagged on another, and remediation has to be done with each engine separately.
No engine commits to a timeline. In our remediation work, straightforward mislabels on properly registered numbers clear in a few business days to two weeks per engine. Numbers with real complaint history take longer and sometimes only step down in severity rather than clear. If you hear nothing for more than two weeks, assume the dispute failed and refile with stronger evidence: business identity, consent basis and realistic volumes.
Yes on both counts. It is jointly operated by the three analytics engines themselves (TNS, First Orion and Hiya) and distributes one registration to all three at no charge. Plenty of vendors resell the same submission for a fee. Its limits are stated on the site: registration establishes your business identity but does not guarantee removal of an existing label, and it is not a monitoring service.
The scoring models cannot see consent. They score behavior: dials per number, call durations, answer rates, complaint reports and hits on honeypot numbers. A fully consented campaign that dials hard, connects rarely and talks briefly produces the same behavioral signature as a nuisance operation. Fixing the label long term means changing the observable pattern, usually by spreading volume across more numbers, tightening list quality and capping attempts per record.
No. Attestation confirms the caller has the right to use the number; it says nothing about whether recipients want the calls. The engines treat attestation as one input among many, so A-attestation traffic still gets labeled when its behavior looks like spam. Weak attestation does make things worse, so it is necessary hygiene, just not sufficient on its own.
Nobody publishes a threshold, and anyone quoting an exact safe number is guessing. The engines score velocity in context: registration, answer rate, durations and complaints all interact. Our operating practice is to spread volume so no single DID works excessively hard, watch per-DID answer rate daily, and rest any number whose answer rate decays against its own baseline. That behavioral trigger matters more than any fixed daily cap.
It depends on call economics. Published market rates cited by TransNexus run 3 to 12 cents per branded call, which is material at outbound scale. For high-value, lower-volume calls such as collections or mortgage retention, the uplift in answers can justify it. For high-volume cold outbound it rarely does. Branding also does not exempt you from analytics scoring: badly behaved campaigns still get labeled, logo or not.
Briefly, then it usually makes things worse. Fresh numbers carry no positive history, and churning through new blocks at high velocity is exactly the pattern the engines associate with spam operations warming inventory. The labels follow the behavior onto the new numbers, and you have taught the models that your organization cycles numbers. Rest and remediate existing DIDs, register everything, and fix the underlying dialing pattern instead.
We run outbound floors daily, and caller ID reputation is part of how we manage them: registration, disputes, DID rotation and weekly monitoring as standing routine. If your numbers are flagged now, or you want them never to be, talk to us.
Talk to us