VICIfast
All guides

Spam Likely vs Scam Likely: what each label means

Spam Likely, Scam Likely, Potential Spam and Spam Risk are labels set by analytics engines, network by network. What each means and what drives them.

VICIfastLast updated

"Spam Likely" and "Scam Likely" are not two entries on one national blacklist. They are display strings that a carrier, or an app on the handset, puts on an incoming call after an analytics engine scores the caller ID. Different networks use different engines, so the same number can ring clean on one phone and arrive labelled on the next.

This guide covers what the labels are, which engine sits behind which network, how labelling differs from blocking, and what you can and cannot do about it from the dialer side. If you run VICIdial outbound, the useful part is the middle: why one number can look fine on AT&T and flagged on T-Mobile, and how you can see that in your own call logs.

What do "Spam Likely" and "Scam Likely" actually mean?

Both are labels a terminating carrier shows on the called party's screen when its analytics engine scores a caller ID as probably unwanted. "Spam" generally points at unwanted marketing volume. "Scam" generally points at suspected fraud. The label is a probability, not a finding that you broke a rule.

Nobody files your number onto a list. The engine watches traffic and complaints about the caller ID, scores it, and the carrier decides how to show that score. The decision is made per network, so there is no single place to check or clear it.

The exact wording varies by carrier, by calling app and by handset. Treat the four common strings below as families, not as a spec.

Label wordingGeneral meaningWhere you tend to see it
Spam LikelyScore suggests unwanted bulk callingCarrier-level display, handset caller ID
Scam LikelyScore suggests fraud or impersonationCarrier-level display, handset caller ID
Potential SpamSofter wording for the same kind of scoreVaries by carrier, app and handset
Spam RiskSofter wording, often shown with a risk levelVaries by carrier, app and handset

The difference between "Spam" and "Scam" wording is the engine's own tiering, not a legal distinction. Do not read more into it than that.

Which analytics company does each carrier use?

T-Mobile uses First Orion, AT&T uses Hiya, and Verizon is commonly reported to use TNS. The first two are documented by First Orion's own 2018 congressional testimony and Hiya's GSMA profile. The Verizon pairing is only commonly reported, so check it before you build anything on it.

NetworkAnalytics engineHow solid is it
T-MobileFirst OrionFirst Orion's 2018 congressional testimony
AT&THiyaHiya's GSMA profile
VerizonTNS (Call Guardian)Commonly reported; TNS says 3 of the top 5 US wireless carriers use Call Guardian
Other mobile, cable, VoIPVariesOften a mix of the same three plus handset apps

Two cautions. Carriers change vendors, and a carrier can run more than one engine for different products. And handset apps keep their own scores that do not have to match the carrier's.

That is why "which list am I on" has no answer. You are on several scores at once, each built from different traffic.

Is a label the same thing as a block?

No. A label lets the call ring and adds text. A block stops the call before the phone rings. The rules differ, the signalling differs, and the way you find out differs.

For analytics-based blocking, the FCC's Eighth Report and Order (FCC 25-15) allows only SIP 603+ and requires it to carry redress contact information, with full compliance from March 2026. That order does not cover labelling. Earlier, the FCC declined to extend required redress to erroneous call labeling.

In practice you see a block as a rising count of SIP 603 results on a trunk. You see a label as a drop in people answering, with no error code at all. The full comparison is in SIP 603+ vs Spam Likely.

Why can a number be clean on AT&T but flagged on T-Mobile?

Because each engine scores the number from its own traffic and its own complaint data, using its own thresholds. A number that placed 400 calls into T-Mobile subscribers and 40 into AT&T subscribers gives one engine ten times the evidence. Unequal exposure alone can produce unequal labels.

The engines also weigh signals differently. One may weight short calls heavily; another may lean on user reports. Neither tells you the weights.

You can see the effect from your side without reading any carrier label. The DID Reputation Center reads your VICIdial call logs and breaks pick-up rate down by the called person's network (T-Mobile, Verizon, AT&T, other mobile, landline/VoIP). If a caller ID is answered at a normal rate on three networks and at half that rate on one, a label on that network is a reasonable inference.

It is an inference, not a reading. We do not read carrier labels and we do not claim to show what a called person sees. The network is derived from the called number's exchange, so ported numbers blur it. Treat the breakdown as approximate and look for large gaps, not small ones.

What drives a label?

Mostly the shape of the traffic from one caller ID: volume, short calls, low answer rates, complaints, and little call history. No outside source publishes the engines' weights, so anything more precise than that list is guesswork.

  • Volume and pattern from one number. Matt Florell, VICIdial's author, said at Astricon in November 2021 that labels are based on the caller ID number and that "a minimum of 5 minutes and 200 phone calls can get you flagged". That is a field observation from 2021, not a current threshold. Use it as a reminder that the number is small.
  • Short calls and low answer. Calls that end in seconds, or that mostly go unanswered, look like bulk dialing to a scoring engine. Hiya's State of the Call 2026 found 86% of unknown calls go unanswered, so a low answer rate alone is the baseline, not a signal. The trouble starts when your number sits well below its peers.
  • Complaints. User reports and handset "report spam" taps feed the score. You do not see them.
  • Little history. Hiya reported in January 2025 that 18% of reported US spam calls come from numbers with minimal call history. A brand-new number has no record to argue with.

The last point matters for how you operate. A stable pool of numbers you own, used at a steady pace, builds history. A pool you keep swapping never does. The FCC is also asking whether to limit "number cycling", so churning numbers is a poor plan on both counts; see The FCC's number cycling proposal.

The engines look at the pattern, so the fixes are operational: fewer calls per number per day, more local numbers per area code, fewer instant hang-ups. The VICIdial side of that is in How to rest a flagged caller ID in VICIdial and VICIdial CID groups.

Does branded calling or CNAM change the label?

Not directly. Branded calling is a separate paid product from the analytics vendors and carriers that adds a business name, and sometimes a logo and call reason, to the call screen. CNAM is the older caller-name record, and it shows a name mostly on landlines and VoIP, rarely on mobile.

Neither is a switch that removes a label. They add identity; the spam score is still computed from behavior. A branded call from a number with poor pick-up history is still a call from a number with poor pick-up history.

If you are weighing the CNAM side, branded caller ID (CNAM) through your carrier covers what the record does and where it shows.

Does STIR/SHAKEN A attestation prevent labels?

No. STIR/SHAKEN signs the call to say the originating provider vouches for your right to use the number. It does not say anything about whether people want the calls. An A-attested call can be labelled.

The scale of signing makes this plain. TNS reported in February 2026 that 85% of Tier-1 traffic was signed in 2025, and 93% of that at A. It also reported that up to 13% of traffic from invalid numbers was signed A. A signature is table stakes, not a reputation score.

Get attestation right, because a missing or low one hurts. Just do not expect it to clear a label. The long version is in Why A attestation doesn't stop Spam Likely, with the basics in What is STIR/SHAKEN?.

What can a caller do about a label?

Monitor, rest the number, and file for redress where you are eligible. No one can promise a label will clear, including us.

  1. Measure pick-up per caller ID. Compare each number to your other numbers in the same campaigns and to its own past. The free health view shows each number's status on every VICIfast server.
  2. Rest the number. Stop dialing from it. Florell's 2021 talk put the time for a label to clear after a number stops being used at "2 to 21 days" - again, a 2021 observation.
  3. Register if you qualify. The Free Caller Registry, run by First Orion, Hiya and TNS, is for businesses that call on their own behalf. BPOs and other third parties must contact the providers directly. It "does not guarantee redress" and is "not a replacement for caller reputation monitoring".
  4. Use the carrier redress paths. USTelecom's redress best practices, adopted in April 2026, cover labelled as well as blocked calls. Members commit to a conclusion on at least 95% of qualifying requests within 2 business days. Participation is voluntary.
  5. Keep your caller ID lawful. Telemarketers must transmit a caller ID that lets the called party make a do-not-call request, and spoofing with intent to defraud is barred. If your practice touches either rule, talk to your counsel.

For the step-by-step version, see How to get Spam Likely removed and Avoiding Spam Likely labels on your outbound calls.

Spam Likely vs Scam Likely: which terms mean what?

One more pass, side by side. Wording differs by carrier, app and handset, so use this to classify what someone tells you, not to predict what a given phone will show.

TermTypeSet byStops the call?Typical causeClears by
Spam LikelyLabelCarrier analytics engineNoBulk-calling pattern, low answerResting, redress
Scam LikelyLabelCarrier analytics engineNoHigher-confidence fraud scoreResting, redress
Potential SpamLabelCarrier or handset appNoLower-confidence scoreResting, redress
Spam RiskLabelCarrier or handset appNoLower-confidence scoreResting, redress
SIP 603+ blockBlockTerminating networkYesAnalytics score over a block thresholdRedress contact in the 603+ response
Branded callAdded identityPaid productNoNot applicableDoes not clear labels
CNAM nameName recordCaller-name databaseNoNot applicableDoes not clear labels

If you want the pick-up numbers per caller ID, per network and per trunk in one place, the DID Reputation Center reads them from your VICIdial logs. It infers likely labels from pick-up; it does not read, remove or appeal them.

Sources