Who actually applies the Spam Likely label?
A business number shows up as Spam Likely because an analytics engine inside the terminating carrier’s network scored its recent behavior, not because the carrier that placed the call labeled it. The receiving carrier turns that score into a label. Koda Voice, as an originating carrier, controls who it onboards, how calls are signed and how traffic is monitored.
Three vendors describe the engines they supply to carriers. Hiya states on its own site that its service is embedded in mobile operators’ networks and handsets to identify and stop spam and fraud calls, and its blog records a 2018 AT&T supplier award for its contribution to developing AT&T Call Protect. TNS describes Call Guardian as a reputation engine built from real-time cross-carrier call events and crowd-sourced data. First Orion lists reputation monitoring, branded calling and call-blocking products and names T-Mobile among the businesses it serves. Each statement is the vendor’s own description of its product.
A call can therefore leave the originating carrier clean, cross the interconnect normally and still reach the handset with a label. The label is a prediction about the number, computed where the call lands.
What the labels mean: Spam Likely, Scam Likely, Spam Risk, Potential Spam
The labels Spam Likely, Scam Likely, Spam Risk and Potential Spam are different wordings of one output: a reputation score crossed a threshold that the carrier receiving the call chose. Each carrier sets its own wording and tiers, and the handset software displays them; scam or fraud wordings generally mark the higher-severity tier.
Three things a label is not:
- A label is not a legal finding. The engine predicted, from patterns, that the recipient would not want it.
- A label is not a STIR/SHAKEN result. Attestation and labeling are separate systems, covered below.
- A label is not permanent. Scores move with behavior, so a number that stops producing the signals below can recover.
What behavior triggers a spam label?
Behavioral signals about the calling number are what analytics engines weigh. The vendors do not publish their weightings, but TNS describes its inputs as real-time call events combined with crowd-sourced data: how the number behaves on the network, and what recipients say about it.
| Signal | What the engine observes | Why it reads as spam |
|---|---|---|
| Volume per number | A large burst of attempts from one number in a short window | A single business line does not normally behave that way |
| Short call duration | Calls that end within seconds of being answered | Recipients hang up on calls they did not want |
| Low answer rate | Most attempts ring out or reach voicemail | People screen numbers they do not recognize |
| Complaints and blocks | Recipients report or block the number on the handset | TNS names crowd-sourced data as an input |
| Calls outside daytime hours | Attempts before 8 a.m. or after 9 p.m. in the called party’s time zone | Outside the federal window under 47 CFR 64.1200(c)(1); draws complaints |
| Number churn | Rotating through fresh numbers as each one is labeled | A pattern associated with callers trying to outrun labels |
A small team doing legitimate customer follow-up from one number, at high call volume, with most attempts reaching voicemail, produces the same signature as an unwanted caller. The engine sees the pattern, not the intent.
Why is my new business number already showing up as spam?
A new business number shows up as spam early because the engine has nothing else to weigh. With no history, the first days of traffic are the whole record, and high call volume with short durations from day one looks like a number acquired to send spam. Two other causes are common:
- Recycled numbers. Numbers are reassigned, and a number used badly by its previous holder can carry that reputation into your account; the engine has no way to know the holder changed.
- Concentration. Putting a whole team’s outreach conversations on one new number compresses the volume signal into one place.
The remedy is not to buy more numbers and rotate them, which engines read as churn. The remedy is to introduce a number the way a real business line behaves: moderate volume, conversations that last, consistent use over weeks. Koda provisions local numbers and ports existing ones; see the phone numbers page.
Does STIR/SHAKEN attestation prevent spam labels?
STIR/SHAKEN attestation does not prevent spam labels, because attestation answers a different question. Under the SHAKEN standard ATIS-1000074, the signing provider gives full (A) attestation when it originated the call, has a direct authenticated relationship with the customer and has verified the customer’s association with the calling number; partial (B) when it knows the customer but has not verified the number; and gateway (C) when it has no relationship with the call’s originator. The value travels in the signed token as the attest claim defined in RFC 8588. All of that describes identity and authorization, not behavior.
| Question | STIR/SHAKEN attestation | Carrier analytics |
|---|---|---|
| Who answers it | The originating carrier, at signing | The terminating carrier’s engine, at receipt |
| What it asserts | How well the signer knows the caller and the caller’s right to the number | Whether the number’s recent behavior looks unwanted |
| Output | A, B or C in the signed token | A score, and a label when the score crosses a threshold |
So a call can carry A attestation and still be labeled. What attestation adds is a verified identity behind the call and the originating carrier’s own name on it, which is why Koda signs under its own certificate rather than a reseller’s; see the STIR/SHAKEN page and the attestation levels article.
What actually helps: registration, disputes and calling patterns
What actually helps is changing the signals first and using registration and dispute channels second; a dispute filed while the behavior continues is likely to see the label return.
- Fix the traffic pattern. Spread outreach conversations across numbers in proportion to real, ongoing use, keep each number in service for months, and give conversations room to last.
- Stay inside the federal calling window. Under 47 CFR 64.1200(c)(1), telephone solicitations may not be made to a residential subscriber before 8 a.m. or after 9 p.m. local time at the called party’s location. Solicitation calls outside that window fall outside the federal rule and draw complaints.
- Call people who expect the call. Consent-first practice reduces complaint and block signals directly. A recipient who asked for the call has little reason to report it.
- Register the business behind the numbers. Some analytics vendors and carriers offer channels where a business can register its identity and its numbers. Ask your carrier for the current ones; this page does not reproduce addresses that change.
- Dispute a specific label through the carrier that applied it. The label is applied where the call lands, so a label seen on AT&T, T-Mobile or Verizon handsets is disputed with that carrier; ask it for its current review process. Expect to show business identity and use case, and expect the result to hold only if the pattern that earned the label has stopped.
- Check the carrier you send through. Under Federal Communications Commission (FCC) rules at 47 CFR 64.6305, providers may accept calls directly from a domestic voice service provider only if that provider’s filing appears in the Robocall Mitigation Database (RMD). An unlisted carrier is one that other providers may not accept traffic from directly. A carrier that signs under another provider’s certificate is not the signer of record for your calls, so the attestation your numbers receive depends on that other provider’s relationship with you. The carrier verification checklist walks through the lookups.
What can a carrier do about labels, and what can it not do?
A carrier can shape every signal that originates on its side of the call and none of the scoring on the other side; it sells the trunk and the numbers, so its leverage sits upstream of the label.
| The originating carrier controls | The originating carrier does not control |
|---|---|
| Who gets onboarded: Know-Your-Customer (KYC) review of the business, use case and consent practices | The analytics engine’s scoring model |
| How calls are signed: its own certificate and the attestation level for each customer’s numbers | The threshold at which a terminating carrier turns a score into a label |
| Which numbers a customer uses and whether the customer is authorized to use them | Which label wording the handset displays |
| Per-customer monitoring of answer-seizure ratio and call duration | Reports and blocks filed by recipients |
| Acceptable-use enforcement, including suspension when traffic does not match the stated use | Removal of a label already applied |
| Cooperation with the Industry Traceback Group (ITG), the FCC-designated traceback consortium | Reputation inherited from a number’s previous holder |
The honest summary: a carrier cannot promise a label-free number. It can state what it controls: its customers are vetted at onboarding, its calls are signed in its own name, and it monitors answer-seizure ratio and call duration per customer so that changes in behavior surface early.
How this applies to Koda Voice
Koda Voice LLC is a licensed U.S. voice carrier for conversational business calling. Its number-reputation posture is the set of controls above, stated as policy, not results.
- Vetted customers. Koda runs a Know-Your-Customer review at onboarding, publishes an acceptable-use policy, monitors traffic quality per customer and per route, and cooperates with industry traceback, as summarized in the robocall-mitigation summary. The onboarding questions are previewed in the KYC article.
- Signing in its own name. Koda Voice is an authorized STIR/SHAKEN service provider with the STI Policy Administrator (STI-PA), iconectiv, under Service Provider Code 583H, and holds an STI certificate issued by an approved STI Certification Authority. Calls originated on Koda’s network are signed under Koda’s own Service Provider Code, not a reseller’s certificate. Attestation follows the framework: A when the customer is verified and authorized to use the number, B when the customer is verified but number authorization cannot be confirmed, C otherwise.
- Registry standing. Koda Voice is listed in the FCC Robocall Mitigation Database and holds Operating Company Number (OCN) 583H, assigned by the National Exchange Carrier Association (NECA). Both are verifiable in the Trust Center.
- Numbers and calling-hour awareness. Koda provisions local numbers across U.S. markets, ports existing numbers and supports caller-ID name (CNAM). Calling-hour awareness and number-reputation monitoring are part of the service, as the real estate page shows for prospecting conversations.
Koda does not offer label removal or a promise about any engine’s score. If your numbers are labeled today, contact Koda with the numbers, traffic pattern and use case; an engineer will look at the signals with you.
This article is a summary, not legal advice.
FAQ
Frequently asked questions
Who decides that a call is Spam Likely?
The terminating carrier's analytics engine decides. Vendors such as Hiya, TNS and First Orion supply engines that score calling numbers; TNS describes its inputs as real-time call events and crowd-sourced data. The carrier receiving the call sets the threshold at which a score becomes a label. The originating carrier does not apply the label and cannot remove it directly.
What is the difference between Spam Likely and Scam Likely?
Spam Likely and Scam Likely are two wordings of the same mechanism: a reputation score that crossed a threshold set by the carrier receiving the call. Each carrier chooses its own words and tiers; scam or fraud wordings generally mark the higher-severity tier. Neither label is a legal finding, and neither is a STIR/SHAKEN result.
Why is my new business number already showing up as spam?
A new number has no history of its own, so an engine can only weigh its first days and its previous holder's record. Heavy volume from day one, short calls and low answer rates on a number with no track record read as spam, and a recycled number may inherit the reputation its previous holder earned.
Does A attestation stop Spam Likely labels?
A attestation does not stop spam labels. Under ATIS-1000074, attestation states how confident the signing carrier is that the caller is entitled to the number; it says nothing about how the number has behaved. Analytics engines score behavior separately, so a number signed with full attestation can still be labeled if its calling patterns look like spam.
Can my carrier remove a spam label from my number?
A carrier cannot remove a label that another carrier's engine applied. An originating carrier can sign calls under its own certificate at an attestation level it stands behind, monitor each customer's answer-seizure ratio and call duration so problems surface early, help bring a number into service like a business line, and point you to registration and dispute channels.