A Spam Likely label is the most expensive thing that can happen to an outbound team, because it is invisible. Nobody tells you. Connect rates fall, reps assume the list got worse, and six weeks later somebody calls their own desk number from a personal phone and sees it.
Who actually decides
Not the carriers directly. Three analytics companies, each contracted to a different network:
| Engine | Network |
|---|---|
| Hiya | AT&T |
| TNS | Verizon |
| First Orion | T-Mobile |
They score independently, using their own models and their own data. A number can be clean on AT&T and flagged on T-Mobile. There is no central registry, no single appeal, and no way to check all three at once without a paid monitoring service.
That fragmentation is the practical fact most guides skip. Remediation means filing with each engine separately.
What they measure
None of them publishes its model. The signals are consistently reported across the industry as:
Call volume from a single number. High and sustained volume from one number is the entry condition for scrutiny.
Average call duration. Short calls look like robocalls. A number whose average call is nine seconds is behaving like a dialer that hangs up on voicemail.
Answer rate. The strongest signal. A number that nobody answers is, statistically, a number nobody wants to hear from.
Abandoned calls. A connected call with nobody on the line is the single clearest robocall signature. This is why parallel dialing carries the highest flagging risk of any dialing mode.
User reports. Anyone can report a number in their carrier's app. A handful of reports on a low-volume number moves it faster than thousands of calls on a well-behaved one.
Complaint velocity. Reports clustered in time weigh more than the same number spread over months.
The threshold question
The most common question is how many calls a day is safe, and there is no published answer.
What is observable is that flagging commonly begins somewhere north of 100 to 150 dials a day from a single number when the answer rate is low. That figure is a rule of thumb from operator experience, not a rule from the engines.
The ratio matters far more than the volume. Consider two numbers:
| Number A | Number B | |
|---|---|---|
| Dials per day | 200 | 80 |
| Answer rate | 20% | 3% |
| Average duration | 3 min | 12 sec |
| Flagging risk | Low | High |
Number A places more than twice the calls and is much safer, because everything about its behaviour looks like a person having conversations. Number B looks like a machine.
That is the useful insight on this page: the fix for spam labelling is usually a better list, not more numbers.
Why STIR/SHAKEN does not help
STIR/SHAKEN authenticates caller ID. An A attestation means your carrier knows who you are and has verified your right to use the number.
It says nothing about behaviour. An A-attested number placing four hundred short unanswered calls a day is an authenticated robocall, and the analytics engines will label it accordingly. Vendors that imply STIR/SHAKEN attestation prevents spam labels are either confused or selling something.
What actually works
1. Rotate numbers, properly. Splitting 300 daily dials across six numbers puts each at 50 a day, comfortably below the range where flagging usually starts. This is the reason serious dialers give you DIDs in quantity: Readymode includes 30 to 75 per licence, and Koncert sells managed caller ID reputation with a health heat map. On a bring-your-own-Twilio setup, numbers cost $1.15 a month, so a pool of ten costs $11.50.
2. Use local presence. A number matching the prospect's area code is answered more often, and answer rate is the strongest positive signal. Local presence improves connect rate and number health at the same time, which is unusual.
3. Fix the list. Every disconnected number, every wrong number and every never-answers record pushes your answer rate down. Scrubbing invalid numbers before dialing improves reputation directly.
4. Stop calling people who never answer. After six to eight attempts with no answer, the record is costing you reputation with no upside. Dispositioning it out is a number-health decision as much as a productivity one.
5. Register for branded caller ID where available. Several carriers support displaying your business name rather than a bare number, usually via your carrier or a CNAM provider. It raises answer rates, which improves everything downstream.
6. Monitor. Paid services check your numbers against all three engines. Without one, you will find out six weeks late.
Remediation, step by step
- Confirm the label. Check your numbers on real devices across all three networks, or use a monitoring service.
- File with each engine separately. Hiya, TNS and First Orion each have their own business remediation process. They do not share decisions.
- Fix the behaviour before you file. A remediation request on a number that is still placing 400 short calls a day will be reversed.
- Rest the number. Take a flagged number out of rotation for a few weeks rather than continuing to dial from it.
- Do not simply buy a new number. This is the most common mistake. The behaviour caused the label; a new number reproduces it within weeks. Teams that churn through numbers monthly have a list problem they are refusing to look at.
The parallel dialing warning
If you are running or considering a parallel dialer, this is the cost that is not on the pricing page.
Parallel dialing produces abandoned calls by design, generates a high volume of short calls, and depresses average duration, which are three of the four strongest negative signals. That is why every credible vendor in that category sells number rotation, and why the ones that do not are leaving you a bill you will pay in connect rate six weeks after signing.
Before buying a parallel dialer, ask exactly two questions: how many numbers do I get, and how does rotation work. If the answers are vague, the spam risk is yours.
Frequently asked questions
- Why does my number show as Spam Likely?
- Because a carrier analytics engine scored your calling behaviour as robocall-like. The signals are call volume from a single number, short average call duration, a low answer rate and abandoned calls. Hiya, TNS and First Orion each score independently for AT&T, Verizon and T-Mobile respectively, so a number can be clean on one network and flagged on another.
- How many calls a day before a number gets flagged as spam?
- There is no published threshold, and the engines do not disclose one. In practice, flagging commonly begins somewhere north of 100 to 150 outbound dials a day from a single number when the answer rate is low and calls are short. The volume figure matters far less than the ratio: 200 calls a day with a 20 percent answer rate and three-minute conversations is far safer than 80 calls with a 3 percent answer rate.
- Does STIR/SHAKEN stop Spam Likely labels?
- No. STIR/SHAKEN is a caller ID authentication framework that proves a carrier knows you and that you have the right to use the number. It authenticates identity, not behaviour. An A-attested number placing hundreds of short, unanswered calls a day will still be labelled, because the analytics engines score how the number is used rather than who owns it.
- How do you fix a Spam Likely label?
- Submit a remediation request to each analytics engine separately, because Hiya, TNS and First Orion do not share decisions. Then fix the behaviour that caused it: rotate across more numbers, improve list quality to raise the answer rate, and stop calling numbers that never answer. Simply buying a new number without changing the behaviour reproduces the label within weeks.
- Does number rotation prevent spam flagging?
- It delays it and spreads it rather than preventing it. Splitting 300 daily dials across six numbers puts each at 50 a day, well under the range where flagging usually begins. It does not fix the underlying signals, so a team with a 2 percent answer rate will eventually flag all six numbers. Rotation buys time to fix the list, it is not a substitute for fixing it.