STIR/SHAKEN and Caller ID: Why Your Calls Get Flagged

You call a customer who asked to hear from you, and their phone shows "Spam Likely." They don't pick up. For outbound teams, caller ID reputation can quietly cut contact rates, and the causes aren't always obvious.

Two separate systems are at work: STIR/SHAKEN, which verifies who is calling, and carrier analytics, which decide whether to label the call. This article explains both and what you can do.

STIR/SHAKEN in brief

STIR/SHAKEN is a framework that lets phone carriers digitally sign calls to vouch for the caller ID. The originating carrier checks whether its customer is entitled to use the number being displayed, signs the call with an attestation level, and the receiving carrier verifies the signature.

The FCC required most voice providers to implement it in the IP parts of their networks by June 30, 2021, under authority from the TRACED Act, with extensions for some smaller providers.

Attestation levels

• A (full): the carrier knows the customer and confirms they're authorized to use this number.
• B (partial): the carrier knows the customer but can't confirm their right to the specific number.
• C (gateway): the carrier is passing along a call it received from elsewhere and can't vouch for the source.

Calls with A attestation are more trusted. If you use a cloud phone system or dialer, ask your provider what attestation your outbound calls receive. Calls using numbers that aren't registered to you with that provider often get B or C.

Where "Spam Likely" comes from

STIR/SHAKEN verifies identity. It doesn't decide whether a call is wanted. The labels people see ("Spam Likely," "Scam Likely," "Telemarketer") mostly come from call analytics run by or for the major carriers. Those systems look at signals such as:

• How many calls a number makes per day and per hour
• How many of those calls are short or unanswered
• How often people decline the call or block the number
• Complaints and reports from users
• Whether the number recently started making lots of calls
• Attestation level

A legitimate business can trigger these signals easily: a new number making hundreds of calls a day, many unanswered, will look like a robocaller to an algorithm.

How to keep your numbers clean

Register your numbers

The major analytics providers accept registrations from businesses, and there are services that submit your numbers to several of them at once, such as the Free Caller Registry. Registering doesn't guarantee clean labels, but it gives analytics systems information about who you are.

Get A-level attestation

Use numbers you own through your calling provider, and make sure the provider signs your calls with full attestation.

Watch volume per number

High call volume from a single number is one of the strongest spam signals. Spread outbound calling across enough numbers that each one's daily volume looks like a normal business line. Don't rotate numbers constantly to dodge labels, though. That pattern is itself a warning sign, and spoofing numbers you don't own is illegal when done with intent to defraud or cause harm under the Truth in Caller ID Act.

Call people who expect you

Calls to people who asked to hear from you get answered and get fewer complaints. That improves your reputation with every call. Fast follow-up on fresh leads (see speed to lead) also helps, because recent inquirers recognize the business.

Honor opt-outs immediately

People who've asked you to stop calling and get called again are the most likely to report you. Make sure opt-outs reach every system that dials. Our TCPA article covers the rules, and call disposition codes covers getting DNC requests into your data.

Use a consistent, callable number

Show a number that, if called back, reaches your business. A callback that gets "this number is not in service" looks like a scam.

Consider branded calling

Some carriers and providers offer branded caller ID that displays your business name (and sometimes a logo and reason for calling) on supported phones. It can raise answer rates for legitimate businesses.

Monitor your numbers

Check regularly how your numbers are labeled. Several services let you test numbers across carriers. If a number gets flagged, find out why before simply swapping it.

What this means for AI outbound calling

AI voice agents can make a lot of calls quickly, which makes reputation management more important, not less:

• Pace campaigns so no number exceeds a reasonable daily volume
• Use the same registered, A-attested numbers your team uses for other calls
• Have the agent identify the business at the start of every call
• Make opt-out handling bulletproof
• Watch answer rates per number. A sudden drop often means a new label

If your number is flagged

1. Confirm the label by testing on major carriers.
2. Check your recent calling patterns for volume spikes or high short-call rates.
3. Submit a remediation request through the analytics providers' or carriers' business portals.
4. Fix the cause (volume, list quality, opt-out handling) before resuming at full speed.

How a call gets labeled, step by step

To make the moving parts concrete, follow one outbound call:

1. Your dialer places a call from a number you lease from your provider.
2. Your provider checks that you're its customer and that the number is assigned to you, then signs the call with A-level attestation.
3. The call passes through one or more carriers to the recipient's mobile carrier.
4. The recipient's carrier verifies the signature. It's valid, so the caller ID is authenticated.
5. Separately, the carrier's analytics engine looks up the number's reputation. It sees that this number made 600 calls yesterday, most under 10 seconds, and several recipients blocked it.
6. The analytics engine decides the call looks like unwanted bulk calling and adds a "Spam Likely" label, even though the caller ID was authenticated.
7. The recipient sees the label and declines.

Step 2 went well. Step 5 is where the problem came from. Fixing labels is mostly about the behavior analytics engines observe.

Habits that protect reputation over time

Ramp up new numbers gradually. A brand-new number that suddenly makes hundreds of calls looks suspicious. Start with modest volume and increase over a couple of weeks.

Keep short-call rates down. Calls that hang up in seconds, from dropped connections or aggressive dialer settings, hurt reputation. Tune your dialer so calls ring long enough for people to answer.

Avoid unnecessary abandoned calls. Predictive dialers that connect more calls than agents can handle create silent calls, which are frustrating and harmful to reputation. AI agents help here, because there's always something available to speak when the call connects.

Match numbers to the audience. Calling from a number with an area code that makes sense for your business and audience can help answer rates, but don't use local numbers to mislead people about who's calling.

Watch complaints. If your answer rate drops suddenly, check for new labels and review recent campaigns for list or script problems.

Inbound matters too

Reputation isn't just about outbound. Make sure the numbers you call from can receive calls and reach a helpful destination: a person, an AI agent, or at minimum a clear voicemail greeting naming your business. When people call back an unfamiliar number to find out who it was, a clear answer builds trust and reduces reports.

A monthly checklist

• Test labels on your active numbers across major carriers
• Review answer rate by number and by campaign
• Confirm attestation level with your provider
• Check that registrations with analytics providers are current
• Retire or remediate numbers with persistent labels
• Review opt-out and DNC handling for errors

Frequently asked questions

Does STIR/SHAKEN stop spam labels?

No. It verifies the caller ID. Labels are decided separately by analytics systems, though good attestation helps.

How many calls per day per number is safe?

There's no published threshold, and it varies by carrier and by how calls are received. Start conservatively, watch answer rates and labels, and spread volume across numbers.

Will calling people who opted in prevent labels?

It helps a great deal, because those calls are answered and rarely reported, but high volume from one number can still trigger analytics.

Written by the Voxvencer editorial team. We build and run AI voice agents for call centers and small businesses, and we write about what we see on real phone lines. Questions or corrections: info@voxvencer.com.