STIR/SHAKEN Attestation: Why an A Is Not Enough

Your carrier says your calls go out signed at A. Your reps still watch a spam label land on a third of the screens. Both of those things are true at once, because STIR/SHAKEN attestation and spam labeling are two different systems that happen to ride on the same call. Getting an A is worth doing. Expecting it to fix your answer rate is where teams waste a quarter.

What does STIR/SHAKEN attestation actually claim?

STIR/SHAKEN is the framework US carriers use to digitally sign outbound calls so the number in your caller ID cannot be quietly forged while the call is in transit. When your originating provider signs a call, it stamps one letter on it. That letter is the entire claim the framework makes about you.

The FCC's description of the three levels is narrow on purpose. A provider may assert A when it put the call onto the IP network, has a direct authenticated relationship with the customer, and has confirmed that the customer is associated with the number being presented. B covers the first two conditions but not the third. C is what a gateway asserts when a call arrived from somewhere else and it knows neither the caller nor the number.

• A: the provider knows you and knows the number is yours to use.
• B: the provider knows you but cannot confirm the number.
• C: the provider knows neither. Usually a gateway passing traffic along.

Now read that again and notice what is missing. Nothing in the definitions touches whether the person wanted your call, whether you have consent, or whether your list is any good. Attestation is a statement about your carrier's paperwork on you. It was never a quality score, and it was never designed to be one.

Does A-level attestation stop a spam label?

No. And the evidence against it is stronger than most sales blogs let on.

Search this topic and you will find A-level sold as the cure, with a promised lift of 10 to 20 percent in answer rate and warnings that weak attestation drags connect rates down to 2 to 5 percent. Not one of those pages publishes a sample size, a date, or a definition of a connect. Read them as marketing copy and move on.

The FCC's Wireline Competition Bureau submitted its second triennial assessment of the framework to Congress on 19 December 2025, and the record in that report to Congress points the other way. USTelecom told the Commission that TNS honeypot data shows scam calls carrying A-level attestation more than half the time, and that 43 percent of spam traffic in TNS production data is signed at A. CTIA described the same thing from the carrier side: unwanted calls arriving with A-level signatures from parties nobody can identify.

The Commission has said it in its own voice too. In the call branding rulemaking it opened on 28 October 2025, the FCC wrote that STIR/SHAKEN information does not give consumers real information about who is calling, and that an A-level indicator on its own does not tell someone whether a call is worth answering.

So an A is the floor, not the edge. The people you are being confused with already have one.

How often does the signature even arrive?

There is a second problem and it is physical. The framework only works across IP networks. The moment a call crosses a segment still running on older TDM equipment, the signed token gets stripped and the terminating carrier sees an unsigned call.

The Bureau named non-IP networks the single largest obstacle to the framework being used everywhere. In the same proceeding, the VON Coalition cited an industry estimate that more than 60 percent of calls signed by an originating provider reach the terminating provider with no attestation details attached. TransNexus added that the share arriving with authentication intact appears to be falling, possibly because some providers route deliberately over non-IP segments.

Put that beside the 84 percent signing figure T-Mobile reported for traffic between major US providers and the real picture shows up. Most calls get signed. A large share of those signatures do not survive the trip.

What actually decides whether your call gets labeled

Labeling is done by analytics engines that sit in front of the receiving handset. TNS, Hiya and First Orion are the names you will run into. They are sold to the terminating carriers, and attestation is an input to them rather than the thing itself.

INCOMPAS and the Competitive Carriers Association put it plainly in the same FCC record: the terminating provider or its analytics partner uses the signing information as one factor in deciding whether to block or label a call. Numeracle went further, telling the Commission that the presence or absence of a signature is often irrelevant to the treatment decision, and that hospitals, schools and financial firms still get wrongly labeled despite following every published best practice.

The other inputs are the ones sitting on your own floor. How many people report your number. What share of your connected calls die inside six seconds. How many distinct numbers one line dials in an hour. How old the number is and whether it has any history behind it. We went through that scoring in detail in why calls show as spam likely, and through the reason swapping numbers has stopped helping in phone number rotation.

What an outbound team should do about attestation

Attestation is not something you buy off a shelf. Your originating provider assigns it per call, based on what it can prove about you. So the work is mostly about being easy to vouch for.

• Ask your provider in writing what level your traffic is signed at, and whether it signs with its own certificate. Since 18 September 2025, a provider with an implementation obligation has to make its own attestation decisions even when a partner runs the signing workflow.
• Only present numbers you actually control. Displaying a number your carrier cannot tie back to you is the quickest way to fall from A to B, and it is the pattern the whole framework was built to catch.
• Keep your numbers. A number with months of steady, moderate volume behind it is worth more than a fresh one. Thin history now reads as a risk signal all by itself.
• Fix the behavior the analytics engines score first. Branded caller ID will put your company name on the screen, and it still will not outrun a number people keep reporting.
• Stay inside calling hours and honor opt-outs fast. Complaints are the input with the most weight, and both of those reduce them. Your internal do not call list is doing more for your answer rate than your certificate is.

On the platform side, the useful thing is having the dial pacing and the compliance rules in the same place as the call history, so a number's reputation is visible next to the behavior creating it. SellifyGPT runs its predictive dialer at one to three lines per rep, holds calling hours to 9am through 9pm in the contact's own time zone with daylight saving handled and a per-campaign override, and defers a contact to the next compliant moment instead of skipping them. None of that changes your attestation letter. It changes the inputs that the labeling engines are reading. You can start a 14-day free trial and watch it against your own list.

One honest caveat on the compliance pieces. Calling hours, consent and opt-out handling are areas where the rules shift and the details matter. Treat everything above as background reading and take the specifics to your own counsel rather than to a blog post.

How do you measure any of this on your own list?

Attestation does not move day to day, so measure the thing downstream of it instead. Five cuts, each with a real minimum sample behind it.

1. Answer rate per outbound number, weekly. At least 100 dials behind a number before you read it as anything.
2. Share of connected calls under six seconds. A rising short-call ratio usually means people are seeing a label and hanging up on it.
3. Answer rate on a control number you dial slowly and never rotate. It tells you how much of a drop is the number and how much is the list.
4. Report rate per thousand dials, pulled from whatever your carrier or analytics vendor exposes to you.
5. A monthly spot check of your own numbers on the free lookup tools the major analytics providers publish, so you find a label before your reps do.

Run all five for a month before you change anything. One bad afternoon on a small list tells you nothing, and most of the advice in this category is built on exactly that.

The short version

Get your calls signed at A, because being unsigned or C-attested is genuinely worse and because it costs you nothing but a conversation with your provider. Then stop thinking about it. The FCC's own record shows scammers holding A-level signatures at scale and legitimate callers getting labeled anyway, and it shows the majority of signatures never reaching the far end intact. Your answer rate lives in your list quality, your complaint rate and your calling behavior. Those are the levers that were always yours.