Robocall mitigation is the layered apparatus that carriers, federal regulators, and third-party analytics providers use to block, label, or downrank unwanted calls. Calls flagged as \"Spam Likely\" see answer rates drop 50-80%. Application developers operate inside this system and must build for it.
\n Originating carrier (signs with STIR/SHAKEN)\n ↓\n Intermediate carriers (pass through, may add tags)\n ↓\n Analytics providers (Hiya, First Orion, TNS, YouMail)\n ↓\n Terminating carrier (applies tags, may block outright)\n ↓\n Recipient's phone (displays tag, allows answer)\nEach layer makes independent decisions. A call rated \"trusted\" by your originating carrier can still be flagged \"Spam Likely\" by the terminating carrier based on independent reputation data.
\nThe Pallone-Thune TRACED Act of 2019 requires:
\nFor voice service providers, RMP registration is mandatory. For application platforms (SignalWire, Twilio), the originating carrier (the platform) handles compliance.
\nMost carriers contract with one of these for reputation scoring:
\n| Provider | Carrier customers | Branded tag |\n|---|---|---|\n| Hiya | Samsung, AT&T (partial), T-Mobile (partial) | \"Spam Risk\" / \"Scam Risk\" |\n| First Orion | T-Mobile (Scam Shield), some MVNOs | \"Scam Likely\" |\n| TNS (Transaction Network Services) | Verizon (Call Filter), AT&T (Call Protect) | \"Telemarketer\" / \"Spam Risk\" |\n| YouMail | Free consumer app, also carrier-licensed | \"Robocaller\" / \"Scam Caller\" |
\nA call gets a \"Spam Likely\" tag when:
\nEmpirically, in the first 30 days after a number is provisioned, reputation forms based on:
\n| Signal | Impact |\n|---|---|\n| Volume of outbound dials | High volume + low answer rate = bad reputation |\n| Average call duration | Very short (< 10 sec) = likely robocall |\n| Diversity of destinations | Calling many unrelated numbers in a small window = bad |\n| Manual user spam reports | Single biggest negative signal |\n| STIR/SHAKEN attestation level | Attestation A protects, C hurts |\n| Caller ID consistency | Changing caller ID mid-campaign = bad |\n| Geographic mismatch | LA number calling Phoenix mostly = neutral, calling random country codes = bad |\n| Time-of-day pattern | Calling 2 AM local time = bad |
\nWhat you can do as someone running SignalWire or other voice platforms:
\nRegister your numbers in the analytics provider databases:
\nBranded caller ID shows your business name + logo on the recipient's phone, bypassing \"Unknown Caller\" and \"Spam Likely\" tags.
\nIf your caller ID is +13105551234, your outbound traffic should originate from a network that legitimately owns that number. STIR/SHAKEN Attestation A requires this.
Most carriers' reputation engines flag:
\n\n\n100 dials per minute from one number
\n
\n\n1000 unique destinations per hour from one number
\n
\n\n5000 short-duration (< 10 sec) outbound calls per day from one number
\n
Pace below these thresholds. Spread campaigns across many numbers with similar profiles.
\nA number's reputation improves when calls get answered and conversations happen. Low answer rates (\"ghost callers\") drive reputation down.
\nTactics:
\nSnowshoeing means using many different numbers to spread call volume evenly so no single number trips reputation thresholds. Carriers detect this via:
\nWhen detected, the whole number range gets flagged together. Don't snowshoe. Build reputation per number organically.
\nIf your legitimate business number gets tagged \"Spam Likely\":
\n| Carrier | Dispute path |\n|---|---|\n| AT&T (TNS) | https://www.callprotect.tns.com/ |\n| Verizon (TNS) | Same TNS portal |\n| T-Mobile (First Orion) | https://callscoreinfo.com/ |\n| Hiya | https://www.hiya.com/business |\n| YouMail | https://directory.youmail.com/ |
\nSubmit business documentation. Resolution: 5-30 business days. Re-tagging is possible if behavior continues to look spammy.
\nRotate numbers every 30-90 days for outbound campaigns. Reputation deteriorates gradually with sustained outbound; fresh numbers start clean. Don't rotate so fast that you can't build branded caller ID.
\nCarrier analytics inspect the audio of recorded sample calls. \"Free trial,\" \"you've been selected,\" \"limited time offer\" trigger ML models that flag the number.
\nSignalWire doesn't expose analytics-provider scores directly. Check externally via:
\nIf a number is consistently labeled \"Spam Likely\" across multiple lookups, retire it.
\nSignalWire automatically signs outbound calls with Attestation A when the number is owned by the project. For BYOC scenarios, attestation may drop. See STIR/SHAKEN.
\nSignalWire integrates with Hiya, First Orion, and TNS for branded caller ID. Setup is via SignalWire support — submit your brand info and they provision via the analytics provider.
\n| Symptom | Cause | Fix |\n|---|---|---|\n| New number gets \"Spam Likely\" within days | Snowshoeing detected, or content flagged | Pace dialing, vary content, register with analytics providers |\n| Branded caller ID not showing despite registration | Analytics provider hasn't propagated yet (24-72 hours) | Wait, re-confirm in lookups |\n| T-Mobile blocks outright but other carriers don't | T-Mobile Scam Shield is most aggressive | Submit dispute, reduce volume, vary content |\n| Answer rate dropped 40% overnight | Number got flagged | Lookup reputation, dispute if false positive, rotate to clean number if not |\n| Calls \"completed\" but actually ringing-then-blocked | Carrier blocked at terminating switch | CDR will show short duration; cross-check with carrier |
\nRetire when:
\nRetire by stopping outbound and either reassigning to inbound-only use or releasing back to the carrier pool (after 30 days, the next subscriber inherits a fresh reputation).
\nToll-free numbers (after TFV) and short codes have much stronger reputation footing because:
\nFor high-volume outbound that survives carrier filtering, toll-free or short code is often the right answer. See toll-free vs short code.
\n