Why Medicare Leads Aren't Converting in 2026: Call Filtering, Reputation Scoring, and the Collapse of Phone Trust

Introduction: Something Changed — and It Is Not What Most Think

There was a time when outbound Medicare sales followed a predictable and repeatable pattern. Call 20 people, have real conversations with several of them, and leave two or three appointments on the calendar. The math worked. The effort translated directly into results.

That same effort today often produces a single appointment — if that — from 100 calls. This is not a minor statistical fluctuation. It is a 5–10× collapse in efficiency, and it has been building quietly over the past several years, accelerating noticeably since 2023.

The conventional explanations — bad leads, weak scripts, poor follow-up — do not hold up to examination. Agents who have not changed their scripts, their work ethic, or their lead sources are seeing the same pattern. Something deeper shifted.

The thesis of this study is simple and documented: the phone system itself no longer guarantees call delivery. Modern telecom networks evaluate calls before they reach the consumer, and that evaluation can result in calls being blocked, labeled, suppressed, or silently diverted — regardless of whether the call is legitimate or whether the prospect actually requested contact.

This is not a complaint. It is a documented shift that any agent who wants to compete in the current environment must understand. KundPro is an insurance agency CRM platform built for Medicare agents — and this study comes directly from field experience managing a book of business in this environment.

The Hidden Shift: Calls Are Now Filtered Before They Ring

For most of the history of telephone communication, when a call was placed, it was delivered. The network's job was transport — neutral, reliable, end-to-end. Calls might go unanswered, but they arrived.

That model no longer accurately describes how calls are handled.

Under guidance and explicit rulemaking from the Federal Communications Commission, telephone carriers now operate analytics-based filtering systems as a standard feature of call delivery. The FCC has explicitly permitted carriers to block calls using analytics[1] and has provided a "safe harbor" — legal protection — for providers that block calls based on reasonable network analytics.[3]

The three largest carriers in the United States each operate and publicly document these systems:

• Verizon Call Filter detects spam automatically and forwards high-risk spam calls directly to voicemail.[4]
• AT&T ActiveArmor provides automatic fraud call blocking and routes spam categories based on user and network settings, with fraud calls set to automatic block by default.[5]
• T-Mobile Scam Shield is built into the network at the infrastructure level and labels calls with categories including "Scam Likely," "Potential Spam," and "Telemarketing."[6]

A call being placed does not guarantee it will ring. It may be blocked, labeled, or silently forwarded to voicemail — without the agent knowing, and without the prospect ever having a real chance to answer.

The Reputation Layer: How Calls Are Scored Before Delivery

Behind carrier-level filtering sits an additional infrastructure layer managed by analytics providers whose products determine call treatment in real time. Industry platforms such as Numeracle and First Orion describe systems that evaluate calls across multiple behavioral dimensions simultaneously.[7][8]

These systems do not ask "Is this call legitimate?" — because that question cannot be answered behaviorally. Instead, they ask: "Does this call behave like something we trust?"

The factors evaluated include call volume and frequency, dialing speed and burst patterns, answer rates, call duration, and the consistency of caller identification across systems. Numeracle's published research specifically documents that even calls from verified, compliant business numbers can be mislabeled due to behavioral signals — a problem that industry analysis estimates affects more than 25% of legitimate business calls.[7]

The Illusion of Contact

This creates a structural gap between what outbound activity metrics report and what is actually happening at the point of attempted contact.

From the agent's perspective and from every system log: call placed, ring detected, voicemail left — all confirmed. From the prospect's perspective: phone never rang, message never reached a monitored voicemail, no meaningful contact opportunity occurred.

This is what this study terms the Illusion of Contact: a condition in which outbound activity metrics appear normal while actual human contact approaches zero. As call treatment increasingly occurs at the network and device layer before the consumer's phone rings, activity volume becomes a progressively weaker proxy for actual outreach effectiveness.

The operational implication is significant. An agent who believes they are failing at persuasion may actually be failing at delivery — two problems that require completely different responses.

The Ghost Voicemail Problem: Leaving Messages That Are Never Received

Beyond calls that are blocked before ringing, there is a second and lesser-documented failure mode that field observation confirms is real: voicemails that are recorded by the agent but never delivered to the recipient. This phenomenon is referred to in the industry as Silent Voicemail Filtering or "ghost voicemail."

The mechanism works as follows. Carriers and third-party filtering applications have moved beyond blocking calls outright. They now analyze and intercept audio data after a voicemail is recorded. If a number has been flagged as marketing or spam-likely, the carrier may allow the voicemail recording to complete — giving the caller every indication that the message was successfully left — while silently discarding it before it ever triggers a notification on the recipient's phone.

There are three primary mechanisms through which this occurs:

• Carrier-level junk voicemail filtering: Carriers including Verizon and AT&T now use AI to analyze the content and metadata of recorded messages. Numbers flagged as marketing or spam-likely may have their voicemails quarantined rather than delivered — the recipient's phone never receives a missed-call notification and the voicemail never appears in their inbox.
• A2P (Application-to-Person) traffic classification: If an agent is dialing from a VoIP system or business dialing platform, carriers may classify the voice traffic like SMS messaging traffic and apply A2P filtering rules. High-volume patterns or repetitive scripts can cause the final delivery of the recorded message to be blocked even after the call path appeared to complete normally.
• Third-party app interception (Answer Bots): Applications such as RoboKiller and Hiya can act as a man-in-the-middle. When a call is deemed low-trust, these apps may answer automatically, play a generic voicemail greeting to the caller, and route the recording to their internal spam dashboard — never delivering it to the subscriber's actual phone inbox. Some apps use "Answer Bots" that play audio to suspected marketing callers specifically to waste time and prevent the call from reaching the actual recipient.

This Is Not a Sales Problem Alone: When Doctors Get Flagged as Spam

The systemic nature of call filtering becomes clearest when examining its effects outside of commercial outreach. Healthcare providers — calling their own patients for appointment reminders, test results, and prescription follow-ups — are routinely caught by the same filtering systems that target telemarketers.

This is not anecdotal. The FCC's Hospital Robocall Protection Group specifically addresses the impact of robocall filtering on healthcare communication infrastructure.[13] Industry research indicates that a substantial share of healthcare-related calls never reach the intended patient due to spam labeling or suppression.

From field experience: calls from a physician's office appear as "Spam Likely" on a patient's phone. These are expected, legitimate calls from a known provider to an established patient — and the filtering system assigns them the same label as a robocall scam. The system does not evaluate intent. It evaluates patterns. And the call-volume patterns of a busy medical practice superficially resemble the patterns of an outbound call center.

The Overreach Problem: Consent Does Not Guarantee Delivery

The most acute frustration in outbound practice is this: a prospect fills out a form, explicitly requesting to be contacted about Medicare options. The agent calls within minutes. The call is blocked, labeled, or silently diverted — because the calling number's behavioral pattern has triggered a carrier analytics threshold.

Written consent, documented lead sources, and immediate follow-up provide no immunity from filtering. The system does not have access to the prospect's form submission. It sees call behavior patterns only.

The FCC has acknowledged the risk that call blocking can "inadvertently intercept wanted calls" and has built redress mechanisms into its blocking policy framework — but those mechanisms operate after the fact.[2] In the moment a call is placed and filtered, consent is invisible to the network.

This is the overreach problem: a system designed with legitimate public interest goals — reducing harassment and fraud — has been calibrated to a sensitivity that also suppresses legitimate, permission-based commercial and medical communication. No individual carrier or regulator intended this outcome, but it is the documented result of deploying pattern-recognition systems at scale.

Why Predictive Dialers Are Losing Ground

Predictive dialing systems were engineered for an environment in which call delivery was largely neutral. Their design goal was to maximize agent utilization by minimizing idle time — dialing enough numbers in parallel to ensure a live connection was always available when an agent finished the previous conversation. In a neutral delivery environment, this was effective.

In the current environment, dialing behavior is itself part of the evaluation. Predictive dialing creates behavioral signatures structurally difficult to distinguish from automated spam:

• High call velocity — often one call every two to five seconds
• Predictable, repetitive sequences with consistent timing
• Elevated abandonment rates when agents are unavailable at connection
• Short call durations when connections are abandoned or immediately rejected

Carrier analytics systems — and the machine learning models documented in academic research on call filtering[15] — are calibrated specifically to identify these patterns. The outcome is counterintuitive but documented: dialing faster, at higher velocity, can reduce the number of people actually reached, because the behavioral pattern itself triggers suppression.

An agent or operation can simultaneously increase raw call volume and decrease effective contact rate — because the dialing behavior is consuming the reputation capital of every number in the pool.

Observed Performance: Old Model vs. Current Environment

The Real Shift: From Speed-to-Lead to Speed-to-Trust

The single most practically important reframe in this study is the following: the competitive advantage in outbound sales has changed.

For years, the advantage belonged to the agent who called first. Lead response speed was the primary determinant of contact rate, and contact rate drove outcomes. Being first was almost always better than being second.

That advantage has shifted. In a filtered phone environment, being first means nothing if the call is suppressed before the prospect can answer. Speed now matters — but only when paired with trust signals strong enough to survive carrier and device-level filtering.

In practice, improving trust signals often means making your identity clear before the call ever happens — whether through pre-call messaging, consistent caller ID, or tools that allow prospects to recognize who is reaching out before the phone rings.

What Actually Works: Evidence-Based Adaptation Strategies

The following recommendations are grounded in the documented behavior of carrier filtering systems, industry research on call reputation, and field observation. There is no single intervention that restores 2015-era contact rates — the environment has fundamentally changed. But there are adaptations that consistently produce measurable improvement.

1. Pre-Call Context — The Highest-Leverage Single Action

The most consistent improvement in answer rates comes from establishing context before the call is placed. An SMS sent within one to three minutes of a new lead arriving, identifying who is calling and why, converts the subsequent call from "unknown number" to "expected contact." This intervention produces a meaningful improvement in answer rates on fresh leads when the call follows within minutes of the message.

2. Register with the Three Major Analytics Providers

Carriers do not maintain their own spam lists. Verizon, AT&T, and T-Mobile each rely on third-party analytics companies to flag and score calls. Understanding which provider serves which carrier is actionable knowledge:

• Hiya — the primary analytics partner for AT&T
• First Orion — the primary analytics partner for T-Mobile
• Transaction Network Services (TNS) — the primary analytics partner for Verizon

The Free Caller Registry (freecallerregistry.com) allows businesses to register their number identity with all three of these providers simultaneously. This is the single most direct step an agent can take to verify business identity with the networks that determine how calls are scored. Registration does not eliminate the risk of spam labeling, but it provides the baseline identity signal that all three systems use when evaluating call treatment.[8]

Number reputation must then be maintained over time. Phone numbers accumulate scoring history based on behavioral patterns. Numbers with high complaint rates, high volumes of short or unanswered calls, or aggressive dialing patterns receive progressively worse treatment. Retiring damaged numbers, limiting calls per number per day, and rotating pools are all documented as effective.[10][11]

One often-overlooked requirement: number warm-up. New numbers have zero reputation history. Beginning with high call volume on a fresh number signals the same behavioral pattern as a robocall operation — causing the number to be flagged before it ever builds positive history. Industry practice is to start new numbers at low volume (10–15 calls/day) and increase gradually over several weeks. Sudden high-volume bursts from a new number are one of the fastest ways to trigger spam labeling.

3. Achieve Level A STIR/SHAKEN Attestation — and Understand the Gap

STIR/SHAKEN is the federal framework that authenticates whether a call is genuinely originating from the number it displays. Calls are assigned one of three attestation levels:[2]

• Level A (Full Attestation): The originating carrier fully verifies the caller's identity and their right to use the number. Calls with Level A are treated with the highest trust by downstream filtering systems.
• Level B (Partial Attestation): The carrier verifies the call's origin but cannot confirm the caller's right to use that specific number.
• Level C (Gateway Attestation): The call arrived from outside the domestic telephone network and identity cannot be verified. Often treated as high-risk.

The Attestation Gap is critical for agents using third-party dialers or separate SIP trunk providers: routing calls through these systems can automatically downgrade attestation from Level A to Level B or C — even if the agent has done everything correctly on their end. A Level B or C attestation is a significant trigger for "Spam Likely" labels and silent filtering by all three major carriers. Agents experiencing unexplained call suppression should verify their attestation level with their phone or VoIP provider, and work with a STIR/SHAKEN-authorized provider to ensure calls are digitally signed with Level A credentials.

4. Human-Pattern Dialing — The Modern Optimal Configuration

Based on call center research, industry best practices, and the behavioral logic of reputation scoring systems, the following configuration represents the current best practice for outbound Medicare calling. Note the specific thresholds: calls under 15 seconds are a primary negative signal to spam-detection systems — they register as hang-ups or robocall patterns. Calls reaching 60+ seconds are a strong positive signal indicating real human engagement.

5. Caller Identity Consistency + Rich Call Data

CNAM (Caller ID Name) should be consistent, accurate, and registered across all relevant databases. Mismatches between the name associated with a number in CNAM records, the business's web presence, and any registered calling certification create identity fragility that filtering systems detect.

Beyond CNAM, agents should be aware of Rich Call Data (RCD) and Branded Call Display (BCD) — a technology that allows your business name and logo to appear on the recipient's screen even if they have not saved your number. This directly bypasses many "unknown caller" filters by making the call visually identifiable before the recipient decides whether to answer. Implementation requires working with a carrier or VoIP provider that supports RCD delivery.

One practical way to strengthen identity consistency before any call is placed is to create a contact card for your outbound calls so prospects immediately recognize who is reaching out. In a filtered phone environment, unfamiliar numbers are often ignored or flagged, but recognizable identity changes how both the device and the person interpret the call. When a prospect has already seen your name, your message, or your contact details, the call shifts from an unknown interruption to an expected interaction — improving both answer rates and trust before the phone ever rings.

6. Multi-Channel Sequencing

The phone is no longer a standalone channel. An effective outreach sequence for a warm Medicare lead in 2026 looks like: immediate SMS providing context → call within 2–5 minutes → if unanswered, brief personalized voicemail followed by email → follow-up the next day with a new angle. Each channel reinforces the legitimacy of the others and increases the probability that the prospect sees the outreach as expected rather than intrusive.[14]

7. Monitor Reputation and Track Your Own Evidence

Agents who log call-level outcomes — not just total dials, but immediate-to-voicemail rates, callback rates after SMS, and answer rates by number and time-of-day — develop proprietary data that identifies problems and opportunities invisible in aggregate metrics. Tools such as Numeracle and Caller ID Reputation allow agents to see how their calls actually appear on real devices across different carriers before they dial. Checking these tools on a regular basis — not only after problems emerge — is the operational equivalent of monitoring a business credit score.[7]

Conclusion: The Communication Layer Has Changed

The performance decline in Medicare outbound is real, measurable, and not primarily attributable to agent skill, lead quality, or script quality. It reflects a documented change in how telephone calls are evaluated, treated, and delivered by modern telecom infrastructure.

Carriers are permitted and encouraged by federal policy to filter calls analytically. They have built and deployed systems that do exactly that, at scale, automatically. Those systems evaluate behavior patterns — not intent — which means legitimate, permission-based calls are regularly affected. This is an unintended consequence of deploying aggressive pattern-recognition filters against a problem — robocalls — that genuinely warranted intervention.

The path forward is adaptation, not more volume. Trust signals, identity consistency, human-pattern dialing, and multi-channel sequencing are not workarounds — they are the appropriate response to a changed communication environment. The agents who understand this first will regain the contact rates that others will continue to struggle to explain.

How to Cite This Study

Scallion, David. Why Medicare Leads Aren't Converting in 2026: Call Filtering, Reputation Scoring, and the Collapse of Phone Trust. KundPro / Lehigh Partners Senior Benefits. April 2026. https://kundpro.com/learn/why-medicare-leads-arent-converting-2026/