Last updated: September 12, 2026
This page quotes federal regulation and is not legal advice. Check every line against the current CMS text and your own counsel before any of it becomes agency policy.
What a Medicare sales call scorecard measures
A Medicare sales call scorecard grades a recorded marketing, sales, or enrollment call against the rules that govern it. Benefit accuracy and the TPMO disclaimer carry the heaviest weights, and both of those, plus the Scope of Appointment, carry a critical fail, because a missed disclaimer is a regulatory finding rather than a few lost points.
Fourteen criteria across 100 points, and six of them are absolute.
The 100 point Medicare call scorecard
Thirteen criteria carry points during the Annual Enrollment Period. The fourteenth switches on outside it, so the card has two point columns and each one adds to 100. Section totals in parentheses are the AEP column.
| Section | Criterion | AEP points | Points outside AEP | Critical fail |
|---|---|---|---|---|
| Recording and notice (15) | Call recorded in its entirety from greeting to disconnect | 8 | 8 | yes |
| Recording and notice | Recording or monitoring notice delivered where state law requires it | 7 | 7 | no |
| TPMO disclaimer (20) | Required TPMO disclaimer delivered before any discussion of plan benefits | 12 | 12 | yes |
| TPMO disclaimer | Disclaimer delivered in full, from the current approved language, without paraphrase or omission | 8 | 6 | no |
| Scope of Appointment (15) | SOA agreed and documented before plan specific discussion began | 8 | 8 | yes |
| Scope of Appointment | Discussion stayed inside the product types listed on the SOA | 7 | 6 | no |
| Permission to contact (10) | Prior express written consent on file before the outbound call | 5 | 6 | no |
| Permission to contact | Consent scope covers the products actually discussed | 5 | 5 | no |
| Beneficiary initiated contact (0) | Call came from the beneficiary rather than from outreach to someone eligible for the Medicare Advantage open enrollment period | 0 | 10 | yes |
| Benefit accuracy (25) | Benefits described match the current approved plan materials | 12 | 12 | yes |
| Benefit accuracy | No unsupported superlative or unapproved comparison to another plan or carrier | 7 | 6 | yes |
| Benefit accuracy | Correct plan year and effective date stated | 6 | 5 | no |
| Enrollment handling and close (15) | Confirmed the beneficiary's understanding before submitting an enrollment | 8 | 5 | no |
| Enrollment handling and close | Delivered the required next steps and confirmation details | 7 | 4 | no |
| Total | 14 criteria | 100 | 100 | 6 critical fails |
CMS publishes no point scale, so every weight above is our editorial judgment. What a weight buys is a number that comes out the same when two supervisors score the same call. Move them to match how your upline reviews you, keep the total at 100, and how the weights were chosen covers where the critical fails sit.
Two things changed in this file since the version linked from QA for Medicare agencies: the column for calls outside AEP, and the beneficiary initiated row that only scores in it. AEP weights are untouched, so a call scored last season still compares against one scored this season.
Critical fails
Six criteria void the score regardless of what the rest of the call did. Each one is a rule with a consequence attached.
| Critical fail | Why it is absolute |
|---|---|
| No recording exists for a marketing or sales call | The record CMS requires was never created |
| Benefits discussed before the TPMO disclaimer was delivered | The disclaimer timing requirement was missed on the call |
| Plan specific discussion held with no SOA on file | The appointment had no documented scope |
| Stated a benefit that does not appear in the approved plan materials | The beneficiary was given information they may act on |
| Used a prohibited superlative or an unapproved plan comparison | Marketing content restriction breached on a recorded call |
| Outreach to an OEP eligible beneficiary, outside AEP only | The call may sit inside a marketing prohibition rather than a sales rule |
Download the scorecard
All fourteen criteria download as a CSV, ungated, carrying the section, the criterion, both point columns, and an auto_fail column of yes or no.
Keep TPMO in the file name wherever it lands in your compliance folder, because a file called scorecard.csv is unfindable in March.
What CMS requires you to record
CMS requires third party marketing organizations to record marketing and sales calls in their entirety, including the audio portion of calls held over web based technology. That text sits at 42 CFR 422.2274(g)(2)(ii), with the Part D counterpart at 423.2274, checked on 2026-09-12. For most agencies the recording already exists, which is why the first criterion asks whether the file covers the whole call rather than whether a file exists. What the rule leaves open is review. Recording and retention are covered in full in CMS call recording requirements.
What changed for the 2027 plan year
Three changes take effect on October 1, 2026, two weeks before AEP opens.
Retention for marketing and sales calls dropped from ten years to six, held as audio for the first three. Enrollment records stay on the ten year path, so a blanket delete at six is a mistake.
The disclaimer moved off the clock, from delivery inside the first minute to delivery before any discussion of plan benefits, so a script written against a stopwatch fails the moment an agent names a plan at second twenty.
Last, the 48 hour wait between a Scope of Appointment and the appointment is gone, and the SOA itself stayed. Guidance written a year ago still describes that one as current, and wording detail sits in what the TPMO rules require on the call.
Weighting for AEP against the rest of the year
An October enrollment call and a February service call carry different exposure, so the same fourteen criteria get different weights depending on which window produced the recording.
| Section | AEP | Outside AEP |
|---|---|---|
| Recording and notice | 15 | 15 |
| TPMO disclaimer | 20 | 18 |
| Scope of Appointment | 15 | 14 |
| Permission to contact | 10 | 11 |
| Beneficiary initiated contact | 0 | 10 |
| Benefit accuracy | 25 | 23 |
| Enrollment handling and close | 15 | 9 |
| Total | 100 | 100 |
During AEP most of the risk sits inside the call, because most calls end in an enrollment. Outside it the exposure moves upstream to whether the call should have been placed at all, which is where the ten points come from.
We nearly left that row off. Whether a call should have been placed is a campaign decision made weeks earlier by somebody who never appears on the recording, and scoring an agent for it reads as unfair. It stayed because the recording is where that decision surfaces, and a compliance director reading a scored January call wants the line visible rather than filed under marketing.
Weights in the other column are a starting position, since almost nobody publishes anything about scoring mid year Medicare calls.
Inbound and outbound are scored differently
Permission to contact is an outbound question. A beneficiary who dialed you did not need your consent record to do it, so both consent criteria come off an inbound call and the card scores out of 90 rather than 100.
That subtraction is the part agencies get wrong. An inbound call scored 88 out of 90 is a 98, and leaving the denominator at 100 makes the inbound team look worse than the outbound team for no reason anybody can trace.
Everything else holds, and the disclaimer applies the moment benefits come up regardless of who dialed.
Auditing more than two calls a week
Say an agency runs 12 agents through AEP, each taking 35 calls a day across the 54 days of enrollment. That is 22,680 recorded calls. Two per agent per week, the common pattern our own research into this vertical turned up rather than a published benchmark, gets you about 185 reviews across the season.
At 12 minutes per review, listening plus writing it up, that is about 37 hours of supervisor time. The other 22,495 recordings pile up whether anybody opens them or not.
Sampling at that rate works fine for coaching trends. It falls apart on the six criteria above, because one call is the finding. A disclaimer that lands correctly on 99.5% of 22,680 calls still leaves 113 where benefits came up first, and no sample of 185 tells you which ones.
What a scorecard cannot settle
Two rows here are record checks rather than call checks. Whether an SOA was on file and whether consent was captured both live in your CRM, so a reviewer working from audio alone is inferring them. Score those from the record and mark what you could not verify, or the card reports a compliance rate it never measured. The Scope of Appointment checklist splits the audio lines from the file lines, and broader regulatory ground sits in the compliance checklist for the whole operation.
Benefit accuracy has a softer edge than it looks. The criterion asks whether the description matches the approved materials, and it cannot decide whether a loose summary of a dental benefit left the beneficiary with the wrong impression. Where the line falls between a deduction and a critical fail belongs in calibration week one rather than call by call.
Machine scoring holds up on criteria with a phrase or an event behind them and gets shakier on anything needing a read of the conversation, which Aircall and StackAI both note about automated QA. Disclaimer timing is the reliable end, since your approved language goes into a vocabulary library and every flag carries a transcript timestamp. Knowing which voice is the agent matters to several rows, so confirm your dialer exports channelized audio first.
Score last season before this one opens
Run the AEP column against six months of the archive you already hold. Those recordings sit in the dialer under the same rules you are about to work under again, and scoring them produces a disclaimer timing number from calls already on file.
A clean result from that archive is written evidence for a plan partner, and a bad one shows where the script needs fixing. The paid pilot loads the archive and scores it against this card. Agree the pilot scope and processing estimate before files move.
Frequently asked questions
What should a Medicare call scorecard include?
Seven sections: recording and notice, the TPMO disclaimer, the Scope of Appointment, permission to contact, beneficiary initiated contact, benefit accuracy, and enrollment handling. Six of the fourteen criteria carry a critical fail, two of them under benefit accuracy.
When does the TPMO disclaimer have to be read?
Before any discussion of plan benefits, under 42 CFR 422.2267(e)(41)(ii), checked on 2026-09-12 and due for another read in the publication week. Delivery inside the first minute was the previous standard, so a script written last season can pass a stopwatch and still miss the current requirement.
How many Medicare sales calls should be audited?
Agencies commonly audit one to two calls per agent per week, which is what our own client research turned up rather than a published statistic. Against a full AEP season that lands well under one percent, and anything scored as a critical fail needs coverage closer to all of it.
Does CMS require recording every Medicare sales call?
Third party marketing organizations must record marketing and sales calls in their entirety, including the audio portion of web based calls, under 42 CFR 422.2274 and 423.2274. Whether your arrangement makes you a TPMO turns on how you are compensated, so confirm your status with counsel.
How long do Medicare call recordings have to be kept?
Marketing and sales calls run six years under the CY2027 rule effective October 1, 2026, held as audio for the first three years and as audio or a complete and accurate transcript after that. Enrollment records stay on a ten year path, and both obligations sit at 42 CFR 422.2274(g)(2).
Sources
- 42 CFR 422.2274(g)(2), for the recording requirement, the six year retention, and the audio and transcript split. Cornell LII 2026-09-09, rechecked against eCFR 2026-09-12.
- 42 CFR 422.2267(e)(41), for the disclaimer content and the timing at (e)(41)(ii), reached from 422.2274(g)(2)(iv). Cross reference verified 2026-09-12, wording not read.
- 42 CFR 423.2274, 422.2264 and 423.2264, 422.2268(b), and 422.2262, for the Part D counterpart, the Scope of Appointment, the OEP marketing prohibitions, and the marketing content standards. None read during drafting. Unverified, and the beneficiary initiated criterion rests on 422.2268(b).
- CY2027 final rule (CMS-4208-F3), for the retention cut, the disclaimer timing change, the SHIP referral removal, the 48 hour SOA removal, and the October 1, 2026 effective date. Secondary, from agent facing summaries of 2026-09-08. Verify against the Federal Register.
- Ten year enrollment record retention, secondary and unconfirmed, and the October 15 AEP start, worth a check on cms.gov.
- One to two calls per agent per week. Our own client research, 2026-09-08. Estimate, unverified.
- Aircall and StackAI on rule based scoring holding up better than tone scoring. Third party, unverified.
- Point weights, critical fail placement, and the column for calls outside AEP. Editorial judgment.