CMS Just Let You Call Your Plan 'The Best' Again

CMS dropped the proof requirement for superlatives

If you write Medicare Advantage or Part D marketing copy, you've spent years working around a specific friction point: a sentence like "the best plan in the area" or "the most complete Medicare coverage" couldn't stand alone. Under the prior version of 42 CFR 422.2262, any material using a superlative or comparative claim had to carry its supporting documentation or data source referenced directly on the same material. That's not a light lift for a landing page, a postcard, or a thirty-second radio spot, so most compliance-conscious agencies simply avoided the word "best" altogether and wrote around it.

If you write across Medicare Advantage, Part D, and anything else in an agency's book, this is the kind of rule that's easy to half-hear about during AEP prep season and then apply too broadly, or not catch at all until a compliance reviewer flags it. It deserves a plain statement of exactly what changed, exactly when, and exactly what's still true underneath it, because the gap between those three things is where the actual risk sits.

That specific friction is gone. CMS finalized its Contract Year 2027 Medicare Advantage and Part D policy rule this spring, and the agency's own partner tip sheet, "Changes to the Medicare Advantage Organization and Medicare Drug Plan Sponsors Marketing Guidelines," lists six marketing and communications guideline changes that apply to all CY2027 materials starting October 1, 2026.¹ Under the heading "Advertising Language Restrictions," CMS states it plainly: "Plans no longer need to include supporting documents on materials that use superlatives (e.g., 'best,' 'most'). In many cases, plans must still provide supporting data if CMS requests it."¹

This is one of six changes CMS grouped into the same tip sheet, and it's worth seeing where it sits next to the others, because three of the six are already their own Playbook guide and one more sits in a different part of the rule entirely.

Why the documentation requirement existed

It helps to understand what the old rule was actually trying to stop before deciding what the new one changes. A superlative claim is, by definition, a comparison: calling a plan "the best" implicitly ranks it against every other plan a reader could choose instead. CMS oversees a program where a confusing or misleading comparison has real consequences, a beneficiary picks the wrong plan based on an unverifiable claim and finds out at the pharmacy counter or the specialist's office that the plan doesn't actually cover what the ad implied it did. Requiring the supporting data to sit right there on the page was a blunt but effective way to make a plan sponsor do the substantiation work before publishing, not after a complaint.

That requirement was also, by CMS's own account, somewhat redundant with a rule CMS was already enforcing elsewhere: the general prohibition on misleading or confusing materials. If a claim can't be substantiated, it's arguably misleading regardless of whether the documentation requirement exists as its own separate rule. CMS's broader CY2027 final rule leans on that redundancy argument repeatedly. Its own fact sheet groups several of the deregulatory provisions, the superlative change among the general category, under the heading "Reducing Regulatory Burden and Costs in Accordance with Executive Order 14192," describing the agency as "removing restrictions on the time and manner by which beneficiaries can have conversations with licensed agents and brokers."⁵ Read plainly, CMS's own framing is that it's trimming procedural steps it considers duplicative of other safeguards already in force, not relaxing the underlying safeguards themselves.

The Notice of Availability change in the same tip sheet works the same way. CMS dropped its own duplicate notice requirement because HHS's Office of Civil Rights already requires an equivalent one.¹ That's the pattern across most of the six changes: CMS removing a specific procedural requirement it judges redundant, while the substantive protection the procedure existed to enforce, no misleading claims, accessible language assistance, a documented compliance trail, stays intact somewhere else in the regulatory structure. The superlative change fits that pattern exactly. The documentation-on-the-page requirement is gone. The documentation-on-request power and the ban on misleading claims are the "somewhere else" it still lives.

How the six changes split: what you publish vs. how you sell

Group the six changes by what they actually touch and a pattern falls out that's useful for deciding where your own attention belongs. Four of the six affect what ends up published, printed, or displayed: the Notice of Availability requirement, the advertising language change this guide is about, the TPMO disclaimer wording and timing, and the mid-year supplemental benefits notice. Only one, the outreach timing change covering Scope of Appointment and educational events, touches the live sales process itself. The sixth, call recording retention, is purely a recordkeeping change with no content implication at all.

That's worth knowing because it reframes how much this specific rule update actually affects your website. It's not one small item in a list of mostly-sales-process changes. Four of the six, two thirds of this year's update, land on the same surface your website already lives on. If you've been tracking this rule change as a sales-floor compliance update and nothing more, you've been under-covering where it actually applies.

What CMS didn't touch

Read the current regulatory text next to the old text and the shape of the change becomes obvious. The current version of 42 CFR 422.2262, as mirrored on Cornell Law School's Legal Information Institute and amended by the Federal Register entry tied to this rule, still opens its list of prohibited activities with "activities that could mislead or confuse beneficiaries, or misrepresent the organization."² That's paragraph (a)(1)(i), and it carries no exception for superlatives. Everything the old superlative provision was backstopping is still backstopped. What's gone is the narrower, specific requirement that documentation live on the same page as the claim.

That last row matters if your agency writes across more than one line of business. This change lives in 42 CFR Part 422 and Part 423, the Medicare Advantage and Part D regulations. It has no bearing on how a state insurance department treats a comparative claim in an ACA marketplace ad, a life insurance brochure, or a final expense mailer. Most states still enforce some version of the NAIC's unfair trade practices framework against a misleading comparison in those lines, with no federal loosening attached to it. A multi-line agency that reads this rule as blanket permission to write "the best" everywhere is solving a Medicare Advantage problem with a rule that doesn't reach the rest of its book.

The trap the loosened rule sets

Here's where this gets worth a full guide instead of a one-paragraph compliance update. A rule that removes a specific procedural requirement without removing the underlying obligation creates exactly the kind of gap that's easy to walk into by accident. The temptation is straightforward: documentation isn't required on the page anymore, so why not finally write "the best plan in the area" the way marketing copy has always wanted to?

Two separate problems sit inside that temptation, and they don't fix each other.

The first is the compliance problem, and it's the narrower one. CMS's own tip sheet says "in many cases, plans must still provide supporting data if CMS requests it."¹ That phrase does real work. It means the obligation to be able to substantiate the claim didn't go anywhere, it just moved from "prove it now, in public" to "be ready to prove it later, on request." An agency that writes "best plan in the area" with nothing behind it, because the page no longer requires a citation, is one documentation request away from discovering it can't actually back up what it published. The compliance risk didn't shrink. It moved downstream and got quieter, which is a worse position to be caught in, not a better one.

The second problem is the one most marketing-rule explainers skip entirely, and it's the one that actually costs you organic visibility: a page built around an unsupported superlative was never going to perform well with an answer engine, and nothing CMS did this year changed that.

A worked example of how the documentation request actually plays out

Picture two versions of the same agency homepage, both published the week AEP opens. Version one reads "Call us today for the best Medicare Advantage plans in [county]." Version two reads "We write Medicare Advantage plans in [county], including the area's only 5-star-rated plan for 2027." Under the CY2027 rule, neither one needs a footnote or a citation printed on the homepage itself. Both are, on their face, compliant with the documentation-on-the-page requirement that no longer exists.

The difference shows up the moment anyone asks a follow-up question, whether that's a beneficiary, a state regulator fielding a complaint, or CMS exercising the "on request" power its own tip sheet describes. Version two has an answer ready: the current CMS Star Ratings release, the plan ID, the contract year. Version one has an adjective and nothing behind it. The agency running version one isn't necessarily doing anything CMS will catch this week. It's carrying a liability that sits quietly until the one time someone asks, and "we meant it generally" is not a response that satisfies a documentation request. Plan sponsors remain responsible for the marketing activities of the agents and brokers who represent them, which is exactly why a sponsor's compliance team is often the one asking the question before CMS ever does.

Check your own site before the next AEP mailer goes out

You don't need to wait for a documentation request to find out whether your own pages would survive one. This takes about five minutes and uses nothing but your own site's search box and a text editor.

First, pull every instance of a superlative. Search your own site for "best," "top," "most," and "#1." Don't stop at the homepage. Agency sites tend to collect these phrases fastest in email footers, PDF rate sheets, and seasonal landing pages built fast during a prior AEP and never revisited.

Second, delete the adjective and read what's left. "Call us for the best Medicare Advantage plans" becomes "Call us for Medicare Advantage plans." If the sentence still makes sense and still does its job without the superlative, the adjective was decoration, not information, and you can either cut it or replace it with the fact it was standing in for.

Third, for every superlative you decide to keep, write down what backs it up. A star rating, an enrollment count, a specific benefit a competitor's plan doesn't carry, each with a source and the plan year it applies to. If you can't fill in that blank in under a minute, that's the signal the claim isn't ready to publish yet, documentation requirement or not.

Fourth, date it. A star rating or an enrollment figure is only accurate for one plan year. A page that doesn't say which year its superlative applies to reads as stale, and arguably misleading, the moment the next plan year's data comes out, which is exactly the kind of drift the general misleading-information ban still covers.

Why answer engines never dropped their own proof requirement

CMS and an AI answer engine are not enforcing the same rule, and they were never trying to solve the same problem. CMS cares whether a beneficiary gets misled into the wrong plan. Google, ChatGPT, and Perplexity care about something narrower and, for your purposes here, more immediately relevant: which source is reliable enough to name out loud as the answer to a question.

Google's own documentation on what it calls helpful, reliable, people-first content is explicit about what earns that trust. Among the questions it tells creators to ask about their own pages: "Does the content present information in a way that makes you want to trust it, such as clear sourcing, evidence of the expertise involved" and "Does the content provide original information, reporting, research, or analysis?"³ It also flags the opposite pattern directly, asking whether "the main heading or page title" avoids "exaggerating or being shocking in nature," and whether the content is, plainly, "factually accurate."³ None of that is Medicare-specific guidance. It's Google's general standard for what a trustworthy page looks like, and a bare superlative with nothing behind it fails on every count: no sourcing, no original analysis, and nothing a reader or a model could verify.

Think about the mechanism from the model's side for a second. An answer engine deciding whether to cite your page is effectively asking the same question CMS used to force you to answer on the page itself: can this claim be checked against something specific? CMS stopped requiring you to show your work. A language model summarizing the open web to answer "what's the best Medicare Advantage plan in [county]" has no such requirement removed, because it was never CMS's rule that gave it that instinct in the first place. It's trained to prefer the passage that names a specific, checkable fact over the one that makes an unverifiable claim, and that preference doesn't update because a federal regulator eased a documentation requirement.

Go back to the two homepage versions from the worked example above. Ask an AI assistant "what's the best Medicare Advantage plan in [county]" and picture which of the two pages it has more to work with. Version one, "the best Medicare Advantage plans," gives a model nothing to anchor a citation to, no rating, no benefit, no number, just an adjective that a dozen other competing agency pages also claim about themselves. Version two hands the model a named, dated, checkable fact: a specific star rating for a specific plan year. If the model is building an answer and needs a source to name, the page with the fact is the one it has a reason to reach for. The page with only the adjective is indistinguishable from every other agency site making the same unverifiable claim, which is exactly the crowd an answer engine is built to filter out, not promote.

That's also why this isn't really a story about artificial intelligence being stricter than a federal regulator. It's a story about two different systems rewarding the same underlying behavior, specificity backed by a source, for reasons that have nothing to do with each other. CMS's version of that reward was a compliance requirement. An answer engine's version is simply which page it finds useful enough to repeat. Writing to satisfy the second one tends to satisfy the first one too, almost as a side effect.

What to publish instead of a bare superlative

None of this means you have to avoid the word "best" forever, or that persuasive marketing copy is off the table. It means the fix is the same one this blog comes back to for a reason: replace the adjective with the fact underneath it, and cite the fact.

Notice what the right-hand column has in common across all three rows: a number or a fact that's true independent of your opinion, attached to a source a reader or a model could go check. That's not a workaround for the old documentation rule. It's a better sentence on its own terms, and it happens to satisfy both the compliance instinct the old rule was built around and the trust signal an answer engine is actually looking for.

How we build this in

This is the same discipline we hold every guide on this blog to, including this one: a figure doesn't go on the page unless we fetched it this session, named the source, and stated the date it covers. We re-fetched our own pricing before writing this, because quoting a price from memory is exactly the kind of error this guide is about.

Digital Foundation's Starter tier runs $247 a month and ships 7 new pages a month, including 2 blog posts, 2 location pages, 2 answer pages, and 1 custom page, plus the full 60-point SEO/AEO checklist.⁴ Pro runs $497 a month and adds up to 15 to 18 new pages a month across blog, location, and answer pages, plus a 24/7 AI voice receptionist.⁴ Scale runs $997 a month and ships up to 76 new pages a month, one blog post, one location page, and one answer page every business day.⁴ Every tier includes a 14-day free look before billing starts.⁴

The reason this matters for a guide about superlative claims specifically: every page our build ships is written to the same rule this guide argues for, a specific, sourced, dated fact instead of an adjective with nothing behind it. That's not a separate feature bolted on for compliance. It's the same checklist running whether the topic is a plan's star rating, a county's enrollment count, or a premium figure for the current plan year.

We build on Astro, which ships zero JavaScript by default, so a page like this one arrives as finished HTML instead of a runtime the browser has to boot before anything loads.⁴ Our builds score 90+ on Lighthouse performance on both mobile and desktop, and some score 100, which matters here for a reason beyond raw speed: a page that loads slowly on a phone is a page fewer prospects and fewer crawlers ever finish reading, no matter how carefully it's sourced. A perfectly cited page that never finishes loading on a phone earns the citation nobody sees.

What changes for your agency

Nothing here promises a specific citation, a ranking position, or a guaranteed pass on a future compliance audit, because nobody can source a guarantee like that honestly. What changes is more concrete than that, and it starts before any answer engine or any CMS documentation request ever enters the picture.

None of that requires treating this as a bigger regulatory event than it is. It's one procedural step removed from one part of one rule, for one line of business. The reason it's worth your attention anyway is the gap it opens between "technically compliant" and "actually works," the gap between a page that satisfies the letter of the new rule and a page that still earns a click, a citation, or a documentation request answered without panic. That gap was always there. The only thing that changed on October 1, 2026 is that CMS stopped papering over it for you.

Questions agents ask