We found no published CMS 85% accuracy requirement

The figure is quoted as settled CMS policy in vendor compliance guides. We went to the primary text to find it.

By the Rollcall Health research team · 21 August 2026

Ideon's compliance guide, published 27 February 2026, states it plainly: “CMS requires a minimum 85% directory accuracy rate for Medicare Advantage and ACA marketplace plans.” The page uses the threshold six times and hangs an enforcement ladder off it, and Google prints that sentence as the answer when somebody searches what CMS requires. Other vendors carry the same figure. None of them says where it came from, so we went looking.

Where we looked

The documents the claim would have to live in are the Medicare Advantage directory rule as finalised in CMS-4208-F2, both its codified text at 42 CFR 422.111(m) and its preamble; the Marketplace provision at 45 CFR 156.230; the REAL Health Providers Act, enacted as section 6220 of the Consolidated Appropriations Act, 2026; and CMS's own technical guide for supplying MA directory data to Medicare Plan Finder, in the November 2025 draft and all three revisions since.

The figure is in none of them. The one standalone 85 in the Federal Register document is “85 FR 25510”, which is a volume citation to a rule from 2020. The current technical guide, published on 14 August 2026, contains no 85 at all.

What CMS actually wrote

CMS gives three reasons it will suppress a plan's data from Medicare Plan Finder. The organisation misses the attestation, the submission fails validation, or CMS “reports on data quality issues that exceed a threshold published by CMS”.

That last phrase has survived every revision of the guide, and each time it is followed by a line saying further guidance on suppressions will be sent in a future communication. As of 14 August 2026 it has not been sent, so the trigger exists and the number behind it has never been written down.

Where the number probably came from

CMS has published an 85% about provider directories. Once. It comes from the secret shopper review of Medicare Advantage online directories that CMS reported in January 2017, and it describes what that review found rather than anything a plan is required to achieve. The sentence reads:

“In considering the deficiencies that are most likely to impact access to care, 85% of those identified (4,476/5,352) were of the highest weighted, most egregious errors.”

That sentence counts errors: of the problems CMS found, most were the serious kind. Run the arithmetic and even the quote's own figure wobbles, because 4,476 of 5,352 is 83.6%, which CMS rounded up. Somewhere between 2017 and the current crop of compliance guides that sentence came to mean that 85% of a directory has to be correct, which is close to the reverse of what it says.

We cannot prove that is the source and there is no point pretending otherwise. One other 85 sits near enough to have done the same job: individual plans set themselves internal directory accuracy goals of 85%, and some publish them. A goal a plan chose for itself is not a threshold CMS imposed, and a guide that presents one as the other is where the confusion becomes expensive.

The marketplace half

Medicare Plan Finder is a Medicare Advantage surface. So is the attestation, and so is the technical guide. Marketplace directories run under 45 CFR 156.230, which asks a QHP issuer to publish an up-to-date, accurate and complete directory in a machine-readable file, sets no percentage, and has nothing to do with Plan Finder. A threshold set for one programme would not reach an issuer in the other even if it existed.

So what is the requirement

None of it is written as a percentage. 42 CFR 422.111(m) tells an MA organisation to publish through CMS, to submit in the format and at the times CMS sets, to update within 30 days of learning of a change, and to attest at least annually that what it submitted is accurate. The REAL Health Providers Act adds, for specified MA plans from plan year 2028, verification of directory information at least every 90 days, though for hospitals and other facilities the Secretary designates it can run as infrequently as once a year. A provider comes out of the directory within five business days of the organisation determining they have left the network, so the clock starts at the determination rather than the departure. CMS-published accuracy scores follow from plan year 2029, with the methodology left to the Secretary.

What the rules do not give is an error allowance or a published safe harbour. They require submitted information to be accurate, and CMS has not translated that obligation into a stated pass mark. That is a harder thing to budget against than a number you can pass, which may be part of why an invented one travelled so well. CMS is blunt about the limits of its own process: “The process outlined above will not validate the accuracy of the provider directory information submitted by MA organizations. Data accuracy is the responsibility of the MA plan.”

Sources reviewed

Every document below was fetched and read for this page on 21 August 2026.

• 42 CFR 422.111, codified Medicare Advantage directory rule, current text.
• 45 CFR 156.230, Marketplace network adequacy and directory provision, current text.
• CMS-4208-F2 final rule, Federal Register, 19 September 2025.
• Technical Implementation Guide for MPF MA provider directory data, CMS, revision dated 14 August 2026, with the November 2025 draft and the February and May 2026 revisions.
• Online Provider Directory Review Industry Report, CMS, 13 January 2017.
• Consolidated Appropriations Act, 2026, section 6220, enacting the REAL Health Providers Act.
• Ideon, CMS Provider Directory Requirements, published 27 February 2026, the page quoted above.

Why we bothered

A compliance team planning against 85% is budgeting for a target nobody set, and will eventually be scored on one that has not been written yet. We publish our own figures with the file, the date and the method attached. We would rather be held to that than benefit from a rule nobody wrote.