In June 2025, more than fifty health plans covering roughly 270 million Americans signed a voluntary commitment to simplify prior authorization. The first wave of promises took effect January 1, 2026: fewer services subject to prior authorization, a 90 day honor period for existing authorizations when a patient changes plans, and clearer denial explanations with appeal guidance.
A year later, AHIP reports that participating plans have eliminated about 6.5 million prior authorizations, an 11% reduction.
Here is the part worth sitting with. When the American Medical Association surveyed physicians about two of the largest participating plans, only 16% working with UnitedHealthcare and 16% working with Cigna said the changes had reduced the number of prior authorizations they complete.
Both numbers can be true at once. That gap is the thing your practice should be interested in.
An 11% reduction across a national book of business is an aggregate. Your practice does not experience an aggregate. It experiences a specific payer, in a specific market, for a specific set of codes.
Plans committed to reducing prior authorization requirements “as appropriate for the local market each plan serves.” That language does real work. A plan can meet its commitment by removing requirements on high-volume services with clear clinical guidelines and stable utilization, which is exactly where the easy volume is. If those are not the services you bill, your burden did not move.
So the useful question is not whether reform worked. It is which of your payers changed which codes, and whether your front-end workflow ever caught up.
When a payer removes a service from its prior authorization list, nothing arrives to tell your staff. The requirement quietly stops existing. Your workflow does not.
We regularly see practices still obtaining authorizations for services that no longer require them, sometimes many months after the policy changed. That is real staff time spent on work the payer stopped asking for. It does not show up as a denial, it does not show up in a report, and nobody notices because the claims pay.
The inverse also happens. A plan reduces requirements in one category while quietly adding them in another, and the first sign is a denial for no authorization on a service that never needed one before.
Pull your top five payers and your top twenty-five CPT codes by volume. For each pair, confirm the current authorization requirement directly against the payer’s published policy, not against your internal reference document. Internal cheat sheets drift.
Compare the authorizations your staff obtained in the last six months against that list. Anything you authorized that no longer requires it is recoverable staff capacity. Quantify the hours. It is usually a larger number than people expect.
Check whether your payers are honoring the 90 day transition. When a patient switches plans mid-treatment, the new plan committed to honoring the prior authorization for benefit-equivalent in-network services. If your staff is starting from scratch on those cases, that is a commitment your practice can hold the payer to by name.
Track denial reason quality. Plans committed to clear explanations with appeal guidance. Vague denials are now inconsistent with a public commitment, which is useful leverage in an appeal and useful evidence in a payer meeting.
The 2027 commitments are more substantial than the 2026 ones. Signatory plans have targeted at least 80% of electronic prior authorization approvals answered in real time, built on standardized FHIR APIs. That aligns closely with the CMS-0057-F API deadline we covered last month, which means the technical foundation is arriving whether individual plans move quickly or not.
Practices that know their own baseline will be able to tell whether 2027 delivers. Practices that do not will be reading press releases and guessing.
If you want help building that baseline, or you suspect your team is doing authorization work no payer is asking for anymore, schedule a discovery call or book a free audit. It is a straightforward thing to measure, and the answer usually pays for itself.
