clinicians.build · July 13, 2026

The Denials Nobody Fought

Today's Big Thing: prior authorization isn't an administrative accident — it's a cost-control valve, and the denial is the point. In 2023, Medicare Advantage plans made ~50 million prior-auth determinations and denied 3.2 million of them. Here's what happened next, in 1,000 dots. Reveal the denials — then follow them.

Built on today's Big Thing: “Prior Authorization Is a Fight and AI Won't End It” (MedCity News, Jul 2026)
Data: KFF analysis of CMS 2023 Medicare Advantage prior-authorization determinations

The field

1,000 requests. Watch how few of the “no”s ever get challenged.

6.4%
of ~50 million requests were denied — 3.2 million “no”s
Each dot is one of 1,000 representative prior-auth requests. 64 are red — the 6.4% that plans denied. At industry scale that's 3.2 million denials in a single year. Now press step 2.
Press “Now follow the denials.” The field reframes: all 1,000 dots become the 3.2 million denials. Watch how many were ever appealed — and what happened when they were.
The critical lens

“82% overturned” is true and misleading at the same time

The reversal rate is real: of denials that reach appeal, most are overturned — strong evidence many should have been approved at first pass. But two things complicate the headline. First, the 6.4% average denial rate hides where it bites: KFF found post-acute and skilled-nursing requests denied at multiples of the overall rate — the sickest, most expensive patients. Second, the 82% overturn figure is an industry average that varies enormously by insurer — from about 42% at Kaiser to 94% at Centene. A single number flattens a machine that behaves very differently plan to plan.

The denials that never became a statistic. This whole picture starts after a request was submitted and adjudicated. It can't see the care a clinician never ordered because the prior-auth burden wasn't worth it, or the patient who abandoned treatment mid-appeal. The AMA's 2024 survey puts the friction at ~39 prior auths per physician per week and ~13 hours of staff time — a deterrent that works precisely because most people give up. That's the machine functioning as designed.
Why it matters for builders
Today's thesis in one image: a denial you prevented is worth more than one you overturned three weeks later. The overturn funnel is the appeal-bot's whole market — and it's downstream, slow, and mostly unused. The bigger prize sits upstream: check the payer's own published criteria before you submit, so the “no” never happens and the 13 hours are never spent. Don't build the tool that fights the denial. Build the one that makes the denial impossible to justify.
Read the piece — “Prior Auth Is a Fight and AI Won't End It” → Or read today's newsletter → · See the KFF 2023 analysis →