CMS just proposed giving traditional MIPS an expiration date — 2029 — and pointing clinicians at ACOs instead. So what does the destination actually look like? Here is every one of the 476 Medicare Shared Savings Program ACOs from performance year 2024, one dot each. Hover any dot. Then drag the size floor and watch the "miracle" savings rates dissolve.
The program CMS wants everyone in already moves real money.
Buried in the 1,803-page CY2027 physician fee schedule rule: traditional MIPS sunsets starting performance year 2029, and the Shared Savings Program gets easier to join. In PY2024, per CMS, 75% of the 476 participating ACOs earned shared savings — $4.1B in payments, with ~$2.5B in net savings left over for the trust funds. This page plots the actual per-ACO results file behind that press release.
476
MSSP ACOs, PY2024
$4.10B
earned shared savings
75%
of ACOs earned savings
10.3M
assigned beneficiaries
The explorer
476 ACOs, one dot each
Horizontal: assigned beneficiaries (log scale — a 3,000-patient physician ACO and Caravan's 329,000-patient giant share one chart). Vertical: savings rate vs. CMS benchmark. Dot size: earned shared savings dollars. Navy = two-sided risk (downside exposure); red = one-sided (training wheels — no downside). The dashed line is break-even.
Minimum ACO size (beneficiaries)0
Risk model
Find an ACO
Critical lens built in: sort the extremes. At floor 0, the best savings rate in the program is 22.1% (Citrus County ACO — 6,198 patients) and the worst is −10.7% (3,868 patients). Drag the floor to 20,000 and the y-axis calms down: the wildest rates live almost entirely in the smallest ACOs. Small-n variance, not miracle medicine — the same rule applies to every "top performer" list you'll ever be shown.
Two-sided riskOne-sided (no downside)dot size = earned $
476
ACOs shown
75%
earned savings
$4.10B
total earned $
4.2%
median savings rate
What this chart can't tell you (read before you're impressed): ① "Savings" means savings vs. a CMS-constructed benchmark, not vs. doing nothing — benchmark methodology is exactly what the new rule proposes to change, again. ② 35 of these ACOs share the identical quality score of 77.05 — that's an assigned/imputed value, not 35 coincidences, so the quality column is part measurement, part administrative default. ③ One row here is a full year of work by hundreds of clinicians compressed into two numbers; the file says nothing about how the savings happened — coding, network design, or actual care redesign.
The builder read: every dot on this chart needs FHIR-based quality reporting, benchmark forecasting, and beneficiary attribution tooling — and CMS just proposed making the dots multiply while the MIPS registry industry gets a 2029 headstone. The five-physician group staring at this chart with no idea which dot to join? That confusion is a product waiting for a founder who can read both the rule and the workflow.