Who Cleared the Bar
Suki says it will hand back its fees if “agreed-upon performance outcomes” aren’t reached. Medicare has run a deal like that with doctors’ groups since 2012: beat a spending benchmark by enough, and you share the savings. Here is every ACO’s 2024 result against the bar it had to clear. The bar, not the savings, decided who got paid.
Read the Suki guarantee story →
Or read today’s newsletter →
Primary source: Jill Hughes, “Suki debuts agentic operating system,” Healthcare Dive, November 18, 2026 (fees back “if agreed-upon performance outcomes are not reached”). Data via MIMI Labs: CMS
Shared Savings Program Performance Year Financial and Quality Results, performance year 2024 (file of July 20, 2026;
datacmsgov.mssp_results), one row per ACO. The 2025 results CMS announced Nov. 17 are in the
fact sheet; the ACO-level 2025 file was not in the dataset at build time.
Each dot is one ACO. Across: how many Medicare patients were assigned to it (log scale). Up: how far its spending came in under its benchmark. The short red dash at each dot is that ACO’s minimum savings rate, the bar it had to clear to share anything. One-sided ACOs (no downside risk) get a bar from CMS that rises as the ACO gets smaller: 2% for the largest, past 3.5% for most under 6,000 patients. Two-sided ACOs choose a flat bar from 0% to 2% or the same size-based scale.
cleared its bar under benchmark, missed the bar, $0 missed the bar, partial share (2024 rule) spent over benchmark
ACOs shown
under benchmark
saved, got $0
their spend under benchmark
Same ACOs, sorted by how far each landed from its own bar · dot area = assigned patients
Left of the red line: under the bar. The red dots there all spent less than their benchmark. The flat-bar setting above applies here too.
80/20 lens
The bar exists for a fair reason: a small ACO’s spending swings by chance, so a 3% “saving” on 5,000 patients may be noise. Drag the size slider and watch the cloud narrow: among ACOs with fewer than 10,000 patients the middle half of savings rates spans points; among those with 40,000 or more it spans . That is the case for a bar. The catch is who sets it and when. Every ACO knew the rule that sets its number before 2024 began, and CMS computed the result from claims it holds. A vendor guarantee written “subject to contractual terms” and measured “using customer-provided data” needs the same three things in writing: the baseline, the bar, and who runs the numbers.
Read before you quote it
- This is 2024, not the 2025 results in today’s issue. The ACO-level file for 2025 wasn’t in the dataset when this was built. The pattern, not the exact counts, is the point.
- “Under benchmark” is CMS’s benchmark minus actual spending, before the bar or sharing rates. It is not Medicare’s net savings, which subtracts what CMS paid out.
- Half shares below the bar. From 2024, certain low-revenue BASIC-track ACOs that miss the bar but save can share at half the usual rate (CY 2023 fee schedule rule). The pink dots are ACOs below their bar with a positive payout; the file does not say which rule applied.
- Quality gates too. Payment also depends on quality scores, sequestration and caps. Two ACOs cleared a 0% bar with savings under 0.2% and were paid $0; they are drawn navy.
- The flat-bar setting is a count, not a payout model. It shows who would clear a single threshold. It does not recompute dollars, since sharing rates differ by track and quality.