A Half-Star Is Worth a Fortune or Nothing. Most Plans Price It Wrong.
Medicare will spend a record $13.4 billion on Medicare Advantage quality bonuses in 2026 — even though the share of enrollees in 4-star-or-better plans fell from about 90% in 2022 to roughly 64% now. The money went up while the ratings went down. That is the first clue that Star revenue is not what most plans think it is. Six things from building an instrument that prices it.
1. It is a cliff, not a curve. Below 4.0 stars there is no benchmark bonus at all. At 4.0, the plan gets a 5-percentage-point benchmark increase — doubled in some counties — and steps up a rebate tier. Nearly all the money sits at one threshold. So the only honest question is never "what will our stars be" but "how far are we from the cliff, and how stable is the edge."
2. The cut-points move after the game is over. Star cut-points are set at the end of the performance year, from the distribution of every plan's scores. You can improve in absolute terms and still lose a star because everyone else improved more. Then CMS added Tukey outlier deletion, which raised the cuts further. Any projection priced against last year's cut-points is mispriced by construction — the target moved after you shot.
3. The program rewards crossing, not staying. Peer-reviewed analysis: contracts cross the 4-star threshold about 1.6 times on average, and hold the bonus consecutively only about a quarter of the time. A one-year crossing booked at face value into a multi-year forecast ignores the program's central feature — it pays the crossing, not the staying. "We hit 4.0 once" is not a durable revenue line.
4. The interventions vendors sell are the ones that don't move. The only two genuinely causal studies of the bonus program both found it did not buy measured quality. A decade of star gains concentrated in about nine medication measures — provider-led — while patient experience, the thing most improvement vendors sell, stagnated. And nothing peer-reviewed shows a plan reliably moving its own CAHPS. A deck promising half a star from member-experience outreach is promising an effect the literature has never observed.
5. There is no credible ROI figure — and that's the finding. Search for an independent, published cost-per-star or ROI number for Stars improvement programs and you will find only vendor marketing, denominator-free. The absence is not a gap to fill with an assumption. It is a fact to put in the board deck: nobody has measured this, so treat every quoted ROI as a number that has never met a counterfactual.
6. The methodology is in court. SCAN and Elevance forced a program-wide recalculation. UnitedHealthcare won on a single disputed phone call. Humana's appeal is pending. Clover has CMS recalculating every plan on HEDIS/HOS/CAHPS-only data. The rating your projection rests on can be recalculated after it's published — the same shape of risk as a RADV audit, and it belongs in the same disclosures.
We built these into an instrument: the cliff arithmetic, the cheapest-star decomposition, the volatility band, and the stability flags — every threshold cited, the vendor uplift taken as a claim under test rather than an input. Browser-only; nothing entered leaves the machine.
A Star projection without a volatility band is a coin flip with a spreadsheet.
What is your plan's distance to 4.0 worth in dollars — and did anyone price it against this year's cut-points, or last year's?