Benchmarked Medicare-Focused
Plans Outperform the Market
in 2027 Star Ratings

Star Ratings and the Medicare Advantage Market

Each autumn, the Centers for Medicare & Medicaid Services (“CMS”) rates Medicare Advantage and Part D contracts from 1 to 5 stars on about forty measures of clinical quality (“HEDIS”), member experience (“CAHPS”), health outcomes, plan administration and drug plan performance. The 2027 ratings, released on October 8, mostly measure 2025, the year of the costs in the 2026 Medicare edition of the Sherlock Benchmarks, and they set bonus payments for 2028.

Figure 1 weights each contract’s overall Star Rating by enrollment, consistent with CMS fact sheets, and aggregates the contracts by the organization that holds them.¹

The eleven Medicare-focused plans in the Sherlock universe are the participants in the 2026 Medicare edition of the Benchmarks. They average 4.12 stars for 2027, with 76% of their members in contracts rated 4 stars or higher, down from 90% for 2026.

The six national publicly traded companies hold 66% of Medicare Advantage enrollment in rated contracts, and 73% of their members are in contracts rated 4 stars or higher, up from 61% for 2026. Figure 1 shows each company.

Click to enlarge

Figure 2 tracks these groups since the 2020 ratings, holding the Sherlock universe at its eleven current plans. The universe has rated above the market every year since 2022, by 0.13 to 0.52 stars. The market’s high for 2022 came when pandemic-era relief let contracts keep the better of two years’ ratings on most measures. The Sherlock Benchmarks universe’s rating peaked at 4.67 stars for 2023 and was 4.12 for 2027, while the Medicare Advantage market fell from 4.15 to 3.99 over the same years.

Click to enlarge

Quality Bonus Payments and Administrative Costs

CMS pays each Medicare Advantage contract against a county benchmark, set at 95% to 115% of what traditional Medicare spends per beneficiary there. A contract rated 4 stars or higher has its benchmark raised 5 percentage points, or 10 in “double-bonus” counties, unless the county’s cap absorbs some or all of it. When a plan bids below its benchmark, part of the difference becomes a rebate that must go back to members as extra benefits, lower cost sharing or lower premiums. For contracts rated from 3.5 to below 4.5 stars, the rebate is 65% of the difference.

KFF estimates that the bonus will total at least $13.4 billion in 2026. For individual plans it averages $381 per enrollee annually, or $32 PMPM, counting enrollees in plans that earn no bonus.

Figure 3 models these payments in comparison with administrative costs. Median Medicare Advantage administration among the eleven plans was $127.80 PMPM, as reported in our September Navigator. The five Sherlock Benchmarks functions that most directly support Star Ratings (Medical Management, Customer Services, Provider Network Management and Services, Enrollment / Membership / Billing, and Member and Group Communication) cost about $32 PMPM combined. That is one-quarter of administration and about the same as the average bonus. Spread over only the members of bonus plans, KFF’s total comes to about $47 PMPM, or 37% of administration.

Click to enlarge

The lower part of Figure 3 illustrates a plan at the median, with implied premium equivalents of about $1,241 PMPM standing in for its benchmark. Where the cap does not apply, a fourth star raises that benchmark 5 points, about $62 PMPM, and the plan returns 65% of that, about $40, to members as extra benefits. That equals 31.6% of its administration, or 63.1% in a double-bonus county. For a plan that earns it, the bonus is therefore worth about a third of its administrative budget, and somewhat more than the costs of all of the functions associated with such ratings. A contract that fell from 4.0 to 3.5 stars in October loses those dollars in 2028.

Administrative Functions Supporting Star Ratings

Star Ratings measure clinical care by providers, but also administrative work such as closing gaps in care, handling member inquiries, deciding appeals on time and reaching members for preventive screenings. The Sherlock Benchmarks measure the cost of that work by function.

Medical Management is roughly half of the cost of the five functions. Within it, Quality Components, the sub-function for HEDIS data collection, NCQA accreditation and the Star Ratings program, cost $1.77 PMPM in Medicare Advantage in 2025, or 1.4% of administration.

Figure 4 shows how these costs changed. Each figure is the median of the plans’ own changes. From 2022 to 2025, the five functions grew 3.2% per year, compared with 1.2% for total administration. Customer Services grew fastest, at 11.0% per year, and Quality Components fell 2.9% per year.

From 2024 to 2025, spending fell 1.9% in Medical Management and 5.9% in Customer Services, though the five functions together rose 2.4%. The reductions followed the drop in the universe’s rating from 4.67 stars for 2023 to 4.34 for 2024.

Click to enlarge

Sherlock Benchmarks Metrics and Star Ratings

No single administrative function would be expected to move a plan’s overall rating, which combines so many measures. We therefore tested Sherlock Benchmarks metrics, including function costs, staffing, plan characteristics and the call center results in our Operational Metrics, against every CMS measure, domain and summary rating. The tests use the sixteen plans of our three universes (Medicare, Blue Cross Blue Shield, and Independent / Provider-Sponsored) that reported Medicare Advantage results in all four editions from 2023 to 2026.

We ran many regression analyses, and few results were meaningful. Figure 5 shows some of the results, with those that met the finding rule described in How We Performed This Analysis shaded. Each slope is the difference in the CMS rating associated with each additional unit of the Sherlock metric, such as each additional $1 PMPM of spending.

Plans more focused on Medicare had higher overall ratings, 0.47 stars higher for each additional 10 percentage points of membership in Medicare Advantage (P-Value of 0.02). The result holds whichever plan is dropped from the analysis, indicating that no outlier skewed the results. CMS’s 2027 fact sheet notes a separate pattern, that contracts with more years in Medicare Advantage generally rate higher.

Click to enlarge

Plans that spent more on Customer Services scored higher on Reviewing Appeals Decisions and on call center language interpreter and TTY availability (P-Values of 0.04). The second result is tentative. Most plans already score near the maximum on it, and Figure 5 shows it accounts for little of the variation among plans.

Customer Services spending showed no reliable relationship with the service domain ratings, and the five functions combined showed no relationship with the overall rating.

These results should be read with the pandemic in mind. The first year tested, 2022, still reflected COVID and plans recovered at different rates. Each new edition of the Sherlock Benchmarks adds a year of data, and perhaps trends will emerge as it builds.

These results suggest several implications for health plan management. First, for a plan that earns it, the quality bonus is worth about a third of its administrative budget. A plan losing its fourth star keeps those expenses and loses those benefit dollars. Second, plans more focused on Medicare rate higher. Third, Customer Services spending showed no reliable relationship with CMS’s service domain ratings, though plans that spent more on Customer Services scored higher on appeals.

How We Performed This Analysis

Contract-level overall ratings are from the CMS Part C and D Star Ratings Data Tables for the 2020 to 2027 ratings, weighted by CMS Monthly Enrollment by Contract (September 2026 for the 2027 ratings). A plan’s or company’s rating is the enrollment-weighted average of its contracts’ ratings, and its 4-star share is the share of its members in contracts rated 4 stars or higher. National figures are as CMS first announced them, except that comparisons with 2026 use the 2026 figures as CMS updated them in June and July 2026. The updates followed a federal court’s ruling, which CMS has appealed, in Clover’s challenge to its 2026 rating. CMS first published the 2026 average as 3.98 stars, and the updated figure is 4.01.

The six publicly traded companies are UnitedHealth Group, Humana, CVS Health (Aetna), Elevance Health, Centene and Molina, grouped by the parent organization CMS designates each year, with The Cigna Group included through the 2025 ratings. The Sherlock universe is the eleven plans in the 2026 Medicare edition, matched to the Medicare Advantage contracts of the entities that report to us.

Expense data are from the Medicare editions of the Sherlock Benchmarks for 2023 through 2026 (2022 through 2025 results), with the Blue Cross Blue Shield and Independent / Provider-Sponsored editions for the tests. They are for Medicare Advantage only, excluding Special Needs Plans where a plan reports them separately, and before that product’s Miscellaneous Business Taxes. Cost levels are medians of the eleven plans, and the five functions’ cost is the sum of their medians. The $127.80 includes pharmacy and behavioral health administration, which the five functions do not, and excluding that administration the five are 28% of the total. Provider incentives, and supplemental benefits funded by the bonus, are medical or benefit costs, not administrative expenses.

Figure 4 follows the same plans in each comparison, the eight plans in both the 2023 and 2026 editions and the seven in both the 2025 and 2026 editions. That is why its total administration declines for 2025, while the September Navigator’s median, drawn from a changing group of plans, rose 1.1%.

Each test analyzes the relationship between a Sherlock Benchmarks metric and a CMS rating, matching each year’s costs with the ratings that measure that year.

A P-Value is the chance of finding a slope at least this far from zero if there were no relationship at all. A P-Value above 0.10 means the sample could not demonstrate a relationship, not that none exists. A result is a finding only when its P-Value is below 0.10. When we report a relationship, it must also hold on each plan’s average across the years or within plans, keep its direction in most years, and hold when any one plan is dropped. R² is the share of the variation among plans, within each year, that the metric accounts for.

Among all 36 plans of our three universes that reported Medicare Advantage results in any of the four editions, the Medicare Advantage share result is significant as well.

Notes

1. The weighted figures include only contracts with a numeric overall rating. Contracts that CMS reports as too new to be measured, as not having enough data, or as not applicable are excluded. They represent about 2% of enrollment in the market and less than 1% in the Sherlock universe.