Accountable care organization (ACO) models are a leading model of value-based care (VBC). Sponsored by Medicare, Medicaid, or commercial insurers, these models offer providers a chance to improve outcomes while keeping costs under control. But complexity in program rules, shifting benchmarks, and data lags make it difficult to manage performance, predict revenue, and invest strategically.
That’s where predictive analytics comes in—not just as a technical tool, but as a strategic asset that can help ACO leaders confidently plan, budget, and invest. When applied effectively, predictive modeling helps ACOs understand what’s coming—financially and operationally—so they can make smarter investments, manage resources more effectively, and stay ahead of risk. With the right data, an effective approach to predictive modeling, and meaningful support, ACOs can gain the insights and confidence they need to succeed with value-based care.
Why Predicting Shared Savings Is So Difficult
In ACO programs such as the Medicare Shared Savings Program (MSSP), organizations are assigned a financial benchmark. If they spend less than that benchmark while meeting quality targets, they keep a portion of the savings. But two key factors make this hard to predict in advance: the benchmark itself changes after the performance year ends, and the organization’s own spending and performance can vary in unexpected ways.
Savings rates can vary significantly from year to year. Mathematica’s analysis shows that since 2020, typical MSSP savings rates have varied by about two percentage points from year to year—nearly half the average rate of 4%.
The same research also found that initial ACO benchmarks, provided in advance of the performance year, change significantly after the end of the year, with final benchmarks increasing by 12% on average in 2023.
Complicating matters even further, most ACOs receive data on only their own attributed patients—and they receive that data with a lag. They often lack visibility into broader market trends, peer performance, or social risk factors that may influence outcomes. Without timely and complete data generating relevant insights, it’s nearly impossible to course-correct during the year or forecast end-of-year results with confidence.
Why Prediction Matters for ACO Strategy
In a risk-based environment, uncertainty is more than a financial headache—it serves as an operational constraint. ACOs that don’t have confidence in their projected earned savings may hesitate to make critical investments in areas that could improve their performance—such as care coordination, health IT, or patient engagement—or to share performance expectations with participating providers.
Many of these investments need to be made upfront and cannot be done incrementally; for example, hiring a care coordinator, upgrading an EHR system or launching a new program for high-risk patients. Being able to forecast savings more accurately helps ACOs act decisively.
Accurate predictions drive more accurate budgets, more consistent network provider compensation models, and improved financial reporting. When leaders can see more clearly where the organization is headed, they can allocate resources more effectively and manage risk more proactively.
What It Takes to Make Accurate Predictions
Effective prediction requires more than data science—it needs deep program knowledge, broad data, and tailored modeling.
- A Deep Understanding of the Payment Model
Program rules—how benchmarks are set, adjusted, and reconciled—vary by model and evolve over time. Whether MSSP, ACO REACH, a Medicaid ACO, or another alternative payment model (APM), precise understanding is key to forecasting outcomes and interpreting results. - Access to the Right Data
Most ACOs have detailed claims and enrollment data for their patients, but broader visibility is critical. For example, market trends drive benchmark adjustments, so modeling must incorporate complete enrollment and provider claims data. Social determinants of health and negotiated prices also enhance predictive accuracy. - Customized Analytics Modeling
Strong models combine statistical rigor with real-world relevance. Beyond advanced analytics, models must align with the ACO’s context and link high-level forecasts to specific actionable insights. Results must be timely and adaptable as rules evolve.
Investing in Predictive Modeling
Some ACOs, particularly the largest and most well-resourced, may consider building this modeling capacity in-house. But the cost and complexity can be prohibitive. Five years ago, McKinsey estimated the upfront cost of developing an in-house data and analytics solution that could support an ACO at $24 million for upfront development, plus $6 million a year in annual costs, including the data scientists and analysts needed to generate insights.
This kind of investment is likely out of reach for many ACOs. But working with an experienced data and analytics partner can reduce the resource burden on ACOs (e.g., investments in infrastructure and staffing). At the same time, working with a data analytics partner offers ACOs the opportunity to scale a solution to meet their specific needs and resources, as well as the ability to realize results earlier rather than later.
The Advantage of Working with an Experienced Partner
Mathematica has decades of experience supporting the design, implementation, analytics and evaluation of value-based care initiatives. At CMS, Mathematica’s experience includes supporting MSSP, ACO REACH, Primary Care First (PCF), the Comprehensive Care for Joint Replacement (CJR) model, Medicaid and all-payer VBC models, and more, at both the program and participant level. Because of this deep program involvement, we understand the mechanics of these models from the inside out—including how benchmarks are calculated, how performance is measured, and how policy changes affect financial projections.
A partnership with Mathematica includes:
- Analytics based on comprehensive, real-time data, including 100 percent of Medicare enrollment and claims data (Parts A-D), 100 percent of Medicaid claims, reimbursement rates from national and regional health plans, and provider-specific Medicare rates.
- Multidisciplinary expertise, from data scientists, clinicians, statisticians, economists, and operations specialists with extensive experience working with healthcare providers, Medicare and Medicaid programs, and commercial payers.
- Proven delivery of predictive analytics for CMS and private sector partnerships.
- Tailored solutions that that match your organization's maturity, needs, and resources.
- Action-ready results, integrated with the ACO’s operations to support better care, smarter investments, and more predictable performance.
Predictability Builds Confidence—and Confidence Drives Action
When ACOs can forecast performance more confidently, they make smarter, faster decisions—improving care and lowering costs. Mathematica’s unique blend of policy expertise, robust data access, and customized analytics can help your ACO thrive in today’s complex value-based landscape.
Let’s talk about how predictive modeling can help your organization take the next step.
