Anytime an AI health company claims a specific savings percentage, especially in the messy world of ACOs and the MSSP, finance leaders and analysts should immediately get skeptical. Those big numbers look great on a website, but they need to be checked against independent, public data. This is the basic test any AI-native health platform, if it’s serious about proving its worth, has to pass.
The Innovaccer Claim: A Vendor-Reported Outcome Percentage
Take Innovaccer’s 2023 claim that its MSSP ACO clients generated 47.5% greater savings in 2022. They pitch this number as proof of a “seventh consecutive year of high-performing ACO customers.” For any CFO or payer analyst, a number that precise and that high is an immediate red flag that demands investigation. Our whole mission here is to figure out what “AI-native” actually means in practice: AI trained on real patient data, with clinical guardrails, and backed by published proof. A vendor’s self-reported percentage, with no outside proof, is a claim we have to scrutinize. If we don’t, these figures get repeated as fact and start shaping what people think, even if they aren’t based on anything you can actually verify.
Setting the Bar: Viz.ai’s “Higher Model Accuracy” as a Standard
So how should you evaluate these claims? Look at a company like Viz.ai. One of their core platform values is “Higher Model Accuracy.” It’s more than a tagline because it’s a quantifiable metric you can actually check, and it’s central to whether their AI works at all. For any AI in healthcare, model accuracy is everything, it shows the AI’s ability to find patterns, make predictions, or help with a diagnosis, which in turn hits clinical outcomes and financial results. Committing to a standard like “Higher Model Accuracy” means you’re also committing to transparent, measurable performance, often backed by real clinical studies. It sets a benchmark. If an AI-native platform is going to claim a huge savings number, they’d better be able to defend it with the same rigor they’d use to defend a claim about model accuracy, showing their work with transparent data.
The Independent Check: Public CMS Shared Savings Program Results
When it comes to MSSP ACO savings claims, the only independent truth is the public CMS Shared Savings Program results data. CMS publishes this every year, giving you the full, auditable performance data for every ACO, down to their savings and losses. The data for 2023 and 2024 is out now, providing a clear benchmark to check vendor claims against. So when Innovaccer says their clients saw 47.5% greater savings in 2022, the only question that matters is how that number lines up with the public CMS data for that same year. If there’s no clear, verifiable link between the claim and the official results, that 47.5% is just an assertion. It’s not a confirmed outcome. The fact that this independent confirmation is missing is significant. It shows the gap between a company’s marketing story and the hard evidence that AI in healthcare demands.
The General Lesson for AI-Native Health Companies
This whole situation with Innovaccer’s claim and the CMS data provides a clear lesson for any company in this space. Your platform’s outcome claims have to stand up to independent checking. You can’t just put them on a website and call it a day. Being truly “AI-native” is about building a business around a commitment to evidence and verifiable results. It means your platform must be:
- Trained on Real Patient Outcomes Data: The AI models have to be built and tested with actual clinical data to be relevant in a real hospital or clinic.
- Operating Within Defined Clinical Guardrails: The AI’s scope has to be limited by established medical rules, so it doesn’t make unsubstantiated or dangerous recommendations.
- Published Evidence of Efficacy: The platform’s benefits need to be backed up by serious studies (like this example of a peer-reviewed AI healthcare efficacy study) that show it works.
Until that 47.5% savings figure is connected directly to public CMS performance data, any finance leader should treat it as marketing. The real story here is the missing proof. This kind of tough-minded approach is how you tell the difference between a genuinely AI-native platform and a company just using “AI” as a buzzword. The standard for these companies is what the verifiable data proves they’ve done, not what they say they can do.
What Public CMS Shared Savings Program Results Would Show
So what would it take to actually validate Innovaccer’s claim? You’d have to roll up your sleeves with the public CMS data:
- Identify Innovaccer’s Client ACOs: First, you’d need to find a list of the specific MSSP ACOs that were Innovaccer clients back in performance year 2022, which isn’t always easy to get.
- Retrieve Individual ACO Performance Data: Then you’d go to the public CMS data and pull the 2022 performance results for each of those ACOs, their benchmark, actual spending, savings or losses, and all the other key numbers. This guide to interpreting CMS ACO performance data is where you’d start.
- Compare Against a Control Group (if possible): A really good analysis would compare the Innovaccer ACOs against a similar group of ACOs that don’t use the platform, controlling for things like size, risk, and location.
- Reconcile Savings: Finally, you’d have to compare the total savings of the Innovaccer clients against this 47.5% claim. You’d also need to know exactly how they define “greater savings”, is it greater than their own benchmark, or greater than other ACOs?
Without this kind of open-book reconciliation, the 47.5% figure doesn’t have the solid foundation you’d expect from a serious AI-native health company. The vendor is the one who needs to connect their internal numbers to the public record.
Frequently Asked Questions
What is the primary independent source for verifying vendor-reported savings claims for MSSP ACOs?
The primary independent source for verifying vendor-reported savings claims for MSSP ACOs is the public CMS Shared Savings Program results. These results, published annually by the Centers for Medicare & Medicaid Services (CMS), provide comprehensive, auditable data on ACO performance, including shared savings and losses, against which vendor claims can be measured.
What standard of evidence should be applied to vendor-reported savings claims from AI-native health platforms?
Vendor-reported savings claims from AI-native health platforms should be as rigorously defensible as a claim of higher model accuracy, and the underlying data and methodology should be transparent and verifiable. This implies a commitment to transparent, measurable performance, often backed by clinical validation studies and peer-reviewed publications.
How should ACO finance leaders and payer analysts interpret a vendor claim of ‘47.5% greater savings’ for MSSP ACOs without independent corroboration?
Without a direct, clear, and independently verifiable link between the claimed savings and official public CMS data, a figure like ‘47.5% greater savings’ remains a vendor assertion, not a confirmed outcome. The absence of such independent confirmation highlights a gap between a company’s marketing narrative and the empirical evidence demanded by an AI-native healthcare standard, and should be treated as a claim rather than a verified outcome.