A new analysis from former CVS Health rebate pricing lead Karishma Mandal argues that two little-known initiatives, CostVantage and TrueCost, could reshape billions of dollars in U.S. pharmaceutical spending.
For most of the last decade, the way Pharmacy Benefit Managers (PBM) handled rebates was a ledger problem hiding inside a pricing problem hiding inside a trust problem. The people who could see it most clearly tended to work inside it. Karishma Mandal was one of them.
As a Sr. Product Manager on rebate pricing at CVS Health, Mandal spent years working on the systems that handle one of the largest and least-understood flows of money in American healthcare. She argues that the industry’s legacy infrastructure was never built for the scale or complexity of the modern pharmaceutical market, and that the consequences have been quietly trickling into complicated payment structures.
The U.S. spent roughly $378 billion on prescription drugs in 2023. About 27 cents of every dollar in pharmacy spending moves through the rebate machinery. This is the gap, she contends, that CVS started closing in 2023 and 2024 with two initiatives most Americans have never heard of: CostVantage and TrueCost. They are not consumer brands. They are infrastructure. And if they work the way they are designed to, they could quietly change how a meaningful chunk of the U.S. healthcare economy moves money.
Why the rebate problem is so big
Pharmacy benefit managers, or PBMs, sit between drug manufacturers and the health plans that pay for prescriptions. They negotiate rebates from manufacturers in exchange for placing drugs on formulary lists, and those rebates are supposed to flow back to plan sponsors. CVS Caremark alone serves close to 90 million Americans. Three PBMs, Caremark among them, together cover roughly 80 percent of the insured population.
The rebate flow itself is straightforward in theory. In practice, Mandal writes, it runs through a chain of incompatible systems, contractual edge cases, and human reconciliation steps that were never built to scale. Manufacturer invoices contain errors. Months pass between when a drug is dispensed and when the rebate gets remitted. Multiply that across hundreds of thousands of unique drug SKUs, thousands of manufacturers, and millions of claims a day, amounts to massive shortfalls. The surprise, she suggests, is that anyone expected a system designed in the 1990s to hold up under 2020s pharmaceutical complexity.
What had to be built underneath
Mandal’s own contribution is within the analytics architecture underneath all of it. As product lead on rebate pricing, she architected the machine learning and natural language processing systems used to forecast rebate flows and structure the unstructured manufacturer contract data those forecasts depend on. ]
The work produced more than $800 million in automation impact at CVS and closed the gap between when a drug was dispensed and when its rebate could be accurately predicted, invoiced, and reconciled. That forecasting and data-structuring infrastructure is what makes transparent rebate accounting calculable in the first place. Without it, the real-time pass-through reporting TrueCost depends on, and the auditability CostVantage promises, would not have been operationally viable, and the multibillion-dollar accuracy gains those products are designed to capture across the U.S. pharmaceutical market would not be reachable at all. The consumer-facing transparency products sit on top of analytics work that had to be solved first. That is the part Mandal owned.
Why this matters for the country
The math scales in a way that makes the headline numbers worth taking seriously. Mandal’s paper estimates that a one-percent improvement in rebate capture accuracy across U.S. insurance markets generates about $4.5 billion in additional rebate value per year. A three-percent improvement, which she suggests is well within reach if transparent systems work as designed, gets to roughly $13.5 billion annually. Those are not abstract savings. They show up in lower premiums, lower copays, and, particularly for low-income patients, the difference between filling a prescription and skipping it.
There is also a market structure question, Mandal flags that does not get enough public attention. The PBM market is concentrated, with three companies controlling about 80 percent of covered lives. We featured Mandal for her paper addressing the risks of an inefficient architecture. The honest read, in her framing, is that the industry waited too long to rebuild. The companies that did the work first, and CVS is one of them, will define the standard the rest of the market eventually has to meet.
What comes next
A few things are likely to happen over the next two to three years, according to Mandal’s analysis.
The first is that more PBMs will roll out their own transparency products, and the quality of those products will vary significantly. Plan sponsors, she advises, should treat transparency claims with the skepticism they deserve and require contractual access to the underlying data.
The second is that point-of-sale rebate delivery, which has been technically feasible but operationally rare, will become more common as the underlying real-time architecture matures.
The third, and the one she watches most closely, is whether AI-driven contract interpretation and rebate prediction tools end up centralized inside the largest PBMs or whether they get built as independent layers plan sponsors can use to verify what their PBM tells them. The market structure question, she argues, gets answered there.
