You Can Be Exactly Right About the Wrong Number

Gross-to-Net teams are measured on rebate accuracy, and most of them are very good at it. The trouble is that accuracy is a statement about arithmetic, not about whether the number being calculated was the right one to begin with.

Every control in a rebate process validates the same thing: that the math is correct on the units you have. Nothing in that process asks whether the units you have are the units you should have had. A prescription stopped at prior authorization never becomes a claim, never enters the eligible population, and therefore never appears in any reconciliation. The calculation stays clean. The number it produces is quietly wrong, and it is wrong in a direction nobody in finance is looking.

Accuracy is not the same as completeness

Rebate accuracy is a precision measure. It asks whether the contract terms were read correctly, whether the rate was applied to the right base, whether eligibility was determined properly, whether the accrual was trued up against actuals. Those are real disciplines and they deserve the attention they get.

Completeness is a different question entirely. It asks whether the population being measured reflects the demand that actually existed. In a mature process those two questions feel like the same question, because the population is whatever the claims data says it is, and the claims data is the only version of events anyone has.

What a prior authorization actually does to the number

The scale here is not marginal. The American Medical Association’s 2025 physician survey, released in May 2026, found that 95 percent of physicians say prior authorization delays access to necessary care and 79 percent say patients abandon treatment because of authorization challenges. Twenty-six percent reported that prior authorization led to a serious adverse event. Physicians complete an average of 40 prior authorizations a week and spend about 13 hours on them.

Read those numbers as a finance problem rather than an administrative one. Every abandonment in that 79 percent is a unit that was demanded, prescribed and never dispensed. It never enters the rebate calculation, because there is no claim. It never shows up as leakage, because leakage is measured against units that exist.

Three financial echoes of one denial

A denial does not produce one outcome. It produces three, and each lands somewhere different in the P&L.

The first outcome is survivable. The patient waits, the therapy starts late, and the quarter absorbs a timing variance that gets explained in a footnote. The second and third are the expensive ones, because both get attributed to something they are not. A volume miss reads as soft demand. A mix shift reads as competitive pressure. Neither explanation mentions authorization, so neither generates a fix.

Why finance never sees it

This is a visibility problem with a specific shape. Finance receives claims, chargebacks, rebate submissions and utilization files. Every one of those artifacts is generated downstream of a successful dispense. The failure happens upstream, at a counter, between a pharmacist, a payer system and a patient, and it produces a rejection code that lands in somebody else’s data set.

So the information exists. It is sitting in switch data, hub data and pharmacy reject reporting. It is simply not routed to the people building the accrual, and it is not part of the close calendar. That is an organizational choice rather than a technical limitation.

What to measure alongside accuracy

None of this argues for less rigor on rebate accuracy. It argues for a second measure standing next to it, so the precision has something to be precise about.

1. Reject and reversal volume by reason code. The raw count of prescriptions that hit a wall, separated by why. This is the denominator that rebate accuracy is silently assuming.
2. Conversion from approved to filled. An approval is not a dispense. The gap between the two is where the second and third outcomes above live.
3. Time from first attempt to first fill. Delay is the leading indicator of abandonment, and it is measurable long before the abandonment shows up as a volume miss.
4. Reconciliation of forecast units to attempted units. Not to dispensed units. Forecasts are built on patients who need the therapy, so the honest comparison is against how many tried.

Put those four next to the accuracy metrics and the variance conversation changes shape. It stops being a debate about whether the accrual was right and becomes a question about where demand went.

The same discipline, one step earlier

Gross-to-Net is the financial echo of how well the enterprise executed the patient journey. The enterprise develops, manufactures, prices, contracts, distributes and supports. It does not diagnose, prescribe, dispense or treat. What it can do is refuse to let a measurement system stop at the point where its own data happens to begin.

Rebate accuracy asks whether we paid the right amount on what we sold. The harder and more useful question is whether what we sold reflects what patients were actually prescribed. The first question is closed and well governed. The second one is where the money is.

Survey figures are the published findings of the American Medical Association. The operating patterns described reflect my own experience and are offered as commentary, not as legal, regulatory, clinical or accounting advice. The views expressed are my own and are not made on behalf of any current or former employer. All trademarks belong to their respective owners.