What the audit actually hands you.Read it before you share a row.

1. 01

   360 claims were reimbursed below what the drug cost to buy

   $1,658 / mo

   8.6% of the claims in the period, and not the everyday generics. These are high-acquisition-cost items: inhalers, mesalamine, lurasidone, buprenorphine/naloxone, icosapent ethyl. Twenty-three recurring items account for 83% of the loss, which is what makes it fixable rather than random.

   Fix: stand up a weekly MAC appeal routine inside the filing window, and reroute five items on sourcing.

   High
2. 02

   Money is being taken back after adjudication

   $1,057 / mo

   $3,171 of post-adjudication recoupment across the period, spread over 58 claims at $33 to $74 each. These are audit findings and per-claim rate corrections, not routine network fees. Two plans account for 78% of it.

   Fix: price those two plans on net rather than gross before the next contract cycle.

   High
3. 03

   The same 14 generics were bought at materially different prices

   $284 / mo

   Unit acquisition cost varied by more than 15% within a single quarter on high-volume items, with no offsetting change in reimbursement. Pennies a tablet, but multiplied by the units actually dispensed.

   Fix: route the worst five to the best observed source and confirm the wholesaler compliance target.

   Medium

What the whole book earns

A leakage figure means nothing without the book it comes out of. Across the 4,187 claims, reimbursement (ingredient cost paid plus dispensing fee paid plus patient pay amount) exceeded acquisition cost by an average of $13.60 a claim. That is gross margin, and it is a normal figure for an independent of this size. It is not profit. Gross margin minus cost to dispense is net, and once cost to dispense is subtracted what is left is 40 cents a script, about $558 a month.

Which means the honest comparison is the uncomfortable one. The $3,151 a month identified in this report is more than five times what the entire book nets. Recovering it is not an improvement to the profit of this pharmacy. On these numbers it is the profit.

Roughly $526 to $871 per month of this is quick-win work: appeals filing and sourcing routing, two to six weeks of effort. The remainder is structural, meaning plan participation and contract terms, which move on a quarterly cycle rather than a weekly one.

Three hundred and sixty claims in the period returned less than the drug cost to buy. Note what is not on the list below: no amlodipine, no losartan, no sertraline. The everyday generics are where this pharmacy makes its money. The losers are the high-acquisition-cost items, inhalers and specialty generics, where a single fill can cost $47 to $228 to put on the shelf and the PBM pays back less than that.

That distinction is also the arithmetic check on this table. A pharmacy cannot lose more on a fill than it paid for the drug, because reimbursement cannot go below zero. The acquisition-cost column is printed alongside the loss for exactly that reason: every loss below is a fraction of the cost sitting next to it, between 14% and 35% of acquisition. Any report showing a $20 loss on a $4 tablet is describing something that cannot happen.

The seven items below are 44% of the underwater claims and 57% of the total loss, which is what makes the problem tractable rather than diffuse.

MAC appeals: forward rate protection

Twenty-three of the underwater items recur, meaning three or more fills in the period, which is what makes them worth a standing process rather than a one-off email. It is not what makes them appealable. Fill count is not an appealability test. What decides an appeal is the contract and the state statute, and above all the clock: most PBM agreements allow 7 to 30 days from the date of adjudication to submit a MAC appeal, and a number of states set a statutory floor, commonly 14 or 21 days.

Which means the honest framing of a 90-day retrospective is this. Most of the claims analysed here are already outside their appeal window and will not be recovered. A successful MAC appeal usually does not claw back the past fill anyway; it corrects the rate going forward. So the value is not recovery. The value is a weekly appeal routine that catches the next fill inside the window, on the items already known to lose money. The band below applies 25% to 50% to the current monthly run-rate to size what that routine protects each month once it is running, not what can be reclaimed from April.

The reconciliation file shows $3,171 of post-adjudication recoupment across the 90 days, $1,057 a month taken back off claims that were already counted as margin. The shape matters as much as the size. This is 58 claims at $33 to $74 each: audit findings and per-claim rate corrections, which is what genuine post-adjudication activity looks like. It is not a small deduction spread thinly across the whole file, which would be network transaction fees and would belong in the per-claim margin instead. Since 2024 Part D price concessions come off at the point of sale, so what lands here is largely commercial plan activity plus Part D audit recoupment, and the Part D concession pressure shows up earlier, as thinner reimbursement in domain 01. Matching these adjustments back to the claims they hit is what turns a gross margin number into a net one, and it concentrates sharply: two plans account for 78% of the total.

Note what is not in this table: the discount-card book. Those networks settle at the point of sale and take their money as a per-transaction fee, so there is no post-adjudication mechanism for them to claw anything back weeks later. A discount-card book can be badly net-negative, and Plan 4 is, without ever generating a single line here. Any report that attributes retroactive recoupment to a discount-card network has mislabelled a transaction fee.

Contract terms flagged for review

Advisory section. No PBM contracts were supplied with this engagement, so nothing below is calculated. These are the terms worth reading first, based on the plans present in the data and twenty years of sitting on the operator side of these agreements.

Fourteen high-volume generics were purchased at unit costs that varied by more than 15% inside a single quarter, with no corresponding change in reimbursement. These are the profitable items, which is the point: the loss here is not underwater reimbursement, it is paying more than necessary for stock that pays well. Where the low end of the observed range is genuinely repeatable, the difference is avoidable cost.

The exposure column is not an estimate. It is units multiplied by the observed price gap. Gabapentin 300 mg is a 30-day fill of 90 tablets, filled 203 times, so 18,270 tablets moved. The spread between the cheapest and dearest unit cost actually paid in the period was $0.044 less $0.031, or 1.3 cents. 18,270 × $0.013 is $237.51. Every row below is built the same way, and every quantity is the quantity dispensed on the same claims counted everywhere else in this report.

Also flagged, not priced. The operator reported a generic compliance rate of about 81% against an 85% wholesaler target during the kickoff call. Missing that target typically raises the cost of everything bought on the primary agreement, which would sit underneath every figure in the table above. Pricing it needs the invoices and the compliance statement.

Gross margin by payer

Read the column header. Both tables below report gross margin per claim, meaning reimbursement less acquisition cost. That is not net. Cost to dispense, $13.20 a script, has not been taken out of any figure here; it is subtracted once, in the whole-book table in section 01. Gross margin minus cost to dispense equals net.

One book is negative on gross margin while carrying 11.6% of claim volume: Plan 4, a discount-card network the pharmacy is enrolled in through its PSAO. It loses money before cost to dispense is even considered. That is a decision, not an error, but it should be a decision made on this number rather than on the foot traffic it brings through the door, and it is the one book here that can be exited without an access consequence.

The Medicaid MCO line is worth reading twice, and not for the reason most people expect. At $12.75 of gross margin a claim it is neither the disaster of received wisdom nor a windfall. Where the state reimburses ingredient cost at NADAC, the ingredient spread is close to zero by design, so what this book actually pays is the state professional dispensing fee. That fee runs roughly $3 to $15 depending on the state, and sits around $10 to $13 in most of them. It is transparent, it is predictable, and it does not move underneath you the way a MAC list does. This figure is state-dependent and would need to be recomputed against your own state fee schedule. Dropping the book would cost this pharmacy money and cost its patients access. The plan to reconsider is Plan 4.

Gross margin by category

Generics carry this pharmacy. They are 81.5% of claims and 85.5% of the gross margin, at $14.27 a claim. Brand is the reverse: a much larger ticket, $5.15 of gross margin on it, and considerably more capital tied up per unit on the shelf. Note that brand does not cover its own cost to dispense. Compounded work is the highest-yielding line in the file by a distance.

Estimates throughout.Nothing in this section is calculated from the dispensing export, because the activity it describes is not in the export. Each item is sized against the pharmacy’s volume and mix, and carries a low confidence label for that reason.

251 claims (6% of the file) carried a drug description but no NDC, and were matched by normalised name. That is good enough for the margin arithmetic but not for the sourcing comparison in Domain 5, which needs an exact product match, so they were excluded there rather than estimated. If they behave like the rest of the file, the sourcing figure is understated by roughly $18 per month. That estimate is stated here and deliberately not added to the headline.