An accrual is a claim about behavior that has not happened yet. Most of them are built as a claim about arithmetic instead.
This field guide gives Finance four diagnostics, the assumption inventory, the lag map, the denominator test and the single change test, so every Gross-to-Net rate carries documented behavioral assumptions with a named owner. Skip it and the cost is not the variance. It is the credibility lost explaining variances after the fact.
What this guide is for
You are standing up a Gross-to-Net accrual, or you have inherited one and cannot explain why it moves the way it does. This guide is the set of questions I would ask in the room, in the order I would ask them.
It is not a rate methodology. Rates are the output. Everything below is about the inputs the rate is quietly asserting.
The failure it addresses
An accrual is usually built by taking historical claims, dividing by historical sales, and calling the result a rate. That arithmetic is correct and the reasoning underneath it is not, because it assumes next period’s patients will behave like last period’s patients, through a channel that has not changed, under coverage that has not moved.
None of those three assumptions is safe, and none of them is written down anywhere. So when the accrual misses, the conversation becomes an argument about the rate rather than an examination of the assumption that produced it. The rate is not the thing that broke.
The deeper version of the same problem is that an accrual can only be built from what was recorded. A patient who left at Exit 05 or Exit 07 never generated a claim, never generated a rejection, and never appeared in the denominator. The accrual is therefore accurate about the population it can see and silent about the population it cannot, and the silence does not look like a gap. It looks like a smaller number.
The use cases you can run
The assumption inventory
Write down every behavioral assumption the accrual depends on, in plain sentences, before touching a rate. Who is eligible. What share enrolls. What share is approved. What share initiates. What share persists, and for how long. Each sentence is a claim someone in the organization owns, and most of them are owned outside Finance.
The lag map
For each contract type, establish how long it takes a claim to arrive, and whether that duration is stable or drifting. Rebates, chargebacks, copay, fees and returns all lag differently, and a single blended lag hides the one that is moving.
The denominator test
Take one rate and ask what population sits underneath it. Then ask which patients could have been in that population but never appeared. This is where Exit 04 through Exit 07 enter the accrual conversation, usually for the first time.
The single change test
Take the one thing you know is changing next period, a formulary move, a new hub vendor, a copay program redesign, a channel shift, and trace it through the assumption inventory. If it touches nothing in the model, the model is not sensitive to reality.
What good looks like
Every rate has a written assumption behind it, a named owner outside Finance, and a stated lag. The model can represent a change you already know about. And the accrual review is a conversation about behavior, with Finance holding the pen and other functions holding the assumptions.
What it costs to skip it
Not the miss. The miss is survivable and gets trued up. What it costs is the year you spend arguing about a rate while the thing that moved it was a coverage change nobody told Finance about, and the credibility Finance loses each time it explains a variance after the fact rather than before.
Where this connects
The exits that feed this accrual are on the philosophy page. Where they land inside the enterprise is the 5 Layers of GTN™. What happens when the actuals arrive is the next guide, True-Up Governance. The patient-side view of the same ten exits is at Where did your treatment stop?
Former employers and products referenced are matters of professional history. The views expressed are my own and are not made on behalf of any current or former employer, and no commercial outcome for any named product is attributed to me personally. All trademarks belong to their respective owners.