How do I defend a rebate accrual to auditors?
By defending the estimate, not the arithmetic. A rebate accrual is variable consideration under ASC 606 — it is supposed to be an estimate, and auditors know it. What they test is whether the estimate is governed: which method you used and why, what the inputs were as of booking, who owns the finishing-rate call, how it was revised, and whether the number can be reproduced today with its evidence.
The defense that works is evidence created at calculation time. A binder reconstructed at audit time from spreadsheets and email is not evidence of control — it is evidence of reconstruction.
The five requests to expect
- “Walk me through the estimation method, per program.” Expected value or most-likely amount, chosen deliberately, with the reasoning written down — tiered programs with discrete outcomes usually take most-likely; high-volume portfolios of similar programs take expected value. “The spreadsheet has always done it this way” is not a method.
- “Reproduce this accrual.” The same number, computed today, from the inputs as they stood at booking — with the qualifying base, the rate applied, and the transactions underneath it. This is the request that separates lineage from archaeology; what it takes structurally is covered on rebate audit trail.
- “Show me the estimate’s history.” When the finishing-tier call changed, who changed it, and on what evidence. An estimate buried in a cell has no history; an estimate recorded as a decision does.
- “Decompose the true-up.” Year-end catch-up split into named causes — rate estimate, late data, base drift — not one plug. The mechanics are on rebate true-up.
- “Support the classification.” Why these programs sit in contra-revenue, and fair-value support for anything expensed — the treatment logic at reduction of revenue vs expense.
The three artifacts that end the conversation early
1. The calculation record, frozen at booking. For each accrual: the qualifying base and its definition, the rate applied and why, the transactions in scope, and the approval — captured when the number was made. The worked entries this record sits behind are on the journal entry guide.
2. The estimate memo. One page per material program: the method, the finishing-rate reasoning, the constraint consideration (would a change cause a significant reversal?), the owner, and the revision log. This converts “a number in a cell” into “a documented judgment” — which is what the standard actually requires.
3. The true-up bridge. Last year’s decomposed estimate-to-actual walk. Nothing builds auditor confidence in this year’s estimate like evidence that last year’s was understood cause by cause.
Where defenses fail
- The number no longer exists. The accrual was “fixed” by typing over it, so the figure certified at close cannot be produced, let alone defended. Corrections need to reference the original, not replace it — supersession, not overwrite.
- The base moved between periods. “Qualifying volume” computed slightly differently month to month makes every reproduction attempt disagree with the books.
- The binder is assembled the week the auditors arrive. Reconstruction can produce the right number and still fail the control test, because the control being tested is whether evidence exists before anyone asks.
The self-test to run before they arrive
- Pick one material accrual from last quarter. Can you reproduce it today, exactly, with its inputs?
- Pick one estimate inside it. Is there a written method, owner, and revision history?
- Take last year’s true-up. Does it decompose into named causes?
Three yeses and the audit is a review. Any no is where the finding will be — and fixing it structurally, rather than binder by binder, is what variable consideration audit defense covers. The free assessment scores all three in about three minutes.