What Is Review by Exception for CPA Firms?

Review by exception means a reviewer gives most of a return no second look at all. Only what's flagged — a number that moved further than a similar return usually does, a schedule that's new this year, a line that disagrees with a source document — gets read closely. Everything else inherits whatever scrutiny an earlier pass already gave it. That isn't a shortcut CPA firms invented under deadline pressure. It's how the IRS itself decides which returns to look at.

The IRS runs on exceptions too

Every individual return, and some corporate ones, gets a numeric score from the Discriminant Function System after it's processed — a model built from what similar returns have looked like historically. "If your return is selected because of a high score under the DIF system, the potential is high that an examination of your return will result in a change to your income tax liability," the IRS says in Publication 556. The agency doesn't examine every return at the same depth for the same reason no firm reviews every return the same way: there aren't the staff-hours, and most returns don't need it.

The IRS decides what to skip using a model checked against its own results. Most firms skip on instinct, or on whatever a vendor shipped as a default.

When the scope is defensible

Circular 230 already contemplates a reviewer who doesn't personally re-verify everything. 31 CFR §10.22(b) presumes a practitioner exercised due diligence when relying on someone else's work product, provided reasonable care went into engaging, supervising, training and evaluating that person. Applied to review, that's the exact shape of review by exception: a reviewer who reads less isn't automatically undiligent, if what they're skipping is work already checked by a preparer the firm has real grounds to trust — not just a title.

That presumption reaches the preparer's work. It doesn't reach the client's. For information the client furnished, 31 CFR §10.34(d) sets a narrower rule.

A practitioner may not, however, ignore the implications of information furnished to, or actually known by, the practitioner, and must make reasonable inquiries if the information as furnished appears to be incorrect, inconsistent with an important fact or another factual assumption, or incomplete.

— 31 CFR § 10.34(d)

Put together, the defensible version of review by exception rests on two conditions, and most firms have never checked either one on purpose. The reliance on the preparer has to be earned by real supervision and training, not assumed because the return came from someone senior enough. And the criteria deciding what counts as an exception have to be built to catch what would appear incorrect, inconsistent or incomplete on its face — not just built to minimize how much a reviewer has to read.

When it's a diligence failure wearing an efficiency label

Consider a firm whose exception rule flags any line that moved more than a set percentage from the prior year. A client's meals-and-entertainment expense stayed essentially flat two seasons running — not because nothing changed, but because a bookkeeper has been coding a category of vendor bills to the same account regardless of what they actually are, and the total happens to land close to last year's by coincidence. Nothing about the number looks unusual. What's wrong isn't a number that moved. It's a number that should have moved and didn't, and a rule built to catch movement has no way to catch its absence.

§10.22(a) doesn't measure diligence by how much of a return got read. It measures diligence by whether the practitioner exercised due diligence in determining the correctness of representations made to the IRS — and under IRC §6694, that duty sits with the practitioner who signs, not with the software that built the exception rule. A scope that reads less can still be diligent, if what it reads less of is exactly the part someone accountable already checked. A scope that reads less because nobody has revisited whether the checked part is still the right part isn't efficient. It's undocumented.

The honest version of review by exception isn't a rule that reads less. It's a rule that can say, in writing, what it's built to catch and what it isn't — and gets checked against what a return that slipped through actually looked like. Most firms have the first part. Almost none keep the second: a record of which exceptions the current rule would have missed, revisited as a firm's return mix changes rather than left at whatever a review-software vendor shipped as a default three tax seasons ago. That record has to live in the system doing the flagging, not in someone's memory of what last season's exceptions were — otherwise it disappears exactly like the judgment it was supposed to check. Pernee keeps a flag's reasoning and the returns that didn't trip one in the same place, so a scope decision can be checked against what it actually missed, not just defended by how much time it saved.