Say a CPA firm wants to know if its tax reviews are actually getting better. Three numbers are sitting right there in the workflow already: how long a return takes to get signed off, how many times it bounces back to the preparer before it's clean, and how often the same mistake shows up again on a different return. All three sound like good ways to measure review efficiency. All three can also be faked — not by lying, just by a reviewer changing how they work instead of how well they review. Here's exactly how, for each one, and what to track instead.
Efficiency is one part of the whole process
See where cycle time and review passes fit inside the full tax return review sequence.
None of these three numbers measures how careful a reviewer actually was. Each is a stand-in for that — useful right up until someone starts getting judged on the number itself instead of the review behind it.
Cycle time
Cycle time is simple: the time between a preparer finishing a return and a reviewer signing off on it. Say a return sits for three days before it's approved — that's a three-day cycle time. Firms like tracking it because a rising number is a real warning sign that something's stuck, somewhere.
Here's how it gets gamed. A reviewer under pressure to shrink that number doesn't read faster — they move the starting line. They mark a finished return "on hold, waiting on the client" when it isn't, or they don't start the clock until they actually open the file, even if it sat untouched in their queue for two days first. The number drops. IRS Circular 230 still requires the same diligence either way, and nothing about how carefully the return got checked has actually changed.
Want the honest range for review time?
See what the IRS's own numbers say — and what they don't.
Review passes per return
This counts how many times a return goes back to the preparer for fixes before it's approved. Say a reviewer finds three mistakes — a missing signature, an expense coded to the wrong line, a number that doesn't match last year's return. They send it back once with a list. That's one pass.
Here's the trick: instead of sending it back, the reviewer can just walk over to the preparer's desk and say "fix these three things" out loud. The preparer fixes them right there. Nothing gets logged as a review round. On paper, that return looks like it cleared in one clean pass — same as a return that genuinely had nothing wrong with it. The count can't tell the two apart, and a firm chasing a lower average now has an easy way to get one.
Recurring-comment rate
This tracks how often the same kind of mistake shows up again — on a different return, a different preparer, next season. Say five returns this year are all missing the same attachment for a mileage deduction. That's not five unrelated mistakes. That's one broken step in the process, showing up five times.
The AICPA — the accounting profession's own standards body — says firms should trace a repeat mistake back to its root cause instead of just re-flagging it each time, per guidance in the Journal of Accountancy: fix the checklist or template that's letting it through, not just the one return in front of you.
Pull a sample of finished reviews
Check a handful of returns that already got signed off. What did the review actually catch?
Spot the repeat
Group similar mistakes together, even if two reviewers wrote them up differently.
Find why it keeps happening
Trace it to a template or checklist gap — not to whichever preparer happened to make it this time.
Fix the template, not the return
Once the checklist is fixed, that mistake has nowhere left to recur.
This one gets gamed through wording, not workarounds. If every reviewer writes comments in their own words, "forgot Form 8283" and "missing donation paperwork" look like two separate issues, even though they're the same mistake. Nobody's lying, nobody tagged it the same way twice — but the rate drops on paper even though the real mistake keeps happening.
What these numbers are really measuring
This isn't unique to tax review. A call center that grades agents purely on "average call time" ends up with agents who hang up fast, not agents who actually solve the problem. Same idea here: cycle time, passes, and recurring comments are all useful signals — right up until someone's bonus depends on moving just the number, not the review the number was supposed to stand for.
A short cycle time tells you a reviewer signed off fast. It doesn't tell you they were right to.
- Track all three numbers together, not one at a time — a good cycle time next to a rising pass count is a red flag, not a win.
- Keep a record of what a pass or comment actually was, not just a count of it — the count is exactly the part that's easy to fake.
The real problem isn't picking the right formula. It's that most firms only keep the count — not a record of what a pass actually found, or what a comment actually said. Pernee keeps that record attached to the return itself, so a number on a dashboard can always be checked against what actually happened.
