“Two-person review” sounds like a marketing phrase. The intent of this post is to show, concretely, what it is actually catching — on real filings, in the week before they would otherwise have gone out.
Over the past year we have logged every reviewer catch and every EA/CPA catch, by category. The numbers below are from the first production year; your mileage may vary, but the pattern is stable enough to publish.
The structure
Every filing goes through three distinct people:
- Preparer — writes the filing from intake data. Draft complete.
- Reviewer — runs the full checklist against the draft. Pass or back to preparer.
- EA/CPA approver — professional-level review. Signs the engagement if clean.
The preparer and reviewer cannot be the same person. The reviewer and the approver cannot be the same person. The approver’s name goes on your proof pack.
What the reviewer catches (that the preparer missed)
The reviewer uses a 23-item checklist specific to Form 5472 + pro-forma 1120. Across the first year the reviewer catches, per filing, an average of 1.8 real issues — down from an average of 2.6 at the start of the year as the preparer side has learned.
The top categories:
1. Legal-name mismatches (23% of catches)
The preparer typed the LLC’s legal name from the formation document, not from the CP 575. The reviewer compares against the IRS record and flags any character difference — commas, “LLC” vs “L.L.C.”, trailing periods, case differences in states that preserve case.
2. Owner-paid expenses missed (18% of catches)
The preparer tagged everything on the LLC’s own statements and did not look at the owner’s personal card. The reviewer asks the owner directly during a short review call: “Did you pay any LLC expense from a personal card this year?” The answer is “a few, yes” more than half the time.
3. Netting of contributions and distributions (12% of catches)
The preparer helpfully netted inflows and outflows. The reviewer breaks them back out into gross categories because that is what the IRS actually wants to see.
4. “Loans” with no documentation (9% of catches)
The preparer tagged an inflow as a loan on the owner’s say-so. The reviewer asks for the promissory note and interest rate; when neither exists, the inflow becomes a contribution.
5. Missing pro-forma header (7% of catches)
The pro-forma 1120 was generated without the “Foreign-owned U.S. DE” header. This should be impossible in our template, and over the year the rate dropped from ~12% to under 3% as the template got stricter.
6. Address mismatches (6% of catches)
The address on the pro-forma 1120 is the owner’s home address, not the registered agent’s. The reviewer suggests the registered-agent address and asks the owner to confirm.
7. Statement coverage gaps (6% of catches)
The twelve-month statement set has a month missing or an account closed mid-year with no statement after closure. The reviewer blocks submission until the gap is filled or a staff waiver is issued.
8. “Unclear” tags still present (5% of catches)
A transaction was never confirmed by the owner. Submit is blocked until every tag is confirmed.
9. Tax-year mismatch (3% of catches)
A fiscal-year LLC had the tax year on the pro-forma 1120 set to the calendar year by mistake. Rare, but ugly when it happens.
10. Service-fee timing (3% of catches)
An owner service fee was billed in one year and paid in another. The reviewer confirms the cash-vs-accrual method on the LLC’s books and makes sure the fee lands in the correct year.
The remaining ~8% is a long tail of individual-case issues.
What the EA/CPA catches (that the reviewer missed)
The reviewer catches checklist items. The EA/CPA catches judgement items — things that are correct on the mechanical checklist but should be done differently given professional context. Across the first year, EA/CPA catches per filing average 0.4 — which is to say, most filings pass clean after the reviewer stage.
The top categories:
1. Classification nudge (37% of catches)
The owner relocated to a US-residency jurisdiction partway through the year. The reviewer treated them as fully non-US; the EA/CPA caught that the US-residency fact pattern needs a part-year analysis and may push the LLC out of disregarded-entity treatment.
2. Related-party identification (21% of catches)
The reviewer caught the obvious counterparties. The EA/CPA caught a transaction with the owner’s spouse that was tagged as “unrelated third party” and should have been “related party by attribution”.
3. Treaty position (14% of catches)
A transaction is reportable, but a US tax treaty with the owner’s country changes how the withholding or sourcing works. The reviewer does not know treaty analysis; the EA/CPA does.
4. Prior-year consistency (12% of catches)
The current-year 5472 is correct in isolation, but it does not line up with how a prior year was prepared (by another firm, or by the owner directly). The EA/CPA flags the inconsistency and either reconciles it or adds a clarifying statement.
5. Writing clarity on “other” statements (9% of catches)
Where an “other” transaction tag was used, the supporting written statement is accurate but ambiguous. The EA/CPA rewrites it for clarity so the IRS reader can tell at a glance what the transaction was.
The remaining ~7% is scattered individual-case judgement calls.
Why one person cannot do this
A natural reaction, especially from careful founders: “Can’t I just be very careful myself?”
You can be very careful. Many of our customers are. “Very careful” does not scale against the specific categories above because:
- Legal-name mismatches feel invisible when you look at your own document. You know what the name is “supposed” to say.
- Owner-paid expenses are psychologically invisible: you tagged them as “personal” in your mind months ago and will not re-flag them without a direct prompt.
- Netting feels tidier and more defensible. You will not catch it by looking harder.
- Loan-vs-contribution judgement is literally a different person’s job. The owner knows the business context; a credentialed professional knows the regulatory framework.
Review is not about effort. It is about having a different pair of eyes with a different checklist. The reviewer is not better than the preparer; they are a different person, running against a different list, with different incentives.
What this means for self-filers
If you are preparing your own 5472 for a disregarded LLC, the single most useful thing you can do before you fax is: hand it to a credentialed person — a CPA, an EA, or at a minimum a thoughtful fellow founder — and ask them to run the top ten errors (from the previous blog post) against your draft.
If they can read your draft without finding at least one issue, you have found an unusually careful self-filer. More commonly, they will find one or two things to fix. That is the value of review in a sentence.
What this means for us
The headline point of this post is not “look how good our review is”. It is this specific list of catches is the floor. Any filing service that is not catching these at review stage is not reviewing.
If you are evaluating BorderFile against another service, this is a good question to ask them: “What does your review process catch, concretely, and can you show me?” If the answer is a marketing phrase instead of a numbered list, that tells you something.
In one sentence
Two-person review is a specific, numbered, repeatable list of catches; it works because it uses a different pair of eyes against a different checklist, not because the reviewers are smarter than you.
Questions on how we structure review internally? Email support@borderfile.io.