Audit fees: review and comparison
No invented percentiles. The comparison is against peers you name — whose fees are public in their proxies — and against your own prior year.
The four captions, both years
Item 9(e) of Schedule 14A requires all four for each of the last two fiscal years — so you need these numbers for the proxy regardless.
| Caption | Current year | Prior year |
|---|---|---|
Audit Fees The annual audit, 10-Q reviews, and services normally provided in connection with statutory and regulatory filings. | ||
Audit-Related Fees Assurance and related services reasonably related to the audit or review, not reported as Audit Fees. | ||
Tax Fees Tax compliance, tax advice, tax planning. | ||
All Other Fees Everything else. The residual category, which is exactly why it gets read. |
Companies you choose, numbers from their proxies
Leave this empty and no comparison is produced — that is deliberate. Three to six comparables, similar revenue and complexity, ideally the same audit firm. Their fee tables are in their proxy statements under the same Item 9(e) captions.
| Company | Audit fees | Revenue |
|---|---|---|
Where the real exposure is
Two things that get missed
There is no public dataset of audit fees to query. This tool supplies the method and the rule content; the peer numbers have to come from peer proxies, which is half an hour of reading and produces a real answer.
A real comparison, or none
Most audit fee benchmarks sell you a percentile. It is the most satisfying possible output and, for most companies asking the question, the least useful — because the distribution behind it is rarely one you can inspect, the peer definition is rarely one you would have chosen, and the answer arrives with a precision the underlying data does not support. This tool does not have such a dataset, and rather than synthesising one it asks you for three to six genuine comparables. Their fee tables are public, in their proxy statements, under the same four captions you use yourself. Reading them takes half an hour and produces a comparison you can defend to an audit committee, which a percentile from an unnamed dataset cannot.
The more consequential half of this page is not about the level of the fee at all. It is the pre-approval machinery, where the exposure is real and the misunderstanding is near-universal. Almost everybody believes the de minimis exception allows non-audit services up to five percent of fees without audit committee approval. The rule has three conditions and the second one governs: the services must not have been recognised as non-audit services at the time of the engagement. It exists for the case where nobody realised what they were buying. Where a service was known to be a non-audit service and approval was simply not obtained, the waiver is unavailable at any amount, and the five percent — which is measured against total fees, not audit fees — never enters the analysis.
The policies themselves are the second common gap. Where pre-approval runs through a standing policy rather than service-by-service approval, the rule requires the policy to be detailed as to the particular service, the committee to be informed of each service, and the policy not to delegate the committee’s responsibilities to management. The usual arrangement — broad categories with spending caps, management judging what fits — fails two of those three. It is also among the cheapest things in governance to put right.
And one requirement that almost nobody knows exists: if more than half the hours on your audit were worked by people who are not the audit firm’s own full-time permanent employees, that percentage has to be disclosed. You cannot work it out; only the firm can tell you. Component auditors, contract staff and offshore delivery centres all count. It is one question to your engagement partner, and much better asked in the autumn than in the week the proxy is being typeset.
What this tool does not do
There is no public dataset of audit fees, so this tool has none and does not pretend to. Every comparison it produces comes from peers you supplied or from your own prior year - if you supply no peers, you get no peer comparison, which is the honest output rather than a gap. The arithmetic on your own numbers is exact. The pre-approval half is rule text applied to your answers, and the answers are yours: whether a service was recognised as a non-audit service at the time of engagement is a fact about what people understood, not something a tool can determine.
- Any industry or peer-group percentile - there is no dataset behind this tool and it will not synthesise one
- Choosing your peers, which is the whole analysis and depends on judgment about comparability
- Whether your fee is reasonable, which depends on scope facts this tool has not seen
- Reading your engagement letter, your fee proposal, or your audit committee minutes
- The full independence rules in 17 CFR 210.2-01 beyond the pre-approval provisions at (c)(7) - prohibited services, partner rotation, employment and business relationships are all separate
- Whether a particular non-audit service is a prohibited service in the first place, which is a different question from whether it was pre-approved
- Negotiating with your auditor, or what a change of firm would cost
- PCAOB inspection results, audit quality indicators, and anything about whether the audit itself was any good
Frequently asked questions
Why does this tool not show me an industry percentile?
Because it does not have an industry dataset, and showing you a percentile without one would mean inventing a number. Audit fees are disclosed company by company in proxy statements under Item 9(e) of Schedule 14A; there is no public API returning a clean peer distribution, and the commercial datasets that exist were built by people reading those proxies. So the comparison here is against peers you name — real numbers you can read in half an hour — and against your own prior year.
Can we use the 5% de minimis exception to skip pre-approval?
Almost certainly not in the way it is usually understood. Rule 2-01(c)(7)(i)(C) has three conditions, and the second is the one that matters: the services must not have been recognised as non-audit services at the time of the engagement. The waiver exists for the case where nobody realised. If you knew it was a non-audit service and simply did not obtain approval, the waiver is unavailable at any amount — the five percent never comes into it.
Five percent of what?
Of the total amount of revenues paid by the audit client to its accountant during the fiscal year — all fee categories, not audit fees alone. And where the waiver is used, Item 9(e)(5)(ii) requires you to disclose the percentage of the Audit-Related, Tax and All Other Fees services approved under it, which is the point: it makes reliance on the waiver visible in the proxy.
Our policy pre-approves tax services up to a cap. Is that enough?
Probably not. Where pre-approval runs through policies rather than service-by-service approval, Rule 2-01(c)(7)(i)(B) requires the policies to be detailed as to the particular service, the audit committee to be informed of each service, and the policies not to delegate the committee's responsibilities to management. A category with a spending cap is not detailed as to the particular service, and a process where management decides what fits inside it is the delegation the rule names.
What is the audit hours disclosure?
Item 9(e)(6) of Schedule 14A: if greater than 50 percent, disclose the percentage of hours on the audit engagement attributed to persons other than the principal accountant's full-time, permanent employees. You cannot derive it — only the firm knows. Component auditors, contractors and offshore delivery centres all count toward it, and most companies have never asked for the figure.
Is there a limit on non-audit fees as a proportion of audit fees?
No rule turns on that ratio, and it is worth being clear that it is not a limit. It is an optic — the first thing a proxy adviser or governance analyst computes from the fee table, because it is the only independence proxy available from public data. Where it is high, the defence is the description of the services, which is required anyway, plus a committee record showing each was considered against independence rather than approved as a category.