CFO Pay Percentile: What It Means and How It Is Calculated

Your peer table is built, and the percentile still turns on three choices

You have spent an evening pulling DEF 14A filings from EDGAR and keying CFO rows into a spreadsheet. The compensation committee wants your review conversation prepared by the end of the month. The table has twelve rows, your own figures sit at the bottom, and a cell somewhere reads "62nd percentile." Before that number goes into any document, three questions need answers. Which peers are in the denominator? Which pay measure was compared? Which proxy years were pooled?

A percentile feels like a fact about you. It is really a fact about a particular peer set, a particular measure and a particular vintage, and changing any one of them can move the result without any change in your pay. Most of the confusion around a cfo pay percentile comes from leaving those three choices unstated.

This article defines the term once, works through the arithmetic with clearly labelled sample figures, and shows how each choice shifts the result. It then explains why a percentile should be withheld when the peer set is too small, and how to report it by pay component rather than as one number.

What a CFO pay percentile means, defined once

Throughout this article, a percentile is the share of peer CFO totals at or below your figure. If seven of ten peer totals are at or below yours, you sit at the 70th percentile. The definition is inclusive: a peer who matches you exactly counts as at or below.

Other tools and textbooks use different conventions. Some count only peers strictly below you, and some interpolate between ranked values. Two tools can therefore report different percentiles for the same data and both be internally consistent. When you read or report a percentile, state the convention beside it.

The median is the 50th percentile, the middle value of the ranked peer totals. The quartiles are the 25th and 75th percentiles. Published studies report them as bands, and Equilar's 2023 study of S&P 500 CFOs reported a median total reported compensation of $4.7 million in both 2021 and 2022. Over 2022 it put the 25th percentile at $3.4M, down from $3.5M, and the 75th at $6.8M, down from $7.0M. That is a statement about S&P 500 companies in two specific years under Equilar's own method. It is not a statement about every public-company CFO, and it says nothing about your peer set. Treat any published "cfo median pay" figure the same way: attribute it, keep its universe attached, and do not drop it into your own table as a stand-in for peers you have not read.

How a CFO pay percentile is calculated, step by step

Here is the arithmetic. The figures below are SAMPLE figures invented to demonstrate the method. They are not real pay and not drawn from any filing.

Worked example (SAMPLE). Take ten peer CFO SCT totals, in thousands of dollars, ranked low to high: 1,900; 2,300; 2,600; 2,900; 3,100; 3,400; 3,800; 4,200; 4,900; 5,600.

1. Fix the measure. All ten are SCT totals, meaning the Summary Compensation Table total with equity at grant-date value. Your own figure must be the same measure.
2. Rank the peers. The list above is already ranked.
3. Count the peers at or below your figure. Suppose your SAMPLE total is 3,800. The values at or below it are 1,900, 2,300, 2,600, 2,900, 3,100, 3,400 and 3,800, which is seven.
4. Divide by the number of peers. Seven of ten gives 70 percent. Under the at-or-below definition you sit at the 70th percentile of this peer set.

Now change one input. If your SAMPLE total were 3,500, six peers would be at or below it and the result would be the 60th percentile. The step from one position to the next is ten percentage points, because each of ten peers carries a tenth of the distribution. With a larger peer set the steps shrink; with a smaller one they grow. A percentile reported to the decimal on ten peers is false precision.

Two practical points follow. First, the denominator is the number of peers whose CFO row you actually found and could use, not the number of companies in your intended peer group. Second, a duplicate employer, such as the same company appearing in three proxy years, is not three independent peers, a point the next section returns to.

Why the peer set, the measure and the proxy year each move the result

The peer set. Revenue band, sector and region all change who is in the denominator. A set drawn from a narrow revenue band answers a different question from one drawn across a sector. The same SAMPLE total can sit near the median in one set and well toward the top of another. Peer selection is its own discipline, covered in how to choose an executive compensation peer group, and the percentile only inherits whatever choices were made there. Size matters as well, and CFO pay by company size goes into why a revenue band is a common first cut.

The measure. The SCT total reports stock and option awards at aggregate grant-date fair value, a rule in place since the SEC's 2009 amendments. It is a grant-date measure, not realized pay. Compensation actually paid under Item 402(v) reflects changes in equity value during the year, so it is a different number for the same executive. The pension and non-qualified deferred compensation column can also move a total without any pay decision behind it. A percentile computed on SCT totals cannot be compared with one computed on another measure. The difference between the measures is laid out in grant-date fair value versus realized pay.

The proxy year. A proxy filed in 2025 typically reports the fiscal year that ended in 2024, and companies with off-calendar fiscal years break that simple offset. Item 402(c) requires the SCT to show each named executive officer's compensation for the last three completed fiscal years. A smaller reporting company shows two. So each filing can supply up to three data points per executive, and you must decide whether to compare against the latest year only or to pool. Pooling gives more observations but mixes vintages, and it can count one employer several times unless you handle that explicitly. Whichever you choose, record it next to the result.

Who is disclosed. Under Item 402(a)(3), a non-SRC filer's named executive officers include everyone who served as principal financial officer, so its CFO appears whatever their pay rank. A smaller reporting company's named executive officers are its principal executive officer and its two most highly compensated other executive officers, so its CFO may not appear at all. Emerging growth companies may use scaled disclosure too. A peer set made up largely of such filers will have fewer usable CFO rows, and the denominator shrinks accordingly. Coverage, meaning how many peers in the set actually disclose a CFO, belongs beside every percentile for that reason.

Why a percentile should be withheld below five distinct employers, and how to report it by component

CFOCompensation is designed to withhold a percentile when the peer set has fewer than five distinct employers. That is a design rule, not a statistical law, and it rests on the arithmetic above. With five peers, each peer represents twenty percentage points, so your position can only land on a handful of values, and one added or removed peer can swing the answer by a full step. Counting distinct employers rather than rows matters because several proxy years from one company are not independent observations. A figure printed under those conditions would look more exact than the data supports, so the app is built to show nothing instead.

The same logic argues for reporting by component. A single overall percentile can hide the fact that your base sits in one place in the distribution and your equity grant value in another. The app is designed to calculate percentiles overall and by component (base, bonus including non-equity incentive, equity grant value, and other pay), across the latest three available proxy years. Reading them side by side tells you which part of the pay mix drives the overall position. For equity, remember what is being compared: grant-date value at award, not what ever vests or is realized.

Two further design rules support the result. Former-CFO rows and rows flagged during review are excluded by default, so an interim or departed executive's partial-year figure does not distort the ranking. And SCT CFO totals are designed to be cross-checked against the Item 402(v) average non-PEO named executive officer total. This is a consistency rule that flags rows that do not reconcile. It is not an audit, and it cannot confirm that a figure is right, only that it is not obviously inconsistent with the same filing.

Where the hand-built version breaks, and what the catalogue covers

A spreadsheet plus EDGAR can replicate the core of this. The costs are in five places. You have to identify the CFO row by title, which may read "Chief Financial Officer," "EVP & CFO" or "Principal Financial Officer," with interim and former CFOs appearing as well. You have to normalise the columns across filers. You have to cite each figure to its filing, which a hand-built table rarely does. You have to cross-check for keying errors, and you have to check coverage so you know how many peers named a CFO at all. None of this is difficult. All of it is slow, and it tends to land in the week before a conversation about your own pay.

The catalogue is built around that checklist. The CFO Annual Pay Review Prep Workbook lays out the peer table, the percentile arithmetic above and the coverage count in a form you fill in yourself, with a column for the source filing on each row. If you want the same assembly handled inside an app, CFOCompensation is being built to do it, and you can join the waitlist to hear when it opens. For the wider process around a review, see how to prepare for a CFO annual pay review.

Nothing here is an instruction on what your percentile should be or what to say once you have it. The method organizes cited comparables so you arrive with a clear account of what they are and what they are not.

About these figures: CFOCompensation works from disclosed figures only, shown as filed and cited to the source filing; any third-party survey or study is named and labelled as such, including the Equilar study cited above. The worked example uses SAMPLE figures, which are not real pay. Nothing in this article is legal, tax or negotiation advice. CFOCompensation and Rovaryn Digital Inc. are not affiliated with the SEC or the IRS. Private-company pay is not covered.