For one quarter, I stopped quoting CEO turnover statistics and started counting them myself. I pulled every Item 5.02 filing from the Russell 3000, logged the announcement date, the effective date, whether a successor was named, and whatever reason the company put in writing. It took eleven evenings. I finished with 84 verified chief executive changes for the quarter and a considerably lower opinion of the figure I had been repeating in meetings for two years.
Here is the verdict, stated plainly: the headline executive turnover numbers in circulation measure different universes of companies under different definitions of "departure," and the gap between them is wide enough to justify or defer a succession decision depending on which one you happened to read first.
One disclosure, once: I have no commercial relationship with any data provider named below. I paid for one subscription out of pocket and canceled it.
What eighty-four exits looked like up close
The aggregate number is the least interesting thing in my spreadsheet. What the categories did to it is the interesting part.
Eleven of the 84 were interim executives being confirmed as permanent. Those show up in most counts as a CEO change. From a board's perspective they are the opposite — the resolution of an instability that started a year earlier. Strip them out and the quarter drops to 73.
Thirty-one were orderly: a named successor, an effective date at least 60 days out, an internal candidate already running a division. Six were the other thing entirely — effective immediately, no successor named, a lead director listed as the contact. Nine were founders stepping back, which is a different phenomenon wearing the same label; a founder handing over after a growth round and a retail CEO removed after two soft quarters do not belong in the same row.
Five changed their story. The stated reason in the initial filing was resignation or a pursuit of other interests; a later filing, sometimes a proxy statement months on, reframed it as a termination. I only caught those because I went back. Most trackers publish once and move on.
The median gap between announcement and effective date was 34 days. Nineteen of the 84 were announced on a Friday after the close. I do not know whether that means anything. I noticed it, and I have not been able to unnotice it.
The denominator nobody prints
A count is not a rate. The number of public companies in the United States is substantially smaller than it was in the late 1990s, and the composition has shifted toward larger firms. A flat count of departures against a shrinking universe is a rising rate. A falling count against a shrinking universe may be a flat rate. Very few of the figures that reach a boardroom carry a denominator, which means the year-over-year comparison you are being handed is doing arithmetic you cannot inspect.
This is also why a triple-digit monthly exit figure and my 84-per-quarter tally are not in conflict. The broad monthly trackers count announced CEO exits across all US employers — private companies, hospital systems, universities, nonprofits, government agencies. My count covered roughly 3,000 listed companies. Both numbers are correct. They are answering different questions, and they get quoted interchangeably.
Three definitions, three different worlds
Does a departure count when it is announced, or when it takes effect? A CEO who announces in November and leaves in March lands in one year or the next depending on the convention, and conventions differ between sources.
Does an interim appointment count as one event or two? Most broad counts book the exit and then book the permanent hire, producing two data points from one vacancy.
Does a chairman-and-CEO who relinquishes the CEO title and keeps the chair count as turnover? Some datasets say yes. If your question is "has effective control changed," the honest answer in many of those cases is no.
None of these are errors. They are choices. The problem is that the choices are usually documented in a methodology note that nobody reads and the number travels without it.
How I'd actually decide which source to trust
Five criteria, in the order I weight them: the universe (which companies, and can you get the denominator), the definition (announced or effective, and how interims are handled), cadence (monthly noise or annual signal), revision behavior (does anyone go back when a resignation is later disclosed as a termination), and cost.
| Source | Universe | Cadence | Best at | Honest negative |
|---|---|---|---|---|
| Broad monthly exit trackers (e.g. Challenger, Gray & Christmas) | All US employers, incl. public sector and nonprofits | Monthly | Fast directional read on the whole labor market | No usable denominator; mixes a 40-person nonprofit with a large-cap; rarely revised |
| Index-scoped succession research (e.g. The Conference Board with ESGAUGE) | Russell 3000 and S&P 500 | Quarterly and annual | Rates, not counts; genuine year-over-year comparability | Lags by months, which makes it useless for a live decision |
| Annual S&P 500 transition studies (e.g. Spencer Stuart) | Large-cap only | Annual | Depth on successor profile, tenure, internal vs external | The 500 largest firms are not the market your mid-cap client competes with for talent |
| Raw 8-K Item 5.02 pulls from EDGAR | Any filer you choose | Real time | You control every definition; catches the reclassifications | Labor-intensive; I spent eleven evenings on one quarter; language is deliberately opaque |
| Commercial executive databases (e.g. Equilar and peers) | Configurable, broad | Continuous | Linking departures to comp, tenure, board composition | Four to five figures a year as of writing; you are still buying someone else's definition |
If you need one answer: for a board discussion about whether your own turnover is abnormal, use index-scoped rate data and accept the lag. For a journalist tracking the labor market this month, the broad monthly count is the right tool as long as you print the universe alongside it. For an executive search professional pricing a market, nothing beats a filtered 8-K pull of your actual competitive set, and it is the only method on this list that catches the story changing after the fact.
Who should act on this, and who shouldn't
Acting on aggregate turnover data makes sense if you are calibrating: setting retention packages, benchmarking your CEO's tenure against a comparable cohort, or advising a board that believes it is an outlier when it is at the median.
It does not make sense as a timing signal for your own succession. A cooling market-wide exit rate tells you what several thousand other boards did last quarter. It tells you nothing about whether your chief executive has stopped being the right person for the next three years. I have watched a board defer a change it had already decided on because a report said turnover was elevated and they did not want to look reactive. That is aggregate data being used as cover for a decision that was individual all along.
Back to the spreadsheet
The number I wanted from my 84 was the one I could not produce: how many of those boards knew a year earlier that the change was coming, and waited.
Six abrupt exits with no successor named suggest at least some boards were surprised. Thirty-one orderly ones suggest planning. The rest sit in the middle, and no filing will ever tell me which side they were on.
Which leaves the question the data cannot close. When a board defers a succession it privately believes is necessary, does that cost show up later — in a worse transition, a shorter successor tenure, a scramble on a Friday afternoon — or does it never show up at all, because half the time the deferral turns out to have been right? The counterfactual is not observable, and I have not seen a dataset that gets near it. Does anyone actually know?
