# Methodology: The 18-month VP of Sales is a myth: the median is 32 months

By Cristian Frunze, Founder of Ken · Oct 12, 2026 · [Ken Research](/research) No. 7 · Sales teams, Hiring and org design, Founders and startups

The full methodology behind [The 18-month VP of Sales is a myth: the median is 32 months](/research/vp-of-sales-tenure.md): the data, every method, every number and every limitation. The same text as the [methodology PDF](/research/vp-of-sales-tenure.pdf).

Methodology and full results.

## 1. Question and pre-registration

What is the median tenure of VPs of Sales, CROs and Heads of Sales, and how does company growth move in the quarters before and after one leaves? The hypotheses, plan, headline metric and the results that would disprove each hypothesis were pre-registered on 2026-10-16, before any data was pulled. In short, we expected:

- H1: a median tenure of 24 to 36 months, longer than the "18 months" folklore, and shorter at small young companies than at companies with 200+ employees. Disproved if the median was 21 months or less.
- H2: slower growth than matched companies in the 4 quarters before the exit.
- H3: no recovery within 4 quarters after the exit.
- H4: a median of 3 months or more before a successor starts, with more than 25% of seats not refilled within 12 months.

The headline metric was the Kaplan-Meier median tenure of the primary sample at companies with 10 or more employees. Every judgment call made after pre-registration was logged in the study's decisions log.

## 2. Data

**Ken internal data.** Ken's dataset of public professional profiles, work histories and company records. The people data holds profiles whose current country is the United States, United Kingdom, Canada, Australia, Ireland or New Zealand; earlier roles can be anywhere. Each role has a title, a company identifier (on about 73% of roles), and month-precision start and end dates. Profiles were last refreshed between 2025-11 and 2026-04. For this study we used two extracts built for several Ken Research studies:

- A sales and marketing careers extract: every role of every person with at least one sales or marketing title, with normalized titles (136,156,211 roles, 23,543,766 people).
- A company headcount history: quarterly headcount per company and function from 2015 Q1 to 2026 Q1, rebuilt by counting the people whose role at the company covers each quarter end, plus company records (industry, founded year).

A one-month list of leadership changes (June 2026) was used as a coverage check.

No campaign or customer data from Ken was used. Only aggregate tables leave Ken's servers, and every group smaller than 50 people is suppressed.

## 3. Sample

**Seat titles.** We classified each normalized title with a fixed set of rules:

- VP: vp, vice president, svp, evp, senior or executive vice president, group VP, combined with "sales" or "revenue".
- CRO: chief revenue officer, or "cro" as the whole title or joined with "and".
- CSO: chief sales officer, chief sales and marketing officer.
- Head of Sales: head of sales, head of global or worldwide sales, sales head, head of revenue.
- CCO (chief commercial officer) is reported separately and is not in the primary sample.

Titles naming a region, segment, channel or sub-function (for example regional, EMEA, enterprise, inside, institutional, sales operations, revenue operations, sales development) were left out of the primary sample. Titles naming a founder, owner, CEO, president (not vice president), managing director, general manager, interim, fractional, acting, deputy, assistant or associate role, or a finance "revenue" job (revenue cycle, revenue management), were excluded. The title rules were checked against every title held by 50 or more people before they were frozen.

**Roles to seats.** 862,231 candidate roles matched a seat family. We dropped roles with no company identifier (218,561), no start date (41,628), or a past role with no end date (41,585). A person's back-to-back or overlapping qualifying roles at the same company (gap of one month or less) were merged into one seat, so a promotion from VP of Sales to CRO counts as one seat. Seats starting 2015-01 to 2025-12 were kept: 222,165 primary seats held by 190,652 people.

**Censoring.** A seat still held today is censored at the month of the person's last profile update, the last date we know they were still in it. 103 roles with an end date after the last update were treated as current. Tenure in months is end month minus start month plus one.

**Headline sample.** Seats at companies with 10 or more rebuilt employees in the quarter the seat started: 166,654 seats, 143,761 people, 105,967 exits.

## 4. Methods

**Tenure.** Kaplan-Meier survival curves of months in the seat, with Greenwood 95% confidence intervals for the median. Cuts: company size at the start (10-49, 50-199, 200-999, 1,000+), company age at the start (start year minus founded year: under 5, 5-9, 10-19, 20+, unknown), title family, industry, start cohort, country, first versus repeat sales leader. A Cox proportional hazards model on the headline sample with size, age, title family, start cohort and a US indicator.

**Growth around the exit (event study with matched comparison).**

- *Exits.* We used sole-leader exits: a finished primary seat at a company with no other person's primary seat that started 6 or more months before the exit and was still active at it, at companies with 20 or fewer primary seats since 2015 (larger companies have no single top seat). We took each company's first such exit between 2017 Q1 and 2024 Q4, required 20+ employees in the exit quarter and headcount in every quarter from 8 before to 4 after. 32,219 companies qualified.
- *Controls.* For each exit quarter, companies that had a primary seat holder in place at the end of quarter t-4 and no primary-seat exit from t-8 to t+8, with the same headcount requirements.
- *Matching.* Coarsened exact matching on exit quarter, industry, size band at t-4 (20-49, 50-99, 100-199, 200-499, 500-999, 1,000+), age band, and quintile of headcount growth from t-8 to t-4. Up to five controls per company were drawn at random with replacement within the cell. 22,054 companies were matched (10,165 had no match), with 74,995 control draws covering 17,536 distinct companies.
- *Outcome.* Log of one plus headcount, relative to t-4, for total headcount and for sales and marketing headcount. The departing leader is removed from the treated company's headcount in the quarters they held the seat, so their own exit is not a mechanical drop.
- *Estimate.* The mean difference between each treated company and the mean of its controls at each quarter from t-8 to t+4, with 95% intervals from 1,000 bootstrap draws over treated companies, reported as percent differences.
- *Robustness (added after the main result).* Controls restricted to companies that lose their own sales leader 9 to 16 quarters later; treated companies with 50+ employees; headcount without removing the leader.

**Replacement.** For each sole-leader exit, the first start of another person's primary seat at the company at or after six months before the exit. Overlaps count as zero months. Exits with no successor are censored at 2026-04. Kaplan-Meier, by company size at the exit.

**Cross-check.** Sales-leader exits in the June 2026 list of leadership changes were matched to the seat table by person and company.

## 5. Results

### 5.1 Tenure

| Group | People | Median months (95% CI) | Still in seat at 18 months |
|---|---|---|---|
| Headline: companies with 10+ employees | 143,761 | 32 (32-32) | 72.4% |
| All company sizes | 190,652 | 33 (33-34) | 72.5% |
| 10-49 employees | 62,650 | 30 (30-30) | 69.2% |
| 50-199 employees | 40,356 | 31 (31-32) | 72.1% |
| 200-999 employees | 26,459 | 33 (33-34) | 74.2% |
| 1,000+ employees | 26,943 | 37 (37-37) | 78.8% |
| Under 5 years old | 13,380 | 24 (24-25) | 61.7% |
| 5-9 years | 19,245 | 25 (25-25) | 63.6% |
| 10-19 years | 24,923 | 29 (29-30) | 69.3% |
| 20+ years | 47,512 | 41 (40-41) | 78.3% |
| 10-49 employees and under 5 years | 9,979 | 24 (24-24) | 60.9% |
| VP of Sales | 102,652 | 35 (35-36) | 74.7% |
| CRO | 14,675 | 26 (26-27) | 68.2% |
| Chief Sales Officer | 3,288 | 33 (32-35) | 74.1% |
| Head of Sales | 28,863 | 26 (26-27) | 66.0% |
| Software development (industry) | 17,977 | 25 (25-25) | 65.7% |

Further results (all company sizes unless noted): completed seats only, 22 months (119,123 people); start cohorts 2015-2017: 36, 2018-2020: 35, 2021-2023: 30 months; first top sales seat 35 versus repeat leaders 27 months; US 35, UK 29, Canada 36, Australia 28 months; chief commercial officers 31 months. Excluding banking, finance and insurance from the headline sample gives 31 months; using only profiles updated in April 2026 gives 33; treating every title change as a new seat gives 32.

Cox model hazard ratios (headline sample, against 200-999 employees, 10-19 years, VP title, 2015-2017 cohort, outside the US; higher means leaving sooner): under 5 years 1.29 (1.26-1.32), 5-9 years 1.20 (1.18-1.23), 20+ years 0.73 (0.72-0.75), CRO 1.25 (1.22-1.27), Head of Sales 1.36 (1.33-1.38), 1,000+ employees 0.93 (0.91-0.95), 10-49 employees 0.97 (0.95-0.99), 2021-2023 cohort 1.08 (1.06-1.10). Concordance 0.58.

### 5.2 Growth around the exit

| Quarter vs exit | Total headcount gap (95% CI) | Sales and marketing gap (95% CI) |
|---|---|---|
| t-8 | -0.9% (-1.2% to -0.6%) | 1.0% (0.5% to 1.4%) |
| t-1 | -0.2% (-0.5% to 0.1%) | -0.4% (-0.8% to 0.1%) |
| t0 (exit quarter) | -1.7% (-2.0% to -1.3%) | -1.6% (-2.2% to -1.1%) |
| t+1 | -4.0% (-4.5% to -3.6%) | -4.2% (-4.8% to -3.5%) |
| t+2 | -6.2% (-6.7% to -5.7%) | -6.2% (-6.9% to -5.6%) |
| t+4 | -10.2% (-10.7% to -9.6%) | -10.1% (-10.9% to -9.3%) |

The value at t-4 is zero by construction. Treated companies grew 6.4% from t-4 to t0 and were 2.2% above t-4 at t+4; matched companies grew 8.3% and 13.8%. The gap widened by 0.090 log points (0.086 to 0.094) from t0 to t+4. Across four random seeds for the control draws, t+4 ranges from -10.20% to -10.14%. Robustness at t+4: controls that lose their own leader later, -12.8% (12,857 companies; t-1 gap -1.0%); 50+ employees, -8.9% (11,625 companies); leader not removed from headcount, -11.7%.

### 5.3 Replacement

Of 93,122 sole-leader exits (83,519 people), 19.0% had a visible successor within 12 months and the median was not reached. At companies with 50+ employees at the exit (29,909 exits), 22.2% had a successor within 3 months, 33.1% within 12 months and 41.5% within 24 months (median 45 months); at 1,000+ employees, 39.3% within 12 months (median 24 months). 8.2% of all sole-leader exits had a successor who started before or in the exit month.

### 5.4 Cross-check

Of 3,752 sales-leader exits in the June 2026 list of leadership changes, 91.3% of the people are in our seat data and 76.2% at the same company.

### 5.5 Hypotheses

- H1: supported. Median 32 months; small young companies 24 months against 36 at 200+ employees.
- H2: supported by the pre-registered test (the t-4 to t0 gap is -1.7% and its interval excludes zero), but nearly all of it falls in the exit quarter. In the four quarters before that quarter, growth matched peers. We describe this as no sign of a slowdown before the exit.
- H3: supported. The gap widened from -1.7% to -10.2%.
- H4: formally supported, but the replacement measure undercounts successors (see limitations), so we do not present it as a finding.

## 6. Verification

Every headline number was recomputed by an independent re-analysis, written from the study plan and definitions alone without seeing our code. All eight headline numbers matched: median tenure 32 months (143,761 people), 72.4% still in the seat at 18 months, 24 and 36 months for small young and 200+ companies, 35 / 26 / 26 months for VP, CRO and Head of Sales, 22 months for completed seats, 25 months in software, and 33.1% of exits at 50+ employee companies replaced within 12 months. The event study matched within the noise of the random control draws: the independent code gave -0.17%, -1.65% and -10.23% at t-1, t0 and t+4 (21,998 companies). Across four random seeds the study's own estimates move by up to 0.06 percentage points and the independent code's by up to 0.18, so event-study effects should be read to the nearest whole percent. Two definitions in the first verification brief were ambiguous and were clarified before the final comparison. The independent check also found that labelling a merged seat by its latest title instead of its first gives CRO tenure of 29 months instead of 26, because some VPs of Sales are promoted to CRO in the same seat.

Alternative explanations considered:

- *Struggling companies push out their sales leader (reverse causality).* If so, growth should slow before the exit. It does not in the main comparison; with later-exit controls there is a small dip one quarter before (-1.0%). We cannot rule out problems that show up in revenue before headcount.
- *Healthy companies keep their leaders.* Controls are chosen because their leader stayed, which may favor healthy companies. Controls that lose their own leader later give a larger gap (-12.8%), so this selection does not explain the result.
- *The exit itself lowers headcount.* We removed the leader from the treated counts; without that adjustment the gap is -11.7%.
- *Company decline or shutdown.* If a company is winding down, the leader leaves along with everyone else. This would produce the same pattern, and the design cannot separate it.
- *Acquisitions.* When a company is acquired, staff often move to the buyer's company page, which reads as falling headcount and may coincide with the sales leader leaving.
- *Short jobs missing from profiles.* This pushes tenure up. The completed-only median (22 months) and the cohort trend show the direction of the bias, not its size.

## 7. Limitations

- Ken internal data is self-reported. People leave short jobs off their profiles, add new jobs late and sometimes edit an old title instead of adding a new role. Deleted profiles are missing, so the earliest years undercount people who left the workforce.
- The people data covers six English-speaking countries, by current country. Headcount misses staff elsewhere; companies are compared with each other rather than read as absolute sizes.
- About 27% of roles have no company identifier and are dropped.
- Dates are month precision and cluster at whole years (visible as steps at 12, 24 and 36 months in the survival curve).
- Founded year is missing for about a third of headline seats; those form an "unknown" age band.
- Title rules work best on English titles.
- The event study is a matched comparison, so it shows what is linked with an exit, not what the exit causes. A third of eligible companies had no exact match and are not in the comparison.
- Replacement counts are lower bounds, since a successor is only seen if their profile is in the data with the company identifier and a top sales title.
- The list of leadership changes covers one month, after the profile snapshot, so it checks coverage, not timing.
- The 2024-2025 start cohort is too recent for its median to be reached.

## 8. Sources

- Ken internal data: Ken's dataset of public professional profiles, work histories and company records (people and company records refreshed 2025-11 to 2026-04).