# Methodology: Do new sales and marketing leaders bring their old team?

By Cristian Frunze, Founder of Kenso · Nov 5, 2026 · [Kenso Research](/research) No. 19 · Hiring and org design, Sales teams

The full methodology behind [Do new sales and marketing leaders bring their old team?](/research/leaders-bring-their-people.md): the data, every method, every number and every limitation. The same text as the [methodology PDF](/research/leaders-bring-their-people.pdf).

Question: when a new sales or marketing leader joins a company (VP, Head, CRO, CMO), what share of the next 12 months of hires into their function come from the leader's previous employer, compared with a placebo base rate?

## Summary

- 120,628 of 8,672,527 external hires (1.39%) in a new leader's first 12 months came from the leader's previous employer (452,932 leader joins at 222,376 companies, 2016-2025).
- Same company, the 24 months before (Placebo A): 0.78% expected vs 1.29% observed on the same 293,340 joins. Lift 1.65 (95% interval 1.56-1.74).
- Same-size companies, same months (Placebo B, the pre-registered headline comparison): 0.01%. Lift 100.02 (81.63-124.26). This comparison is too weak to be read as the effect; see Methods.
- Affinity-matched difference-in-differences (Placebo C, added after the first results): 2.47% observed vs 0.52% expected on 18,883 joins. Lift 4.77 (4.36-5.31).
- The pre-registered hypothesis H1 (about 4% of hires, lift at least 3x over the same company) is not supported: the lift over the same company is 1.65, under the 2x line the plan set as proof against it.

## Data

Kenso internal data: Kenso's dataset of public professional profiles, work histories and company records. This study reads Kenso's shared sales-careers extract (built for the Kenso Research study *Who are GTM engineers, and where do they come from?* on 2026-10-15): every experience spell (title, normalized function and level, company id, month-precision start and end dates, current flag) of the 23.5M people with at least one sales or marketing title. All 136,156,211 spells were read. No new data was pulled. Profiles were last refreshed by 2026-04-29; the snapshot month is 2026-04.

Coverage: people whose current profile is in the US, UK, Canada, Australia, Ireland or New Zealand. Earlier jobs can be anywhere. Leader joins by the leader's current country: US 320,611; UK 80,683; Canada 24,521; Australia 20,717; Ireland 3,967; New Zealand 2,433.

## Definitions

- **Function families.** Sales: SDR/BDR, account executive, account management, general sales, sales engineering, partnerships and business development, enablement, revenue operations, customer success, GTM engineering. Marketing: marketing, growth marketing, demand generation, marketing operations. Function and level come from the shared taxonomy (ordered regex rules on the normalized title).
- **Leader title.** Sales or marketing family and level VP or C-level (CRO, CMO, chief customer officer, president), or level director with the word "head" in the title. Plain "Director" titles and assistant vice president (AVP) titles are excluded.
- **External entry.** A spell at a company id where the person has no spell at that company with an earlier start month. One entry per person, company and family (the earliest; a leader title wins a tie in that month).
- **Previous employer.** For any entry, the company of the most recent spell that started in an earlier month at a different company (ties: longer spell, then the one listed first on the profile). A spell with no company id makes the previous employer unknown.
- **Leader join (event).** An external entry with a leader title, start month 2016-01 to 2025-04 (so the 12-month window closes before the snapshot and the 24-month pre-period starts in 2014), a known previous employer L that differs from the new company C.
- **Hires.** External entries into the same family at C with start month t to t+11 (t = the leader's start month), the leader excluded. Shares use hires with a known previous employer.
- **Mass transfers.** 1,685 company pairs had 25 or more people move from one to the other in a single month (mergers, rebrands, outsourcing). They were removed from events and from every placebo pool.
- **Function-size band.** People with a same-family spell at the company covering the month: under 5, 5-9, 10-19, 20-49, 50-99, 100-249, 250-999, 1000 or more. Team-size groups: under 20, 20-99, 100 or more.

## Sample flow

| Step | Count |
|---|---|
| External entries into sales or marketing spells | 26,969,591 |
| ... with a leader title | 1,056,811 |
| ... starting 2016-01 to 2025-04 | 563,195 |
| ... with a known previous employer other than the new company | 457,961 |
| ... after removing mass-transfer pairs (events) | 452,932 |
| Hires into the function in the 12-month windows (all / known previous employer) | 10,508,204 / 8,672,527 |
| Leader joins with at least one hire with a known previous employer | 285,474 |

## Methods

1. **Observed.** For each join, hires in the window and how many came from L. Pooled share = sum of hires from L / sum of hires with a known previous employer. Pooling weights each join by its hires, so large companies weigh more; the team-size cuts show the typical join.
2. **Placebo A (same company).** Two placebo dates at the same company and family, t-24 and t-12 months, each with a 12-month window (together months t-24 to t-1). Expected hires from L = the join's hires x the company's share of hires from L in that period. Joins with no hires in the period (159,592) drop out of this comparison.
3. **Placebo B (matched companies).** The same 12 calendar months at all companies in the same family-size band, counting only company-months with no leader join in that family within 12 months either side, and excluding L's own hires. Expected hires from L = the join's hires x the pool's share of hires from L. Because the match is only on team size, most pool companies have no tie to L, so this base rate (0.01%) is close to the chance that any given hire anywhere came from L. It was the pre-registered headline comparison and is reported as such, but it is not a fair bar for the effect of a leader.
4. **Placebo C (affinity-matched difference-in-differences, post-hoc).** Joins whose company hired at least one person from L in the 24 months before (18,883 with usable matches). Matches: other companies in the same family-size band at t that also hired at least one person from L in those 24 months (clean company-months only). Expected share for the join = the company's own pre-period share x (matches' post share / matches' pre share). Matches' pooled share from L fell from 1.21% before to 0.33% after (both sides were picked for having hired from L, so both regress toward the mean); the leaders' companies went from 2.17% to 2.47%.
5. **Lift** = sum of observed hires from L / sum of expected. 95% intervals: bootstrap over companies (1,000 resamples, seed 54), placebo rates held fixed.
6. **Overlap (H2).** A hire from L overlapped with the leader if their spells at L shared at least one month.
7. **Any-hire comparison.** Share of joins with at least one hire from L, against its expected value 1 - (1 - p)^hires under Placebo A and B.

## Results

**Headline**

| Measure | Value |
|---|---|
| Hires from the leader's previous employer, first 12 months | 120,628 of 8,672,527 (1.39%) |
| Placebo A share / observed on A subset / lift | 0.78% / 1.29% / 1.65 (1.56-1.74) |
| Placebo B share / lift | 0.01% / 100.02 (81.63-124.26) |
| Placebo C observed / expected / lift | 2.47% / 0.52% / 4.77 (4.36-5.31) |
| Excess hires from L per 100 joins (vs A / vs B) | 14.89 / 26.37 |
| Joins with at least one hire from L | 51,105 (11.28%); expected 6.53% (A), 0.18% (B) |
| Hires from L who overlapped with the leader at L | 114,386 (94.83%), vs 80.85% of Placebo B pool hires from L |
| External hires per leader join (known previous employer) | 19.15 |

**By months since the join**

| Months after the join | Hires (known prev.) | From prev. employer | Share % |
|---|---|---|---|
| 0-2 | 2,324,858 | 56,727 | 2.44 |
| 3-5 | 2,140,364 | 25,800 | 1.21 |
| 6-8 | 2,101,825 | 21,040 | 1.0 |
| 9-11 | 2,105,480 | 17,061 | 0.81 |

**Splits** (Lift C is blank where Placebo C was not computed; join-year rows carry no intervals)

| Group | Leader joins | Hires (known prev.) | From prev. employer | Share % | A subset observed % | Placebo A % | Lift A (95% CI) | Placebo B % | Lift B | Lift C |
|---|---|---|---|---|---|---|---|---|---|---|
| function: sales | 265,020 | 6,597,987 | 94,475 | 1.43 | 1.35 | 0.85 | 1.6 (1.51-1.69) | 0.01 | 118.9 | 4.72 |
| function: marketing | 187,912 | 2,074,540 | 26,153 | 1.26 | 1.09 | 0.58 | 1.89 (1.73-2.09) | 0.02 | 63.56 | 5.25 |
| leader level: vp | 224,985 | 4,399,718 | 73,286 | 1.67 | 1.57 | 0.93 | 1.7 (1.61-1.79) | 0.01 | 140.88 | 5.06 |
| leader level: head | 161,486 | 3,662,785 | 36,628 | 1.0 | 0.91 | 0.65 | 1.41 (1.31-1.55) | 0.02 | 60.89 | 4.28 |
| leader level: C-level | 66,461 | 610,024 | 10,714 | 1.76 | 1.51 | 0.55 | 2.76 (2.41-3.27) | 0.01 | 127.09 | 4.97 |
| team size at join: under 20 | 330,455 | 524,249 | 32,143 | 6.13 | 5.01 | 1.89 | 2.65 (2.53-2.78) | 0.0 | 1345.48 | 9.46 |
| team size at join: 20-99 | 69,742 | 936,888 | 28,130 | 3.0 | 2.99 | 1.52 | 1.97 (1.83-2.12) | 0.01 | 372.12 | 7.89 |
| team size at join: 100 or more | 52,735 | 7,211,390 | 60,355 | 0.84 | 0.84 | 0.62 | 1.36 (1.29-1.44) | 0.02 | 54.54 | 4.25 |
| prev employer size: top 100 previous employers | 20,569 | 1,273,114 | 25,727 | 2.02 | 2.0 | 1.72 | 1.16 (1.09-1.25) | 0.07 | 29.71 | 3.32 |
| prev employer size: all other previous employers | 432,363 | 7,399,413 | 94,901 | 1.28 | 1.17 | 0.62 | 1.88 (1.81-1.96) | 0.0 | 279.1 | 7.23 |
| prev employer size: excluding top 20 | 445,652 | 8,172,598 | 109,218 | 1.34 | 1.23 | 0.7 | 1.75 (1.67-1.82) | 0.01 | 156.11 | 5.66 |
| join year: 2016 | 32,362 | 700,924 | 9,063 | 1.29 | 1.22 | 0.81 | 1.51 | 0.02 | 81.41 | 4.4 |
| join year: 2017 | 35,376 | 751,428 | 8,430 | 1.12 | 1.05 | 0.72 | 1.45 | 0.02 | 73.9 | 2.91 |
| join year: 2018 | 39,830 | 880,044 | 10,491 | 1.19 | 1.11 | 0.78 | 1.42 | 0.01 | 99.13 | 4.62 |
| join year: 2019 | 44,244 | 951,878 | 12,040 | 1.26 | 1.19 | 0.74 | 1.61 | 0.01 | 97.99 | 4.88 |
| join year: 2020 | 44,108 | 1,000,887 | 12,763 | 1.28 | 1.18 | 0.72 | 1.65 | 0.01 | 116.19 | 6.61 |
| join year: 2021 | 56,278 | 1,654,767 | 20,118 | 1.22 | 1.15 | 0.77 | 1.49 | 0.01 | 85.14 | 6.64 |
| join year: 2022 | 59,886 | 1,160,859 | 15,522 | 1.34 | 1.25 | 0.78 | 1.6 | 0.02 | 74.93 | 4.32 |
| join year: 2023 | 57,585 | 615,659 | 12,712 | 2.06 | 1.89 | 0.83 | 2.28 | 0.01 | 139.85 | 5.18 |
| join year: 2024 | 60,670 | 704,259 | 13,991 | 1.99 | 1.76 | 0.88 | 2.01 | 0.01 | 174.98 | 4.52 |
| join year: 2025 | 22,593 | 251,822 | 5,498 | 2.18 | 1.91 | 0.95 | 2.01 | 0.01 | 198.11 | 3.65 |

**Robustness**

| Group | Leader joins | Hires (known prev.) | From prev. employer | Share % | A subset observed % | Placebo A % | Lift A (95% CI) | Placebo B % | Lift B | Lift C |
|---|---|---|---|---|---|---|---|---|---|---|
| main | 452,932 | 8,672,527 | 120,628 | 1.39 | 1.29 | 0.78 | 1.65 (1.56-1.74) | 0.01 | 100.02 | 4.77 |
| drop hires in the join month | 452,932 | 7,833,317 | 85,362 | 1.09 | 1.05 | 0.77 | 1.37 (1.32-1.43) | 0.01 | 78.67 | |
| leader stayed 12 months or more | 383,095 | 7,846,344 | 107,805 | 1.37 | 1.28 | 0.78 | 1.63 (1.55-1.74) | 0.01 | 97.39 | 4.73 |
| first leader join per company and function in 24 months | 283,889 | 663,321 | 31,296 | 4.72 | 3.7 | 1.28 | 2.9 (2.75-3.06) | 0.01 | 514.82 | 6.26 |
| function under 500 people at the join | 434,900 | 3,188,662 | 88,504 | 2.78 | 2.53 | 1.29 | 1.96 (1.88-2.04) | 0.01 | 290.15 | 7.03 |
| US leaders, US hires, US pool | 320,611 | 5,466,666 | 86,322 | 1.58 | 1.46 | 0.85 | 1.72 (1.63-1.82) | 0.02 | 90.42 | |
| previous employer = job that ended 0-12 months before | 392,936 | 7,130,043 | 102,436 | 1.44 | 1.33 | 0.8 | 1.65 (1.56-1.75) | 0.02 | 91.75 | |

**Most common previous employers of new leaders** (company names from the most common spelling on profiles; Pavilion and Chief are membership communities that people list as concurrent roles, see Limitations)

| Previous employer | Leader joins | People in data ever there | Hires (known prev.) | From prev. employer | Share % | Placebo B % | Lift B |
|---|---|---|---|---|---|---|---|
| Salesforce | 1,079 | 52,163 | 95,830 | 3,227 | 3.37 | 0.03 | 108.85 |
| Google | 1,024 | 69,111 | 104,251 | 2,488 | 2.39 | 0.05 | 50.04 |
| Oracle | 988 | 71,078 | 68,546 | 2,331 | 3.4 | 0.03 | 109.23 |
| Amazon | 855 | 134,627 | 31,986 | 765 | 2.39 | 0.32 | 7.57 |
| IBM | 854 | 111,227 | 83,817 | 1,790 | 2.14 | 0.05 | 39.05 |
| Microsoft | 830 | 88,530 | 109,463 | 2,345 | 2.14 | 0.06 | 36.51 |
| Pavilion | 752 | 5,834 | 15,002 | <50 | | 0.0 | |
| Amazon Web Services (AWS) | 603 | 35,558 | 54,700 | 702 | 1.28 | 0.03 | 41.82 |
| Cisco | 577 | 48,272 | 71,720 | 1,174 | 1.64 | 0.02 | 66.35 |
| Wells Fargo | 563 | 109,721 | 16,265 | 578 | 3.55 | 0.07 | 48.34 |
| Chief | 562 | 5,370 | 14,486 | <50 | | 0.0 | |
| JPMorgan Chase & Co. | 470 | 72,129 | 25,652 | 721 | 2.81 | 0.05 | 54.63 |
| Accenture | 458 | 66,997 | 28,657 | 271 | 0.95 | 0.03 | 36.23 |
| SAP | 450 | 24,397 | 51,084 | 664 | 1.3 | 0.01 | 181.13 |
| Dell Technologies | 426 | 72,815 | 48,693 | 1,285 | 2.64 | 0.04 | 63.18 |

## Hypotheses

1. **H1, not supported.** Expected about 4% of hires, at least 4x over Placebo B and 3x over Placebo A. Observed 1.39%, lift B 100.02, lift A 1.65. The plan said a lift under 2x against either placebo proves H1 wrong, and lift A is under 2x.
2. **H2, supported on the registered test, weak test.** 94.83% of hires from L overlapped with the leader at L, more than half and more than the 80.85% for other companies' hires from L. Leaders often spent years at L, so most of L's staff overlap with them; the test cannot separate personal ties from shared employment.
3. **H3, supported.** Lift A for teams under 20 is higher than for teams of 100 or more (splits table), and lift B and C keep the same order.
4. **H4, supported on the registered test.** Lift B for leaders from the 100 largest previous employers is above 2x. Against the same company's history (lift A) it is the weakest cut, close to 1.

## Decisions and deviations

Every judgment call is logged with its reason in the study's decision log. The deviations from the plan were: Placebo C and the any-hire expected values were added after the first results; a robustness check with the previous employer defined as the job that ended 0-12 months before the move was added after the top-employer table showed membership communities. Placebo A and B are reported exactly as registered.

## Verification

**Recomputation (2026-10-16).** An independent re-analysis, written from the study plan without seeing our code, got only the pre-registered plan, the exact definitions fixed before results (including Placebo C), the shared sales careers extract and a list of numbers to recompute, without their values. It had no access to the study's working files, outputs, code or logs, wrote its own code and ran it on all 96 parts of the extract.

| Number | Study | Verifier | Match |
|---|---|---|---|
| Leader joins; companies | 452,932; 222,376 | 452,932; 222,376 | yes |
| Hires with known previous employer; from L; share | 8,672,527; 120,628; 1.39% | 8,672,527; 120,628; 1.39% | yes |
| Placebo A: joins; observed on subset; expected; lift | 293,340; 1.29%; 0.78%; 1.65 | 293,340; 1.29%; 0.78%; 1.65 | yes |
| Placebo B: expected share; lift | 0.01%; 100.02 | 0.01%; 100.02 | yes |
| Placebo C: joins; observed; expected; lift | 18,883; 2.47%; 0.52%; 4.77 | 18,883; 2.47%; 0.52%; 4.77 | yes |
| Joins with at least one hire from L; expected under A | 11.28%; 6.53% | 11.28%; 6.53% | yes |
| Share by months 0-2; 9-11 | 2.44%; 0.81% | 2.44%; 0.81% | yes |
| Team under 20: share; Placebo A on subset (observed; lift) | 6.13%; 1.89% (5.01%; 2.65) | 6.13%; 1.89% (5.01%; 2.65) | yes |
| Overlap share | 94.83% | 94.83% | yes |

Every headline number matched to the reported precision on the first pass. The verifier's notes were interpretation choices that coincide with the study's code: the "<5" band includes companies with 0 people in the function at t; clean company-months use every leader-title entry; unknown-end overlaps count as no overlap (there were none). It did not recompute the bootstrap intervals (not in the list). Separately, running the study's results step twice now gives byte-identical outputs.

**Alternative explanations**

1. **The two companies were already linked** (same industry, city or customer base), so hiring from L would rise anyway. Placebo A holds the company fixed and Placebo C holds the existing pipeline fixed; the effect survives both (1.65x and 4.77x). It is the reason the same-size Placebo B (100x) is not read as the effect.
2. **Big employers are everyone's previous employer.** For leaders from the 100 largest previous employers the lift over the company's own history is only 1.16, so the effect comes mostly from smaller previous employers (1.88). Named in the report.
3. **Time trends in where companies hire from.** Placebo A has none; Placebo C does (match companies over the same months). Every join year 2016-2025 has lift A between 1.42 and 2.28.
4. **Mergers, rebrands and acqui-hires** move many people from L to C at once. Pairs with 25 or more moves in one month are removed everywhere; remaining smaller acqui-hires are a limitation.
5. **The hires brought the leader, or both were recruited together.** The design cannot tell who moved whom. Same-month moves account for much of the first-quarter peak (dropping them: lift A 1.37). Named as a limitation; the report says "linked with", not "caused".
6. **Concurrent roles misread as the previous employer** (memberships like Pavilion and Chief, board seats). Using the job that ended 0-12 months before gives the same lift A (1.65).
7. **Big companies dominate the pooled share and count the same hires for several leaders.** The first-leader-per-24-months check (lift A 2.90) and team-size cuts show the effect is larger, not smaller, where one leader owns the function.
8. **Self-reported, backfilled profiles** (people add jobs late, omit short ones). Affects observed and placebo periods alike; the last join year (2025) shows the same pattern. Limitation.

## Limitations

- **Observational design.** This is a before-after comparison with matched comparison groups, not an experiment. The results show hiring from the leader's old company is linked with the leader's arrival and is concentrated right after it. They do not prove the leader caused each hire; some hires may have been recruited by the same pull that brought the leader, or may have brought the leader.
- **Placebo A has no time trend.** It compares the same company with its own past. If hiring from a given employer drifts up or down over time for other reasons, the lift moves with it. Placebo C addresses this, but covers only the 18,883 of 452,932 joins whose company already had a hiring pipeline from L.
- **Placebo B is too easy a bar.** It matched only on team size; most companies in the pool have no link to L. Its 100x lift should not be quoted as the effect.
- **Pooling weights big companies.** A VP joining a large company is credited with every hire into the function there, and large companies have many VP joins, so the same hires can count for several leaders. The team-size cuts and the first-join-in-24-months check show the effect is larger where one leader clearly owns the function.
- **Same-month moves.** Hires who started in the leader's start month account for much of the first-quarter peak. Some are likely moves agreed together with the leader; dates are month precision, so their order cannot be told.
- **Self-reported, incomplete histories.** Profiles are self-reported. People leave short jobs off, add new jobs months late and edit old titles. Only currently active profiles exist, so older years undercount. Spells with no company id make a previous employer unknown: of 10,508,204 hires, 8,672,527 have a known previous employer.
- **Previous employer rule.** The previous employer is the most recent earlier-starting job at another company, which can be a concurrent role, board seat or membership community (Pavilion and Chief appear among the most common). The check that uses the job that ended 0-12 months before gives the same lift A.
- **Company ids are not merged.** Companies that merged or changed records can appear under two ids. The mass-transfer rule removes the clearest cases.
- **Title-based levels.** VP, Head and C-level come from the title only. Bank and insurance VP titles are often not leadership roles; AVP is excluded, but other inflated titles remain.
- **Coverage.** Six countries by the person's current profile, English titles classified best.
- **Suppression.** Any count of people under 50 is shown as "<50" with its percentage blanked.

## Sources

- Kenso internal data (Kenso's dataset of public professional profiles, work histories and company records), through the shared sales-careers extract built for the GTM engineer study.