Search Fund Study Comparability: Check the Sample

Search fund study comparability starts with the risk that similar-looking headlines describe different populations, periods and measurements. Before comparing results, identify who entered each sample, what the reported metric measures and which outcomes were excluded. A comparison that leaves those details out can turn useful research into an unsupported expectation about an investor's experience.

The mean and median tell different stories

Separate study panels, not a comparable series: Stanford’s 862 core funds show 4.75x aggregate ROIC and 33.9% pre-tax IRR; median not reported here. Yale’s 12 investors, 23 funds and 768 allocation decisions show company-level mean MOIC 2.78x and median 1.60x excluding broken searches, versus median 1.0x including them. IESE international funds show aggregate ROI 2.0x, IRR 18.1%, and median 1.4x.

Stanford GSB 2026 Search Fund Study

862 core search funds raised since 1984 · As of December 31, 2025

33.9% aggregate pre-tax IRR

■ Aggregate ROIC: 4.75x

Median not reported here.

Yale SOM · How are Search Fund Investors Really Faring?

12 investors · 23 funds · 768 allocation decisions

Company-level MOIC · Lazier, Thomas, Wasserstein

■ Mean · excluding broken searches: 2.78x

○ Median · excluding broken searches: 1.60x

○ Median · including broken searches: 1.0x

Broken-search inclusion changes the median.

IESE 2024 International Search Fund Study

International search funds · Sample size not reported here

18.1% aggregate IRR

■ Aggregate ROI: 2.0x

○ Median: 1.4x

Separate international study population.

Source: Stanford GSB 2026 Search Fund Study via ClearlyAcquired; core funds since 1984, as of December 31, 2025¹. Lazier, Thomas, Wasserstein, “How are Search Fund Investors Really Faring?”, Yale SOM (October 27, 2025)². IESE 2024 International Search Fund Study via Search Fund Insider (April 9, 2026)³. Study figures for the asset class, not the performance of any fund, network or firm.

Figure data

The mean and median tell different stories

Study	Population	IRR	Measure	Value
Stanford GSB 2026 Search Fund Study	862 core search funds raised since 1984 · As of December 31, 2025	33.9%	Aggregate ROIC	4.75x
Yale SOM · How are Search Fund Investors Really Faring?	12 investors · 23 funds · 768 allocation decisions	—	Mean · excluding broken searches	2.78x
Yale SOM · How are Search Fund Investors Really Faring?	12 investors · 23 funds · 768 allocation decisions	—	Median · excluding broken searches	1.60x
Yale SOM · How are Search Fund Investors Really Faring?	12 investors · 23 funds · 768 allocation decisions	—	Median · including broken searches	1.0x
IESE 2024 International Search Fund Study	International search funds · Sample size not reported here	18.1%	Aggregate ROI	2.0x
IESE 2024 International Search Fund Study	International search funds · Sample size not reported here	18.1%	Median	1.4x

Search fund study comparability starts with scope

A study can answer its research question well while leaving your question unanswered. Research on acquired companies can describe those companies without describing every search that began. An aggregate can describe a combined result without describing the experience of a typical company or limited partner. Put the research question beside the headline before deciding what the finding establishes.

For accredited investors, family offices and LPs, this distinction matters when research moves into a discussion about an investment. A published asset-class result does not become evidence about a particular investment simply because both involve search funds. The sample and the claim must remain connected as the information moves from the source to your reading notes.

John Steinberg discusses the Stanford research in The SMB Investor podcast episode Inside the Search Fund Playbook. The qualitative boundary Steinberg describes is that evidence about the search fund model belongs to the asset class, rather than to any particular investment organization. That boundary is useful even when you leave the episode's numerical claims out, as this article does.

Keep that distinction visible throughout the comparison. The academic findings discussed here are asset-class or study-sample observations. They are not SMB Investor Network results, and past industry performance does not guarantee future results.

Define the population

Start by writing down what qualifies an observation for inclusion. Does the research follow searches, acquired businesses, investment funds or investors? These labels identify different things. An acquired-business sample begins after a search has found a company; a search sample may also contain attempts that never reached that point.

Next, record the geographic scope using the source's own language. Do not replace a label such as international with a more specific regional definition unless the source supports it. A broad geographic label can hide differences that a summary does not explain. When the boundary is unclear, record that uncertainty instead of supplying a definition from memory.

Also distinguish the population the researchers want to understand from the observations they could actually obtain. A research question may concern investors broadly, while the available information comes from a particular participating group. The size or reputation of the publisher does not eliminate that distinction. Ask how observations entered the sample and what was unavailable.

Andy Allaway offers a related caution in his Empire Flippers interview on The SMB Investor podcast. Allaway describes selective follow-up rather than comprehensive outcome reporting. Paraphrased as a research-reading lesson, an experience that someone returns to report can illustrate an outcome without establishing the outcome of everyone who participated.

That podcast observation does not establish a flaw in an academic study. It supplies a question to ask of any collection of outcomes: how did the author learn what happened, and whose experience is missing? The answer must come from the study being evaluated. Do not treat uncertainty about missing observations as proof that those observations were favorable or unfavorable.

Yale's sample should be read as a case study, rather than an estimate of every LP's outcome. Its findings can help a reader see why the level of measurement and exclusions matter. They do not establish that an investor outside the observed sample had, or will have, the same experience.

Read the observation period

The publication date and the observation period answer different questions. Publication tells you when a document appeared. The observation period tells you when the underlying activity occurred and where the record stops. A recently published summary may describe investments whose histories began much earlier.

Write down the cutoff separately from the publication date. Then ask what status each observation had at that cutoff. A business still being held and a business whose investment has concluded do not provide the same kind of endpoint. If a summary does not explain how ongoing holdings were treated, the comparison remains incomplete on that point.

Look for the starting boundary as well. A sample organized around when searches began may differ from a sample organized around when businesses were acquired or investments concluded. Matching publication dates would not align those starting points. The dates only become meaningful once you know what event each date describes.

This is especially relevant when comparing a rate of return with a capital multiple. Time is part of the question for a rate; a multiple alone does not tell you how long the capital was invested. Similar-looking return language cannot substitute for an observation-period note.

The studies below have different cutoffs and do not necessarily treat ongoing investments the same way. Stanford measures through December 2025; IESE measures through December 2023. Read each study's methods before comparing its result with another study's figure.

Separate measurement levels

Keep the figures in separate rows, with the scope attached to each result. The table is a reading aid, not a ranking of studies, regions or investment opportunities. It deliberately does not pool the samples or calculate a blended result.

Research and source path	Population and measurement scope	Reported observation
Stanford GSB, 2026 Search Fund Study	U.S. and Canadian core search funds launched since 1984; initial search investors' cash flows through December 2025, including unsuccessful searches and operating companies; aggregate, pre-tax IRR, excluding follow-on financing	Stanford reports an aggregate pre-tax IRR of 33.9 percent1.
Yale SOM, How are Search Fund Investors Really Faring?, by Lazier, Thomas and Wasserstein	The observed case-study sample; company-level MOIC, excluding broken searches	The Yale SOM case reports a mean company-level MOIC of 2.78x and a median of 1.60x, excluding broken searches2.
IESE, International Search Funds – 2024	Core search funds outside the U.S. and Canada; eligible concluded funds with return data through December 2023, including unsuccessful searches and operating companies; initial search investors' cash flows, pre-tax, excluding follow-on financing for operating companies but including it for exits	IESE reports an aggregate IRR of 18.1 percent3.

The scope labels are part of the findings. Removing pre-tax from the Stanford label changes what the reader is told. Removing company-level or excluding broken searches from the Yale label conceals the boundary of the observation. Dropping international from the IESE label obscures the population being described.

An IRR and a multiple measure different things. The existing IRR definition explains the rate measure, while the MOIC definition explains the capital multiple. Use those definitions to identify the question each metric answers. Do not derive a conversion from the table or assume that a multiple implies a particular annual rate.

The level of measurement matters independently of the choice of metric. A company-level observation describes a company investment. An aggregate describes a combined calculation under the research methodology. Neither label, by itself, establishes the experience of a particular LP. Keep the unit being measured in the same sentence as the result.

Mean and median also answer different questions about the observed sample. The mean summarizes through an average, while the median identifies the middle observation. A difference between them is a reason to inspect the distribution, not permission to substitute whichever figure best supports a preferred argument. This article leaves that distribution question to the related research series.

Even Stanford and IESE's shared use of IRR does not establish that their figures are directly comparable. Matching the metric is only part of the work. Population, period and calculation basis still need to align before a reader can explain what a difference means. The secondary summaries alone do not justify a geographic ranking.

Record what remains unknown

A useful comparison note states both what the evidence supports and where the comparison stops. Unknowns should remain visible beside the finding. Otherwise, a reader returning to the note later may mistake an omitted qualification for a resolved question.

Use this checklist while reading each source:

• Record the research title and the publisher, then identify whether you read the original work or a recap.
• Describe the population in the source's language and identify what enters the sample.
• Record the observation period and distinguish it from the publication date.
• Identify the measurement level, the metric and any stated calculation qualifiers.
• Note whether unsuccessful searches are included, excluded or reported separately.
• Note whether the source explains how ongoing investments and missing observations are treated.
• Write down what the source does not establish about an individual investor's experience.

Do not fill gaps by borrowing a methodology from another study. A definition found in the Yale case does not automatically apply to Stanford or IESE. Likewise, a recap's silence about an exclusion does not establish that the primary research ignored it. Mark the distinction between missing from the summary and missing from the study.

Keep questions neutral. Asking whether unsuccessful searches are included is a check on interpretation. It is not an accusation that the author concealed losses. Asking how participants were selected does not establish that selection invalidated the work. These questions help define what the evidence can support without dismissing evidence you have not fully examined.

Keep search fund study comparisons traceable

The source trail should survive when a finding is copied into a note or shared with a colleague. Keep the study name, source path, scope and evidence reference together. A bare percentage or multiple loses the context needed to decide whether it belongs in the comparison at all.

Sources

-¹ Stanford GSB, 2026 Search Fund Study: Selected Observations, reports the aggregate return for core search funds in the U.S. and Canada. Its financial return method explains the investor cash flows, pre-tax basis and exclusion of follow-on financing. -² Lazier, Thomas and Wasserstein, Yale SOM, How are Search Fund Investors Really Faring?, reports the company-level observations and broken-search exclusion. -³ IESE, International Search Funds – 2024: Selected Observations, reports the aggregate return for eligible international funds with return data and explains its calculation method.

• John Steinberg, Inside the Search Fund Playbook episode, The SMB Investor podcast. The discussion here paraphrases only the asset-class attribution boundary.
• Andy Allaway, Empire Flippers interview, The SMB Investor podcast. The discussion here paraphrases only the limits of observed outcomes from selective follow-up.

Return to the search fund returns research guide for the broader reading map, and read the broken searches and acquisition risk explanation when a sample begins after acquisition. Leave unresolved fields marked as unknown until the source supports an answer.

Published in partnership with SMB Investor Network. Keep your research notes available so the questions you bring forward retain their source and scope.