What actually happens when people buy small businesses

Buying a small business instead of starting one has been measured properly, at scale, for forty years, and the results are better than most people expect and considerably rougher than the people selling the idea admit. A quarter of the acquisitions lose money. The headline return is carried by a handful of outliers. Both of those things are true alongside a genuinely excellent aggregate.

The one dataset worth arguing from

Almost everything written about acquisition entrepreneurship is argued from anecdote, usually the author's own and usually the good one. There is one exception. Stanford's Graduate School of Business has tracked search funds since 1984 and publishes the results every two years, and the 2026 edition covers 862 traditional funds across the United States and Canada, which the study puts at roughly 99 percent of every such fund known to exist. A further 503 international funds are tracked separately.

A search fund is a specific vehicle: somebody raises a small amount of money from investors to spend a year or two looking for a company, buys one, runs it, and eventually sells it. That specificity is exactly what makes the data useful. It is a defined population with a defined start and a defined end, followed for four decades, which is not something you can say about any other form of small-company acquisition.

The caveats are real and worth stating before the numbers rather than after. These are mostly graduates of a small number of business schools, backed by an experienced investor group who screen the deals, buying in North America. That is a selected population with structural support, and it will outperform a person doing this alone from a standing start. Read the figures as an upper bound on a well-supported version of the strategy.

The aggregate, and then the distribution

Across all 862 funds, the study reports an aggregate internal rate of return of 33.9 percent and 4.75 times invested capital. Funds that both acquired a company and exited it did better still, at 39.3 percent and 5.98 times. Measured against the public market, the aggregate public market equivalent is 2.88 times, so as a class this materially outperformed the index.

Now the part that rarely gets quoted. Strip out the funds returning ten times or more, and the aggregate falls to 2.8 times capital and 27 percent. Still a very good outcome, and a completely different sentence.

Two cumulative figures do more work than any single bar. Twenty-six percent lost money. And adding the first three rows, 46 percent returned less than twice the capital put in, after a search of roughly two years and then several years of running the company. Nearly half of a well-supported, pre-screened population did all of that and roughly got their money back.

That shape matters more than the headline, because the headline describes a portfolio and almost nobody reading this is a portfolio. An investor holding twenty of these gets the aggregate. An individual buying one company gets a single draw from that distribution, and a one in ten chance of losing the lot is a different proposition from a 33.9 percent return.

The average is a fact about the asset class. What happens to you is a fact about one company, and those two numbers have almost nothing to do with each other.

Which is the strongest argument I know for being ruthless about selection rather than optimistic about deal flow. You do not get to average across your mistakes. You get the one you bought.

There is a second asymmetry underneath that one, and it is the more important of the two because it is about you rather than about the data. Everybody in this study had investors. That is what a search fund is: other people's money, committed by a group who screen the deal, sit on the board, and have seen this go wrong before. It changes the arithmetic of failure in a way the return figures do not show.

So the 10 percent total-loss row means something different depending on who is reading it. For the population being measured it is a survivable outcome inside a portfolio of relationships. For somebody putting their own capital into one company it is closer to a terminal event. Same percentage, different consequence, and the study cannot tell you about the second one because it never measured anybody in that position.

The practical conclusion is not to be discouraged by that. It is to buy the boring end of the distribution deliberately. The outliers above ten times are not the target and were never available to be chosen in advance; they are a fact about the population, not a strategy. A business bought carefully at a sane price, that produces cash and does not need heroics, lands somewhere in the middle rows, and the middle rows are a perfectly good life.

Finding one is harder than it looks

Before any of the return figures apply, you have to actually buy something, and this is where the data is most sobering. Across the whole history of the study, 58 percent of concluded searches ended in an acquisition. In the more recent cohorts, from 2021 to 2024, that fell to 48 percent.

So of the people who committed to this full time, raised money to do it, and searched to a conclusion, fewer than half of the recent group ended up owning anything. The median search took around twenty months. That is close to two years of work with a coin-flip chance of a company at the end.

1. The market got more competitive

   More capital chasing the same population of retiring owners, including from people who had read the same books. Competition on price is the visible effect. Competition for the seller's attention is the one that actually slows a search down.

2. Good businesses are scarcer than listings

   A pipeline of companies for sale is not a pipeline of companies worth buying, and the gap between those two numbers is where most of the twenty months goes.

3. Sellers are not on your timetable

   An owner deciding whether to hand over a life's work moves at their own speed, which is measured in years and is not responsive to your fund's clock.

4. Financing conditions moved

   The recent cohorts searched through a sharp rise in the cost of debt, which changes what price works and therefore which deals close at all.

None of that is a reason not to do it. It is a reason to be honest about the timeline, and specifically about the fact that the searching phase is the phase, not a preamble to the real work. Anyone describing acquisition as a shortcut past the hard part of building something has not accounted for two years of rejection.

What actually gets bought

The picture of what these buyers end up owning is more useful than the general advice, because it is revealed preference across hundreds of deals rather than opinion.

For the 2024 to 2025 cohort the median acquisition had an enterprise value of 16 million dollars, on median earnings before interest, tax, depreciation and amortisation of 2.5 million, at a median multiple of 6.2 times. That is not a corner shop and it is not a mid-market buyout. It is a real company with a management layer, bought at a price that assumes it keeps performing.

On sector, the study is clear: services is consistently the most common category, followed by software, with education, and specifically credentialing and vocational training, reaching its highest ever count. That last one is worth dwelling on, because it cuts against the assumption that this strategy is about trades and physical work. It is not. The common thread is recurring revenue, a fragmented market, and an owner who wants out, and that combination exists in software, in professional services, in training, in healthcare administration, and in dozens of unglamorous categories that have nothing to do with a van.

It is also worth noticing what the multiple implies about leverage. A company bought at six times earnings, part-funded with debt, does not tolerate a bad year quietly. The distribution at the top of this article, with a quarter of deals losing money, is not mostly a story about bad businesses. It is substantially a story about ordinary businesses bought at prices that left no room.

What the data says to do differently

Five things follow from the numbers, and each of them cuts against something the genre repeats.

• Selection beats sourcing. You get one draw from the distribution, so rejecting faster is worth more than seeing more
• Price discipline is the whole game at a 6.2 times median, because there is no operational improvement that reliably rescues an overpayment
• Budget two years for the search and treat it as the work, not the wait
• Ignore the sector folklore. The pattern is recurring revenue in a fragmented market with a retiring owner, and it is industry-agnostic
• Read any aggregate return with the outliers stripped out, and ask whether you are buying a portfolio or a company

Reading past an average generalises well beyond this. Whenever a strategy is sold with an average, the honest question is what the distribution looks like and how many draws you get. A 4.75 times aggregate across 862 funds and a 4.75 times expectation for your single acquisition are not the same claim, and the difference between them is where most disappointment in this field comes from.

Why I still think it is the right trade

Having spent this long on the unflattering parts, the conclusion is still that buying beats building for most people who can do either. A company that already exists has customers, cash flow, a team and a track record on day one, and the failure rate of new businesses over the same horizon is far worse than a 26 percent partial-or-total loss rate. The comparison that matters is not between buying and certainty. It is between buying and the alternative.

What the data changes is not whether to do it but how. It makes a strong case for a narrow, well-defined target profile, for walking away often, and for being extremely careful about price, because that is where the losing quarter comes from. It also makes the case for holding rather than flipping, since the study's own exit-driven framing measures something different from what a permanent owner is optimising for. I have written about that difference in permanent capital versus the fund clock, and the case for buying over building in buy a business, or build one. What I actually screen for before any of this is on the buy page.

All figures from the 2026 Search Fund Study, Selected Observations, Stanford Graduate School of Business, covering 862 traditional search funds in the United States and Canada launched between 1984 and December 2025.