How does a search fund buy a manufacturer without overpaying?
A search fund is an operator-led acquisition vehicle: a searcher raises search capital, finds a company, and buys it with investor equity plus debt, often plus a seller note. Stanford's Search Fund Study 2026 (data through 2025-12-31) put the 2024–25 median purchase at $16 million and about 6.2x EBITDA on a company with roughly $8.1 million of revenue, $2.5 million of EBITDA, a ~25% margin, ~30 employees, and ~12% EBITDA growth (Source: Stanford GSB; Five Experts recap). That is the study's median. It is not a CNC job-shop median.
Why manufacturers confuse searchers
Many plants are 10–15% EBITDA with an owner who still estimates. The study median is a 25% margin profile. Searchers who ignore that gap overpay or fail diligence. Searchers who respect it still buy manufacturers — especially certified or proprietary-product shops with a second chair — but they underwrite labor and quality systems harder than they underwrite a contracted services firm.
Deloitte / Manufacturing Institute: as many as 3.8 million manufacturing employees needed 2024–2033, up to 1.9 million unfilled if the skills gap holds (Source: Deloitte/MI, 2024). A searcher who cannot hire a programmer is not buying a lifestyle. They are buying a constraint.
How they underwrite
Durable customer relationships, not a single RFQ book that re-trades every quarter. A replacement GM already in the number or a documented hire. Certificates that survive a stock or asset sale. Concentration they can live with. A machine list that is not a capex cliff in year one. They will ask who quotes, who programs, and who the customers actually call.
Practice, not law: many searchers will look at $1 million-plus of adjusted EBITDA with a story. The study median is ~$2.5 million. SBA individuals live in a different band — see selling to an SBA buyer.
Structure
Bank debt plus investor equity plus, often, a seller note. Hold is operator hold; a later sale to PE is a possible second bite for the searcher, not a promise to you unless you roll. If 7(a) is in the stack, SOP 50 10 8 still requires a 10% injection and counts a seller note as equity only on full standby for the life of the loan and only up to half the injection (Source: SBA). Traditional search notes that pay current interest are not that instrument.
QoE appears more often than on a pure Main Street SDE file. Working-capital pegs still fail on WIP if the LOI copied a retailer.
Qualification
Median search duration in the study is about 20 months. Aggregate acquisition rate since 1996 is 58%; recent 2021–24 launches closer to ~50% (Source: Stanford GSB). Those figures describe a population, not your buyer. Ask for the capital path. An independent searcher and a funded search are different close risks — independent sponsors are a third animal.
Relocation is a diligence item. A searcher who will not move to the plant is telling you the operating thesis is thin.
When a searcher is the right buyer
When you want an operator in the building. When the earnings base looks like a real company, not a job plus assets. When you will finance a piece and you trust the person. When a PE add-on would fold the name and you care about the floor crew seeing a successor.
When it is the wrong exclusive: peak-year tooling shops, 40% customer concentration with no structure, owner-only AS9100, or an asset-floor file. Those need a different conversation — valuation first.
Process notes
Blind teaser still. The searcher community is talkative. Do not let a "quick walkthrough" become the marketing channel. Plant visits after tax returns and the machine list. Customer names last.
Study-wide IRR 33.9%, ROI 4.75x, PME 2.88 as of 2025-12-31 are historical aggregates for the search model, not manufacturing-specific forecasts (Source: Stanford GSB). Do not put them in a CIM as if they were your return.
Educational information only. Confirm structure and securities with counsel. Jason Taken is a business broker, not a search-fund advisor.
What a searcher should send you
A buybox that names a process, a size band, and a geography. A capital path that is more than a Stanford citation. A relocation plan. A view on who runs the bid desk after the owner leaves. If the inbound is "any manufacturer under $10 million," send them back to the buybox. HedgeStone will not waste a teaser on a tourist.
Margin math without folklore
Take trailing EBITDA. Subtract a replacement GM if the owner is still in the number. Look at the result as a percent of revenue. If you are far from the study's ~25% median, say so in the CIM. Some searchers will still underwrite a 14% certified shop. Some will not. Lying about the margin wastes a month.
Seller notes versus SBA notes
Search models often want current-pay paper. SOP 50 10 8 wants full standby if the note is equity for 7(a) (Source: SBA). Those are different instruments. Pick a stack. Do not sign both stories into one LOI.
Read the process page and the valuation pillar before you exclusive with the first searcher who quotes a Stanford multiple at your 11% job shop.
Margin profile first, searcher vocabulary second
Stanford's 2024–25 median purchase was about $16 million at 6.2x EBITDA on a ~25% margin company (Source: Stanford GSB). Many manufacturers are not that profile. Some still fit. Do not borrow the median because the buyer says search fund.
What Stanford's median is — and what it is not
The 2024–25 median purchase was about $16 million at 6.2x EBITDA on a ~25% margin company. That is a filter, not a manufacturing law. Owner-estimator shops with 11% margins are usually a different buyer. Searchers who skip quality-system cost are not ready for a certified plant.
Run them next to SBA and strategics when the file can support more than one class.
How a searcher should look on a manufacturing file
Like someone who can survive a replacement GM, a quality-system cost, and a peg that is not a retailer definition. Stanford's ~25% margin median is the filter. A job shop at 11% is allowed to say no. Run SBA and strategics in the same process when the file can support them. Exclusive LOI is a prize after the margin profile is honest.
Traditional search capital versus a self-funded searcher on a shop floor
A traditional searcher raises a search vehicle, then raises acquisition equity from that same investor group. A self-funded searcher often uses personal capital, SBA 7(a) when the file fits the $5 million box, and a seller note (Source: SBA). Those are different close calendars. Ask which one you are talking to before you grant a long exclusive. Stanford’s 2024–25 median purchase of about $16 million at roughly 6.2x EBITDA on a ~25% margin company is a traditional-search print (Source: Stanford GSB). It is not a 14-person CNC multiple and it is not a self-funded 7(a) box.
A searcher who will sit as CEO still has to survive a replacement wage if the owner’s number is inside EBITDA. They still have to survive quality-system cost if the book is certified. They still have to survive a manufacturing peg that classes raw, WIP, and finished goods. “I ran a software search” is not a plant underwriting memo.
| Searcher claim | Ask for | Walk if you get |
|---|---|---|
| Traditional fund | Investor list and a written equity path | “The board likes manufacturing” |
| Self-funded + 7(a) | Credit conversation and SOP 50 10 8 injection plan | A current-pay note labeled as equity |
| Operator plan | Who quotes on Monday | The owner as unpaid training |
BizBuySell’s 2025 manufacturing median close was 223 days (Source: BizBuySell). Search processes that also need a QoE and a relocation stretch that clock. Do not promise the floor a 90-day close because a searcher is “ready to move.”
Run the searcher next to an SBA operator and a strategic when the file can support more than one class. Exclusive LOI is a prize after the margin profile is honest.
What a searcher QoE will rebuild on a job-cost file
Search boards that have seen software files often meet manufacturing WIP for the first time in quality of earnings. They will rebuild job cost from packing lists, time tickets, and the shipping log. Unbilled first-article hours are not finished goods. Customer-owned tools are not inventory you can borrow against. If the controller cannot produce those classes, budget the accountant before you grant exclusivity.
Stanford’s ~20-month median time-to-buy is a search-model statistic, not a close date you can put on an employee memo (Source: Stanford GSB). A manufacturer that also needs a Phase I or a certificate transfer runs longer than a contracted-services file. Say that in the process letter.