Stanford’s 2024–25 search-fund neighborhood showed a median purchase near $16 million at ~6.2x EBITDA with a ~25% margin profile (Source: Stanford GSB). Quote that print when the file looks like the study. Do not grade a $500k-SDE fabricator against it.
How should you read a margin before you call it good?
Start with the earnings base. Adjusted EBITDA already subtracts a replacement manager and market-rate family wages. A 22% “EBITDA” that still includes the owner’s $200k draw is not 22%. It is SDE wearing a costume.
Then ask whether the margin is mix or luck. One fat aerospace job can hide a weak commercial book. Job costing and gross margin is the sibling extract. This page will not invent a NAICS margin table.
Census counted 284,452 manufacturing establishments in 2023; 68.4% had fewer than twenty employees (Source: Census CBP). Small shops often show lumpy margins because one programmer is also the estimator.
| Signal | What a good file shows | What a fragile file shows |
|---|---|---|
| Recast EBITDA margin | Stable after replacement pay | Collapses when the owner is priced in |
| Utilization vs Fed 76.0% | Honest hours, overtime explained | 95% “utilization” on a two-shift myth |
| ISM climate (PMI 54.6) | Used as weather, not as a grade | “The PMI is up, so pay me more” |
| Concentration | Margin not hostage to one account | 30% customer also the only profitable one |
| Capex | Maintenance visible in the cash | Margin bought by starving the iron |
Federal Reserve G.17 put manufacturing utilization at 76.0% in July 2026 (Source: Fed G.17). ISM’s August 2026 manufacturing PMI printed 54.6 (Source: ISM / Industry Today). Those are national climate cells. They do not certify your bay.
When is a thin margin still a good business?
When the floor is honest, the book is diversified, and the process is transferable. A tooling house with modest EBITDA and a strong OLV can still be a rational asset-aware deal. A 28% margin on one captive OEM can be a worse file.
How much customer concentration is too much belongs in the same sitting. Valuation still wants the size band before anyone argues a turn on a pretty percentage.
Why this FAQ will not publish a “good” percentage by sub-sector
The owner guide extracts the benchmarks you can actually pull from a file: margins, utilization, concentration. It does not invent a city cell or a state margin. The FAQ directory follows that rule.
BizBuySell’s 2.75x SDE median describes price on a sold Main Street set, not a required margin. Mixing a margin target with a multiple target in the same sentence is how people talk themselves into a number.
SBA packaging still cares whether cash after debt service exists inside a $5 million / 10% box. A gorgeous margin with a thin peg and a starved spare crib will not survive diligence.
How should you talk about margin in a CIM without inventing a grade?
Show the bridge. Show mix. Show the last three years, not the one pretty year. If a single aerospace job created the 24%, say so on the same page as the percentage. Buyers who discover the mix in week six do not feel lucky. They feel sold.
Do not grade yourself against a search-fund study unless the file looks like the study. Do not grade yourself against a neighbor’s rumor. Do not write “best in class” above a number you have not recast. The CIM is a conversation starter. It is not a report card from a trade association that does not exist.
Utilization belongs next to margin, not in a footnote. If you are running 50 hours on iron that could run 80, the margin may be fine and the capacity story may be the asset. If you are running 95 hours by starving maintenance, the margin is a costume. Fed 76.0% is climate. Your hours are the file.
Concentration belongs next to margin for the same reason. A 26% EBITDA that lives in one OEM is not a “good margin.” It is a hostage. Put the customer file in the same sitting as the percentage.
What should you bring if you want a margin conversation that is not theater?
Trailing-36 P&Ls. A recast that names every add-back. Hours or machine utilization you can defend. Sales and gross profit by customer. A capex list for the last three years and the next two. Whether you are still on the floor.
A “good” margin is a sentence you can defend with mix, utilization, and a recast — not a grade you copy from a study. If you cannot defend it, do not print it on the CIM cover. The benchmarks owner is where you extract the numbers you can actually pull from the file.
Jason Taken is a business broker. He will tell you if the margin is a costume. He will not certify a “good” percentage for a lender. Margin without mix is a costume. Margin without a recast is a costume. Put both on the same page before anyone grades the plant.
Educational information only. Read the owner, then use the intro call.