How is a manufacturing business valued once you name the size band?
Why one “manufacturing multiple” is the wrong tool
U.S. manufacturing is 284,452 employer establishments; 68.4% have fewer than 20 employees (Source: Census CBP, 2023). A 12-person CNC shop and a 220-person medical molder do not share a multiple. They barely share a buyer list.
BizBuySell’s 2021–2025 manufacturing transactions (2,303 sold businesses in that recap) show a 2.75x median SDE multiple, 2.04x–3.59x interquartile, median SDE $268,321, median revenue $1,127,990 (Source: Iconic recap of BizBuySell, retrieved 2026-09-10). GF Data’s Q1 2026 manufacturing drilldown, as recapped, averaged 7.2x EBITDA and ran from 5.9x under $25 million enterprise value to 8.8x in the $100–$250 million bracket (Source: Iconic GF Data recap). Those are different rooms.
IBBA and M&A Source Market Pulse Q1 2026 (300 advisors, 203 closings) plotted LMM EBITDA multiples and cash-at-close shares by enterprise-value band and reported that 67% of advisors saw no material AI effect on value (Source: IBBA highlights PDF). Pepperdine’s 2025 Private Capital Markets Report, as secondarily cited, put PE-style valuations near 5.5x EBITDA around $10 million of earnings (Source: Exit Equity citing Pepperdine 2025). Use each source in its size band.
Size-band × sub-sector table
Ranges below are synthesized from the sources in /research/valuation-benchmarks.md. They are not an appraisal. On-page they are “reported-deal and advisor-survey ranges.”
| Sub-sector | SDE < $1M (SDE x) | EBITDA $1–3M | EBITDA $3–10M | EBITDA $10M+ |
|---|---|---|---|---|
| CNC / general job shop | 2.0–3.6 | 4.0–6.0 | 5.0–6.5 | 6.0–8.0 |
| Metal fab / commodity stamp | 2.0–3.5 | 4.0–5.5 | 5.0–6.5 | 6.0–8.0 |
| Plastics (IM / extrusion) | 2.2–3.8 | 4.5–6.5 | 5.5–7.5 | 6.5–9.0 |
| Electronics / EMS | 2.0–3.5 | 4.0–6.0 | 5.0–7.0 | 6.0–8.5 |
| Aerospace & defense (AS9100) | 3.0–5.0 | 6.0–8.0 | 6.5–9.0 | 7.0–11.0 |
| Medical (ISO 13485) | 3.0–5.0 | 6.0–8.0 | 6.5–9.0 | 7.0–11.0 |
| Automotive / IATF | 2.0–3.4 | 4.0–5.5 | 5.0–6.5 | 6.0–8.0 |
| Food / co-pack | 2.2–3.6 | 4.5–6.0 | 5.5–7.0 | 6.5–8.5 |
| Packaging (specialty) | 2.2–3.8 | 4.5–6.5 | 5.5–7.5 | 6.5–9.0 |
| Print (commodity) | 2.0–3.2 | 4.0–5.0 | 4.5–6.0 | 5.5–7.0 |
| Wood / millwork | 2.0–3.6 | 4.0–6.0 | 5.0–6.5 | 6.0–8.0 |
| Chemicals / coatings | 2.2–3.8 | 4.5–6.5 | 5.5–7.5 | 6.5–9.0 |
| Tooling / dies / molds | 2.4–4.0 | 5.0–7.0 | 5.5–7.5 | 6.5–9.0 |
| Building products | 2.0–3.6 | 4.5–6.0 | 5.5–7.0 | 6.5–8.5 |
| Textiles / sewn | 2.0–3.2 | 4.0–5.5 | 4.5–6.0 | 5.5–7.5 |
| Machinery OEMs | 2.2–3.8 | 4.5–6.5 | 5.5–7.5 | 6.5–9.0 |
Aerospace and medical cells sit higher because certificates and customer qualification are scarce. Commodity fab and print sit lower because iron is easier to replicate and customers shop. If a cell is being used as a quote, it should be labeled a range, not a promise.
SDE versus EBITDA
SDE (seller’s discretionary earnings) is pre-tax cash the owner-operator can take: net income + owner pay + interest + non-cash + documented personal and one-time items. It is the right language when the buyer is replacing the owner’s job.
EBITDA is earnings before interest, taxes, depreciation, and amortization. Adjusted EBITDA subtracts a replacement general manager and other normalizations an institutional buyer will demand.
A shop with $900,000 of SDE that includes $220,000 of owner salary is not a $900,000 EBITDA company. Subtract the replacement GM before you talk to a searcher or a PE associate. Stanford’s 2024–25 search-fund median sat near $2.5 million EBITDA and 6.2x (Source: Stanford 2026 study recaps). That math already assumes a professional operator in the number.
Add-backs that live, and add-backs that die
| Usually survives QoE | Usually dies |
|---|---|
| Owner W-2 above a documented replacement | “I could hire someone for half” with no market comp |
| One-time move, flood, or lawsuit | Recurring “one-time” every year |
| Personal vehicles / travel clearly on the P&L | Unfunded maintenance called an add-back |
| Non-operating interest in a side LLC | Inventory step-up without a physical |
| PPP or similar dated items, isolated | Future tariff “givebacks” |
Sellers who win QoE have a binder. Buyers who lose money skipped the binder.
Asset value versus earnings value
Enterprise value assumes a going concern. Machinery-and-equipment appraisers will also give:
- FMV in continued use
- OLV (orderly liquidation)
- FLV (forced / auction)
If OLV of the iron + net working capital − environmental and deferred-maintenance corrections exceeds earnings value, the honest valuation is the asset floor. That is not a distressed-only problem. It shows up whenever a shop kept buying machines while margins compressed.
Inventory is not automatically “plus inventory.” Raw, WIP, and finished goods need an E&O pass. Percentage-of-completion and job-cost WIP are where manufacturers and software companies stop resembling each other.
Real estate is outside the operating multiple. Price the plant on its own, then lease or sell.
Value drivers that actually move a turn
- Concentration. Rule of thumb: top customer above 25–30% of sales is a turn or a structure problem.
- Backlog and contracts. Firm releases beat verbal forecasts. Cancel-for-convenience is not “recurring revenue.”
- Certificates. AS9100, ISO 13485, IATF, SQF — current audits only.
- Owner dependence. If quoting dies when the owner fishes, the multiple dies too.
- Equipment age and capex. Deferred spindle rebuilds are a price chip, not a story.
- Union / withdrawal liability. Price the letter from the fund.
- Environmental RECs. Phase I (ASTM E1527-21) is how adults start that conversation.
- Government / ITAR mix. Premium and friction in the same sentence.
Illustrative only: two $2.4 million-EBITDA precision shops. Shop A has 16% top-customer share, AS9100, and a 2018+ CNC mix. Shop B has 41% top-customer share, a lapsed ISO, and 1990s manuals. They do not share a multiple. Anyone who says they do is selling a brochure.
Worked examples (illustrative)
Example 1 — SDE job shop, machining, sde-under-1m. Trailing SDE $380,000 after a $140,000 replacement-supervisor add-back is refused. Apply 2.5x–3.0x SDE (BizBuySell manufacturing band). Illustrative EV $950,000–$1,140,000 before peg and excess cash. If a decent OLV on the cells is $1.05 million and inventory is clean, you are in asset-floor territory. Do not “split the difference” with a 5x fantasy.
Example 2 — Medical molder, ebitda-3-10m. Adjusted EBITDA $4.1 million, ISO 13485 current, top customer 19%, no RE in the deal. Table range 6.5x–9.0x. Illustrative EV $26.7–$36.9 million. The fight will be QoE add-backs, tooling ownership, and the peg on resin and WIP — not whether the multiple “should be 3x because manufacturing.”
Neither example is a quote. Both exist to stop blended math.
Appraisal versus broker opinion
A broker opinion of value is what Jason Taken will discuss on an intro call: a range, a buyer set, and the two or three issues that will move you inside the range. An appraisal for a lender, an ESOP trustee, a partner buyout, an estate, or a divorce is a different product from a credentialed appraiser. This firm will not pretend otherwise.
Partner buyouts, estates, and divorce often need a defined standard of value (FMV, fair value, investment value). That is counsel and appraiser work. The process page is about a marketed sale, not a courtroom.
Data current as of September 2026
ISM Manufacturing PMI 54.6 in August 2026; Prices Index 71.1 (Source: ISM). Those figures inform margin risk. They do not replace the table.
How backlog and contracts should change the number
Backlog is not a multiple by itself. Firm releases with cancel-for-convenience clauses are weaker than they look. Long-term agreements with pricing formulas that lag steel or resin are a margin risk, not a “recurring revenue” slide. Book-to-bill under 1.0 for several quarters is a yellow flag even if trailing EBITDA still looks fine.
The valuation move is to separate (a) firm scheduled work, (b) blanket orders without dates, and (c) forecast. Only (a) supports the high end of a range. Earn-outs on named accounts are how adults bridge (b) and (c). A CIM that sums all three and multiplies is how arguments start in week six of exclusivity.
How owner dependence and equipment age should change the number
If quoting, programming, and closing sit in one person, you do not have a $3 million-EBITDA company even if the P&L says you do. Subtract a real replacement package — burdened — and then decide whether a buyer will believe the customers will stay. The discount is often a turn, a longer training period, or both.
Equipment age is not a year on a tag. It is control generation, crash history, and whether the quoted capacity assumes one shift or hero overtime. Fed G.17 manufacturing utilization was 76.0% in July 2026 (Source: Fed G.17). A shop that claims 95% on one-shift overtime is telling you the second-shift path is unpaid. Price the hiring plan.
Capex should be split: maintenance capex to keep the current output, growth capex to add a cell. Buyers will treat deferred maintenance as a dollar-for-dollar chip. They will not pay a growth multiple for a crate in the aisle.
Union status, real estate, and distress
Union shops are sellable. Withdrawal liability, successorship, and the CBA’s assignability are priced items. Pretending “the union is like family” is not a valuation method. Get the estimate letter.
Real estate inside the multiple is how sellers accidentally give away a building. Price the plant on rent coverage or cap rate, then decide sale-leaseback versus a 504. Industrial cap rates move with the metro; do not invent a national factory cap rate on this page.
Distressed valuation starts at OLV and works up only as far as a buyer will pay for a customer list and a trained crew. Peak-year SDE is not the base. Last six months, annualized, after inventory write-downs, is closer.
Partner buyouts, estates, and divorce
Those assignments often need a defined standard of value and a credentialed appraiser. Fair market value, fair value, and investment value are not synonyms. A marketed-sale broker opinion is the wrong instrument for a courtroom or a buy-sell that says “appraiser.” HedgeStone will say so on the first call rather than stretch a CIM into a sworn report.
How to use the table without lying to yourself
Pick the row that matches the process (CNC versus medical versus print). Pick the column that matches the earnings base after a replacement operator is paid. Then move inside the cell using concentration, certs, backlog quality, and the asset floor. If you are outside the cell, you need a reason a second person can audit — a scarce certificate, a proprietary product, or an OLV floor — not a feeling about the year.
Illustrative only: a packaging converter with $6.8 million of adjusted EBITDA, no customer above 14%, and clean GFSI audits sits in the specialty-packaging $3–10M band (5.5x–7.5x). A commodity printer with the same EBITDA and a 44% customer does not inherit that cell. The table is a fence, not a wish.
Inventory in the valuation, not as a plus-up
Sellers like to say “plus inventory.” Buyers like to say “normalized working capital.” Both can be true only if inventory is counted, aged, and written down before the multiple is applied. Raw that is obsolete is not an asset. WIP that cannot ship is not an asset. Finished goods the customer already rejected are not an asset. A valuation that adds book inventory on top of a cash-free, debt-free, normalized-WC enterprise value is double counting.
The peg exists to keep that fight from becoming a second purchase price. The valuation exists to decide whether the going-concern number is even the right number, or whether OLV plus a clean peg is the honest offer. Those are sequential questions. Mixing them is how people “split the difference” into a number neither appraisal nor earnings can defend.
Machinery-and-equipment appraisals: OLV, FMV, FLV
Hire the appraiser for the standard you need. Lenders often want OLV because they have to sell the iron if the note fails. Going-concern buyers care about FMV in continued use. Auction math is FLV and is the wrong instrument for a healthy shop — unless the shop is not healthy. Useful lives in a CIM should match the appraisal, not a blog. Do not publish asking prices for a 2014 VF-4SS without a sourced comp. The machine list is evidence. The appraisal is the number.
A calculator, later
The manufacturing size-band multiple calculator now ships as its own URL with sourced defaults and a disclaimer — still not a live quote and not an appraisal. The table on this page remains the master grid: pick the cell, then move inside it with concentration, certs, owner dependence, and the asset floor. Do not average the entire table and call it “the manufacturing multiple.” That average is how a $400,000-SDE job shop gets told it is worth a PE headline.
Quality certifications as access, not décor
AS9100, ISO 13485, IATF 16949, ISO 9001, and SQF change who can buy the output. They change the multiple only when the last audit is clean and the certificate can survive the stock-or-asset structure. A lapsed cert is a customer-access discount. A cert the owner treats as a logo is a diligence finding. NADCAP special processes are even narrower: the process, not the building, is what the customer approved.
Do not add “a turn for ISO” as a rule. Add access: if losing the cert loses 40% of sales, you do not have a 40% customer — you have a certificate customer. Price that.
How we will talk about your number on the call
The intro call will not produce a binding number. It will produce a cell in the table, two or three issues that move you inside the cell, and whether the asset floor is the real conversation. If you want a lender or a court to rely on a number, hire an appraiser. If you want a marketed range that a buyer set can understand, that is the broker opinion. Mixing those jobs is how people get angry at the wrong professional.
For partner buyouts, estates, and divorce, stop using this page as the report. Use it to understand why the appraiser’s standard of value will not match a marketed CIM, then hire the appraiser the document requires.
Valuation cluster doors (one question per URL)
Child pages own methods, EBITDA multiples by size, multiples by sub-sector, SDE versus EBITDA, add-backs, asset versus earnings value, OLV/FMV/FLV, inventory, concentration, backlog, certificates, owner dependence, equipment age, union, real estate, distress, worked examples, a calculator spec, appraisal versus broker opinion, and partner/estate/divorce standards. This pillar stays the master table. Do not average the table.
| If you came here asking | Open next | Do not do |
|---|---|---|
| “What multiple?” | Size-band × sub-sector table on this page | Blend Main Street SDE with GF Data EV |
| “Is the iron the story?” | Asset-floor test | Write a growth CIM on an OLV file |
| “Can a lender use this?” | Appraisal vs broker opinion | Hand this page to a court |
| “Why is my neighbor higher?” | Different size band or certificate | Copy their rumor |
BizBuySell’s 2021–2025 manufacturing median remains 2.75x SDE (Source: Iconic recap). GF Data’s recapped 7.2x average remains a larger-EV series (Source: Iconic GF Data recap). Both can be true. Neither is your plant’s quote.
A broker opinion on this pillar is a marketed range with a size band and a sub-sector label. It is not a USPAP report, not a lender appraisal, and not a court exhibit. If those documents are required, hire the credential those forums asked for.
Educational information only. Not an appraisal and not tax, legal, or engineering advice. Confirm with M&A counsel, a CPA, and qualified appraisers. No value is guaranteed.
Cells move when inputs move; slogans do not get a vote
Pick SDE or adjusted EBITDA by size band. Apply the sub-sector cell. Then test the asset floor. A GF Data larger-EV print does not belong on a Main Street SDE shop. A Stanford 6.2x does not belong on an 11 percent owner-estimator plant (Source: Stanford GSB). An aerospace cell does not belong on a lapsed certificate.
Illustrative teaching cases on the hubs are not offers. Change concentration, lapse a certificate, or find a REC without a Phase II plan and the cell moves. That is the point of naming the process. Hire an appraiser for the standard a lender or a court asked for. This page remains a broker-opinion framework.
Why a later calculator still will not average the table
A size-band × sub-sector calculator will ship as its own URL. Until then, and after then, the rule is the same: pick SDE or adjusted EBITDA, pick the process cell, then test the asset floor. Do not average BizBuySell’s 2.75x SDE manufacturing median with a larger-EV sponsor print and call the blend “the market” (Source: BizBuySell / Iconic). Do not drop Stanford’s ~6.2x search median onto an owner-estimator shop (Source: Stanford GSB).
Inventory class, certificate access, and owner dependence move you inside the cell. Climate prints — ISM PMI 54.6 and Prices 71.1 in August 2026, Fed utilization 76.0% in July 2026 — stay in the CIM weather sentence (Source: ISM; Fed G.17). They do not add a turn. Hire an appraiser when a lender, a trustee, or a court asked for a standard this page does not claim.
How owner dependence and certificates move you inside a cell
Owner dependence is a second-chair test. If the owner is the only person who can start the scarce job, you are buying a training period. Certificates are an access test. If losing the scope loses a named share of sales, you do not have a diversified book — you have a certificate customer. Neither fact adds “a turn for ISO” as a rule. Both facts move you inside the size-band × sub-sector cell already on this page. Read manufacturing financial statements explained when the job-cost file, WIP method, and add-back schedule have to sit under that cell.
Hire an appraiser when a lender, a trustee, or a court asked for USPAP or adequate consideration. This pillar remains a broker-opinion framework. How owner dependence affects value and quality certifications own the deeper versions.