Jason Taken will split your invoices before he splits your multiple. HedgeStone is a brokerage, not a branding studio. Sell my manufacturing business remains the pillar. This page is the mix test. Score the same identity as a row-by-row table on the job shop vs proprietary product manufacturer compare page.
What is actually being sold on each file?
A job shop — including many CNC and fab houses — sells available hours, certifications, and the owner’s or estimator’s relationships. The customer owns the drawing more often than not. Tooling may be customer-owned. Tomorrow’s book is an RFQ stack.
A proprietary product company — including many machinery OEMs and catalog component makers — sells a specified item the customer reorders. You own the drawing, the brand, or both. The sales motion can be reps, distributors, or a catalog. The floor still matters. It is not the whole story.
Hybrids are common. A shop that designed one fixture twenty years ago and still job-shops 85% of hours is a job shop. A catalog company that takes custom variants is still a product company if the variants sit on a common platform.
Census CBP 2023: 284,452 establishments, 68.4% under 20 employees (Source: Census CBP, 2023). Most of those plants are capacity businesses. That is allowed. It is why Main Street medians look like SDE multiples rather than larger-EV EBITDA prints.
How do multiples and buyer sets diverge?
Valuation always carries size band × sub-sector. Inside a band, product versus build-to-print is a broker value driver.
| File type | Typical earnings basis | Discussion neighborhood (not an appraisal) | Usual buyer set |
|---|---|---|---|
| General job shop, SDE under $1M | SDE | 2.0x–3.6x BizBuySell manufacturing quartiles | Individual / SBA, some searchers |
| Specialty job shop still < $1M SDE | SDE | 3.0x–5.0x observed cleaner / certified range | Searchers, small strategics |
| Commodity product, EBITDA $1–3M | Adjusted EBITDA | 4.0x–5.5x synthesized general-fab-like cell | Search, SBA if EV fits, small PE |
| Specified / certified product, EBITDA $1–3M | Adjusted EBITDA | 6.0x–8.0x aero/medical-like cell if certs are real | PE add-on, strategics |
| OEM / product, EBITDA $3–10M | Adjusted EBITDA | 5.5x–9.0x by niche and backlog | PE, strategics, sponsors |
BizBuySell’s 2021–2025 manufacturing 2.75x median SDE (Source: BizBuySell / Iconic) is the job-shop gravity well. GF Data Q1 2026 manufacturing recap 7.2x average and 5.9x under $25 million EV (Source: Iconic GF Data recap) is a larger-EV series. Product platforms show up there more often than $400k-SDE brake shops. Do not average the two.
Stanford’s 2024–25 search median of about $16 million at ~6.2x EBITDA on a ~25% margin company (Source: Stanford GSB) is closer to a product or certified file than to a walk-in job shop. Some searchers still buy shops they can operate. Eligibility is not a multiple.
SBA 7(a) remains a $5 million cap with a 10% injection under SOP 50 10 8 (Source: SBA). A product company whose enterprise value sits above that box needs a different stack. A job shop whose SDE fits the box should not exclusive only with a PE associate who wanted a platform.
What diligence questions split the two?
Job shop: who quotes, who programs, who holds the customer, what percent of hours are repeat parts, who owns the fixtures, how much of backlog is cancel-for-convenience, what happens if the estimator leaves. Owner-dependent is the companion page.
Product: who owns IP, who can tool a copy, what is warranty reserve, what is channel inventory, what is returns history, what is the SKU concentration, what is the next redesign cycle. A patent number in the CIM that expired is worse than no patent.
Both: quality of earnings will test whether “repeat” is invoice reality. Both: customer concentration still prices. A product company with one big-box account is not diversified because the SKU has a UPC.
How should the peg and inventory be defined on each?
Job-cost WIP, customer-owned material, and unbilled change orders are the job-shop peg fight. Finished goods, kits, and distributor stock are the product peg fight. The working-capital peg must name the mix. A single “inventory” cell is how true-ups steal cash.
Slow-moving raw on a job shop is often leftover alloy for a program that ended. Slow-moving finished goods on a product company is often an obsolete revision. Write both down before the peg.
IBBA Q1 2026’s 83% of >$5 million deals with three-plus offers (Source: IBBA) helps product files that can attract strategics. It does not rescue a job shop that sent the teaser to one local rival.
What does an illustrative pair look like?
Illustrative only, same metro, same $1.8 million of adjusted EBITDA — already above the Main Street SDE world.
File A: 90% build-to-print, owner quotes, no customer above 19%, 2012–2018 CNC mix, ISO 9001 current. Discussion lives in a precision-machining $1–3M cell (synthesized 5.0x–6.5x), discounted for owner dependence.
File B: 75% catalog components with drawings in the company’s name, distributors in three regions, same EBITDA, ISO 9001 plus a trademark. Discussion can live higher in the same size band if QoE confirms repeat orders. Neither file is a 2.75x SDE shop. Neither is an 8.8x $100–250 million EV print.
If File A’s owner calls File B’s multiple “the manufacturing multiple,” the process starts with a fiction.
How should the teaser tell the truth without giving away the shop?
Say “build-to-print precision machining” or “proprietary catalog components” in the first sentence. Do not say “product company” because you designed one bracket. Do not say “job shop” as an apology if 80% of GP is a branded line.
Keep names last. Keep unique facade photos out. Keep the IP story for the NDA package with documents.
How should warranty, returns, and field service sit in the product-company recast?
A catalog line without a warranty reserve is a QoE gift. Field-service costs that live on the owner’s truck are still costs. Returns that get quietly remade as “goodwill jobs” are margin leaks. Job shops have escapes too, but they usually show up as scrap on a traveler. Product companies hide them in the owner’s weekend.
Document the policy. Age the reserve. If you have never had a reserve and you are going to market, do not invent a tiny one the week of the CIM. Inventing looks like concealment. Disclosing a gap looks like adulthood.
Channel inventory at distributors is not your FG unless the contract says you still own it. Stuffing a channel to juice trailing sales is a QoE classic. Don’t.
What sales motion evidence do buyers want on each file?
Job shop: RFQ log, win rate, estimator capacity, repeat-part percentage. Product: order-book by SKU, reorder rates, rep agreements, distributor sell-through if you have it, website or catalog that matches invoices.
If the “product company” cannot show reorders, it is a job shop with a brochure. If the “job shop” has 60% repeat parts on drawings it owns, it may be closer to a product file than the owner admits. Invoices decide.
Illustrative only: same $1.6 million EBITDA. File J is 88% build-to-print with a strong plant manager. File P is 70% catalog with two distributors and a thin warranty reserve. File P can live higher in the $1–3 million synthesized cell if QoE confirms reorders. File J lives on transition and certs. Averaging them is the error.
How should the teaser split a hybrid without sounding clever?
Lead with the majority of gross profit, not the story you prefer. “Build-to-print machining with a small catalog sidecar” is honest. “Product company with some custom” is a lie if custom is 80% of GP. Buyers forgive hybrids. They do not forgive costumes.
How should tooling title be proven on each file type?
Job shop: purchase orders and tool tags that say customer-owned, plus photos of plates. Product: invoices that show you paid, drawings in your name, and a schedule of molds or fixtures that travel with the SKU. Mixed files need both columns. A teaser that says “we own the tools” without the column is a Friday fight.
If a customer paid for a mold and you capitalized it anyway, say so. QoE and 8594 both care.
Illustrative only: 12 molds, 7 customer-owned, 5 company-owned on a catalog line. Two columns in the room. No Friday surprise. The twin file that capitalized all 12 spends QoE week on title.
How should you treat private-label work that looks like a product but is not yours?
If you manufacture a brand you do not own, you are a capacity file with a specification. Switching costs may still exist. The brand is not your intangible. Do not put the customer’s logo in the teaser. Do not claim a catalog you cannot assign.
Illustrative only: 55% private-label for one brand, drawings owned by the brand. Job-shop gravity with a whale. Honest CIM. The costume “product company” dies in the first NDA package.
How should you treat a SKU that is really a service contract with a part number?
If most of the dollars are field service or programming, you are selling labor attached to a label. Price the labor dependence. Do not borrow a product multiple because the invoice has a SKU column.
A last mix habit: print a trailing-twelve GP split in the room — build-to-print versus catalog — and update it monthly. Arguments about identity end when the split is a number. Teasers can stay qualitative. Rooms should not.
Educational information only. IP ownership and tool title are counsel questions. No mix guarantees a multiple. Jason Taken is a business broker, not a patent lawyer. The longer business-models guide walks invoice identity, drawing custody, and fixture title without treating a homepage refresh as a product company.