Which plant risks are actually earn-out-shaped?
An earn-out is a tool for a measurable open item. Top-customer concentration above a 25–30% rule-of-thumb band. A long-term agreement that still needs a change-of-control consent. Backlog that is real on paper and unreleased on the floor. A quality registration whose next audit sits after close. A key estimator who has not yet signed a stay package. Those are earn-out-shaped. "We disagree on the multiple" is not. That is a price fight.
Concentration is the usual driver. A shop that ships 38% to one OEM can still sell. Buyers haircut the multiple, escrow a slice, or write an 18-month GP test on that account. Contract assignment and consents still has to happen. An earn-out does not create a novation under FAR 42.12 and does not replace a customer letter.
Quality-gated files belong here too. If first-pass yield or on-time delivery is the reason a buyer will not pay the headline, write those numbers from the same reports the plant already runs. Do not invent a new KPI the ERP cannot print.
| Open item | Metric that can be audited | Metric that will fight | Usual horizon |
|---|---|---|---|
| Whale account | Named-account GP from invoices | Company EBITDA after allocations | 12–24 months |
| Unreleased backlog | Conversion of dated releases | "Booked" dollars on a wall | Through the last firm release |
| Quality hold history | First-pass yield or escape rate | A new dashboard the buyer designs | Through the next audit |
| Consent pending | Consent received, then a GP tail | "Relationship strength" | Until the letter arrives |
| Owner-quoted work | GP on jobs the second chair quotes | Hours the seller "helped" | Training window |
| New cell / program | Units shipped on a named program | A hope forecast | Program ramp |
Deloitte and The Manufacturing Institute's 3.8 million / 1.9 million workforce figures (Source: Deloitte/MI, 2024) describe how hard a second chair may be to hire. They do not create an earn-out. They explain why a stay package and a metric sometimes travel together.
How do you write gross profit so a successor cannot starve it?
Name the accounts or the SKUs. Define GP as invoice revenue minus the job-cost elements the plant already uses — material, outside process, and direct labor — not minus a new corporate allocation, a management fee, or "synergy." Say whether scrap and rework on those jobs stays in the metric. Say whether work the buyer moves to another plant still counts. Say who owns the customer relationship and who may re-price.
Set a volume floor so a buyer cannot politely lose the account and call the earn-out zero. Set a cap so the buyer can live with an upside surprise. Set information rights: quarterly packs, invoice inspection, a dispute accountant. Set a period that matches the contract or the program, not a hope decade.
Quality of earnings explained will test the baseline. If QoE kills $90,000 of vibe add-backs and restates GP on the whale, rewrite the earn-out now. The peg can also move inventory methods that change GP. Write the earn-out after those two documents, or write it with an automatic reset to the QoE number.
IBBA Pulse Q1 2026 reported that 83% of deals over $5 million attracted three or more offers (Source: IBBA). That is a process fact, not a promise your file will. It is also leverage: a vague EBITDA-after-allocations earn-out is easier to refuse when another LOI is sitting on the desk with cash.
How do SBA, searchers, and size-band multiples treat contingent price?
Many lenders ignore earn-outs as a repayment source. SBA 7(a) still caps at $5 million with a 10% injection (Source: SBA). Combined 7(a)+504 can reach $10 million for fitting borrowers (Source: SBA). None of those products let you underwrite a hope payment as if it were amortizing debt. If the credit memo needs the earn-out to clear DSCR, the structure is wrong.
Search-fund files that inhabit Stanford's 2024–25 median neighborhood — about $16 million purchase, ~6.2x EBITDA, ~25% margin (Source: Stanford GSB) — still use contingent price on the right concentration files. That median is never a job-shop multiple. A 9% fabricator does not mint 6.2x because an earn-out exists.
Main Street manufacturing in BizBuySell's 2021–2025 set cleared at 2.75x median SDE on about $268k earnings (Source: BizBuySell / Iconic). On an sde-under-1m general CNC file inside the 2.04x–3.59x interquartile, a $70,000 earn-out is a visible piece of a turn and may not be worth the administration. On an ebitda-3-10m precision or aerospace file talking synthesized 5.5x–7.5x or 6.5x–9.0x EBITDA bands, a $400,000 GP tail on two named programs can be the difference between a signed LOI and a walk.
What happens when the buyer can move the work or change the costing?
If the buyer already owns another plant, they can starve the metric by transferring the whale. Write whether transferred work still counts at the historical GP rate. If they change from job-cost to a standard-cost system, write which file governs the earn-out. If they raise price and lose volume, the floor you wrote is the protection. If they cut price to win a corporate volume deal, say whether the GP test still uses the pre-close price list.
Sellers who stay as paid consultants should not be able to manufacture GP by quoting unprofitable work just to hit a threshold. Buyers should not be able to refuse reasonable quotes on the named account to miss the threshold. Those two sentences belong in the schedule.
Jason Taken will not draft the schedule. He will keep the CIM from treating contingent price as cash.
What does an illustrative manufacturing earn-out look like when it is collectible?
Illustrative only — not a listing.
File A: top customer 34% of T12 sales. Headline on an ebitda-1-3m precision shop inside a synthesized 5.0x–6.5x band. $350,000 earn-out over 18 months on that account's GP, definition matching the job-cost extract QoE already tested, volume floor at 80% of T12 units, cap at $350,000, quarterly packs. Ordinary. Collectible if the book holds and the consent arrives.
File B: "EBITDA above $2.1 million after buyer allocations." No named accounts. Buyer adds a $180,000 management fee and a new ERP allocation. Seller collects zero and spends legal fees. Refuse this shape in the LOI.
File C: backlog conversion of $1.2 million dated releases. Earn-out pays as those releases ship at historical GP, 12-month sunset, cancel-for-convenience clauses disclosed. The consents page still has to move the paper.
BizBuySell's 223-day 2025 manufacturing median close (Source: BizBuySell) is a planning number, not a guarantee. Use part of that calendar to negotiate the metric, not the week of signing.
When should you take a lower cash number instead of a prettier headline?
When the whale can leave and the earn-out is the entire premium. When you need a December wire for a house or a tax year. When the measurement system is a junk drawer. When the buyer will not show invoice-level GP. When standby paper and an earn-out would both sit on the same tired seller. A lower cash number from a capitalized party can beat a higher headline that never arrives. The column scorecard is earn-out vs lower cash price.
Rollover is a different instrument. See rollover equity. Do not call equity an earn-out.
Educational information only. Counsel drafts. A CPA owns tax character of contingent payments. No metric guarantees collection. Jason Taken is a business broker with HedgeStone Business Advisors.
How should the LOI treat an earn-out so exclusivity does not freeze a blank?
Write the metric family in the letter of intent: named-account GP, backlog conversion, or a named quality statistic. Write the dollar cap. Write the horizon. Write that the definition will match the job-cost file quality of earnings tests. Leaving "earn-out TBD" inside a long exclusive is how you spend week ten arguing about EBITDA allocations.
Sellers who already know the whale is the issue should put a proposed metric in the CIM instead of waiting for the buyer to invent one. Buyers who need the paper to stretch a multiple should say so. Stretching a multiple with a metric nobody can audit is how both sides waste a 223-day planning calendar (Source: BizBuySell).
Information rights start in the LOI too. If the seller will not share invoice-level GP on the named account under NDA, they are not ready to earn-out that account. If the buyer will not agree to packs after close, they are not ready to pay one.
Census CBP 2023 counted 284,452 manufacturing establishments, 68.4% under 20 employees (Source: Census CBP). Most of those shops can run a simple named-account GP test. They cannot run a private-equity-style EBITDA bridge. Match the instrument to the book.
A last manufacturing-specific trap: warranty and returns on the named account. If the earn-out ignores credits issued after shipment, GP is inflated. If it includes every customer complaint the buyer chooses to credit, GP can be starved. Write the credit rule. Tie it to the plant's existing warranty policy, not to a new one the buyer invents in month two.