Summary
A manufacturer’s books are a job-cost system wearing a tax-return jacket. Census CBP 2023 counted 284,452 U.S. manufacturing establishments, and 68.4% have fewer than 20 employees (Source: Census CBP). Most of those shops keep a cash-basis or hybrid tax file and a job boss that tells a different margin story. The work of this page is to read both without inventing a third.
Start at the tax return. Rebuild the internal P&L so it maps. Then open jobs. Analyzing job costing and gross margin is the diligence pass. SDE vs EBITDA for manufacturers is the earnings-base fork. Add-backs in manufacturing is the survival test. Manufacturing business valuation is where those earnings meet a size band. This guide is the statement map that sits under that table.
Main Street manufacturing in BizBuySell’s 2021–2025 set cleared at a 2.75x median SDE on about $268k of earnings (Source: BizBuySell / Iconic). Stanford’s 2024–25 search median sat near $16 million, about 6.2x EBITDA, and about 25% margin (Source: Stanford GSB). Those prints use different statements. Blending them is how a job-cost P&L gets a multiple it cannot support.
What does a job-cost P&L show that a tax return hides?
A tax return answers the IRS. A job-cost P&L answers whether the shop made money on the work it promised. Quoted margin versus actual, by customer and by cell, is the first extract. Scrap, rework, and first-pass yield tell you whether the quality certificate is a wall hanging. Unabsorbed overhead tells you whether the manuals are being carried by the 5-axis cell. On-time delivery versus promise date tells you whether backlog is a blessing or a past-due pile.
ERP and job-costing systems in diligence is the systems page. If the ERP cannot produce won-job versus actual, budget a quality of earnings that has to rebuild packing lists. That cost belongs in the model. A whiteboard estimating file is not “entrepreneurial.” It is a reconstruction project.
Owner-estimators often run personal expenses through cost of goods or through a catch-all “shop supplies” account. Those lines do not disappear because the tax return was accepted. They become add-back candidates or they become a habit the buyer will inherit. The job-cost P&L is where you see which.
Overhead absorption is the quiet fight. A shop that absorbs burden only when the owner remembers to close jobs will show a fat month and a thin month that have nothing to do with demand. A buyer who annualizes the fat month is buying a closing habit. A seller who annualizes the thin month is hiding a good cell. Read twelve months of job closes, not the teaser month.
| Statement | Question it can answer | Question it cannot |
|---|---|---|
| Tax return | What was filed, and which method | Whether job 4412 made money |
| Internal P&L | How management grouped the year | Whether WIP is a method or a plug |
| Job-cost file | Quoted versus actual by job | What the IRS accepted |
| Balance sheet | AR, inventory classes, AP, debt | Whether inventory exists in the cage |
| Statement of cash flows | Whether earnings became cash | Whether the next job is priced |
How should WIP be read on a manufacturer?
WIP is incomplete work measured by a method. It is not a balancing figure that makes the balance sheet land on a hoped-for number. Analyzing inventory and WIP classes raw, WIP, finished, consigned, and customer-owned. This page asks how the dollars got into the WIP cell.
Ask for the method in writing. Hours times a standard? Percent complete against a quote? Material issued plus a margin assumption? Each method can be honest. Each method can also park losses in the cage. Valuing incomplete jobs at quoted margin is hope. Valuing them at cost with no estimate-to-complete is how a loser stays an asset. The physical count at close is a fact. The method is the argument that has to be finished before the working-capital peg is written.
Unbilled first-article hours are not finished goods. Customer-owned tooling is not yours to count. Consigned resin or steel is not yours to count. Progress-billed aerospace parts can sit as a receivable and still have WIP behind them. A food co-packer’s palletized finished goods are a different animal. Do not copy a retail inventory definition onto either file.
Age the cage. Material that has not moved in a band you would not finance is an excess-and-obsolete candidate. A seller who calls aged WIP “just in time for the next release” needs the release in the backlog, not in a speech. A buyer who skips the physical is purchasing rust at carrying value.
When does percentage of completion beat completed contract?
Percentage of completion recognizes revenue and cost as the job progresses, when estimates are reliable and the buyer will inherit unbilled progress. Completed contract waits until the job is done. Long-cycle tooling, contract manufacturing with milestones, and government work often live on progress logic. Short-cycle job shops often live on completed jobs and a WIP residual that should be small.
The fight is not which method sounds more sophisticated. The fight is whether the method in the CIM matches the method in the tax file and the method the peg will use. A shop that files completed contract and sells percentage-of-completion earnings is selling a timing difference. A shop that files percentage of completion with sloppy estimates is selling optimism. Write the map before exclusivity. Quality of earnings for buyers rebuilds that map. Surviving quality of earnings as a seller is the binder habit: every percent-complete claim tied to a traveler, a shipper, or an engineer’s estimate-to-complete.
Change orders and claims need their own column. Revenue that exists only as “the customer will approve it” is not earned. Cost that exists only as “we will recover it” is not inventory. A QoE that has to reconstruct change-order hygiene is expensive. That cost is information.
| Method | When it fits | How it fails a deal |
|---|---|---|
| Percentage of completion | Long jobs, reliable estimates, inherited unbilled work | Quoted margin parked as WIP |
| Completed contract | Short-cycle jobs that close when they ship | A year stuffed with unclosed losers |
| Hybrid / job-close habit | Many small jobs plus a few long ones | No written rule for which job uses which |
| Progress billing | Contract milestones | Billing ahead of cost with no estimate-to-complete |
How do SDE and EBITDA part ways on the same shop?
Seller’s discretionary earnings is the pre-tax cash an owner-operator can take: net income plus owner pay, interest, non-cash charges, and documented personal or one-time items that will not transfer. Adjusted EBITDA is earnings before interest, tax, depreciation, and amortization after a replacement general manager and other normalizations an institutional buyer will demand. They are different bases. SDE vs EBITDA for manufacturers is the fork page. This guide only shows where the statements force the fork.
If the owner still quotes, programs, or runs the floor, start at SDE. Subtract a replacement package before you speak EBITDA. Discretionary earnings that still contain the owner’s wage are not institutional earnings at the same headline. The replacement seat has to show up as a number, not as a shrug.
If a real general manager is already paid, start at adjusted EBITDA. Searchers, sponsors, and most strategics will recast there whether you like the language or not. Stanford’s 2024–25 search median sat on a ~25% margin profile at about 6.2x (Source: Stanford GSB). That print does not attach to an owner-estimator SDE file. BizBuySell’s 2.75x manufacturing median attaches to SDE (Source: BizBuySell / Iconic). Do not cross-walk the two with a shrug.
SBA credit files often live on cash the lender will treat as available after a replacement wage. The 7(a) box is still $5 million with a 10% injection under SOP 50 10 8 (Source: SBA). A pretty EBITDA slide does not move that box. A dirty job-cost file does not get cleaner because someone labeled it EBITDA.
| Line | SDE treatment | Adjusted EBITDA treatment |
|---|---|---|
| Owner draw / W-2 | Add back, then compare to a replacement | Replacement GM already in the number |
| Spouse AP wage | Add back only the excess above market | Normalize to market and keep the seat |
| Interest | Add back | Add back |
| Depreciation | Add back; then test unfunded iron | Add back; then test unfunded iron |
| Personal truck / travel | Add back with invoices | Add back with invoices |
| Deferred spindle rebuild | Not an add-back; a cash chip | Not an add-back; a cash chip |
Which add-backs survive a quality-of-earnings rebuild?
Documented excess owner pay above a market replacement. Isolated legal, flood, or move costs with invoices and dates. Personal expenses that will not transfer and that leave a paper trail. Add-backs in manufacturing is the longer survival list. This page only says the statement has to support the claim.
What dies: recurring scrap called “unusual,” inventory step-ups without a physical, unfunded maintenance called earnings, future tariff givebacks, “I could hire someone cheap,” and synergies. A crashed spindle is not EBITDA. A lapsed certificate is not a growth add-back. PPP-era items, if they still appear, need dates and a run-rate test.
Quality of earnings for buyers is not optional once add-backs could move a turn. Tell the seller in the letter who pays, which firm, and how long exclusivity lasts if the report slips. If the report kills a slice of earnings, the price moves or the deal dies. Hoping the lender will “see it differently” is not a plan. A CIM recast is a conversation starter. It is not a QoE.
After the instrument is chosen, purchase-price allocation and Form 8594 split iron, inventory, and goodwill under IRC § 1060. That allocation does not change historical earnings. It does change the first tax return the buyer files. Inconsistent Class V versus Class VI/VII numbers are how audits start. A CPA prepares the form. Jason Taken will not.
Worked contrast: two recasts of one year
Illustrative only. Recast G starts with a tax return, maps the internal P&L, and ties job-cost closes to shipments. WIP is hours-and-material with an estimate-to-complete. Owner pay is compared to a replacement GM already identified. One flood invoice is added back. Deferred rebuilds stay a cash chip. The earnings base is named SDE or EBITDA on purpose. A buyer can argue the cell. The buyer cannot argue that the statements were three different companies.
Illustrative only. Recast H starts with a teaser EBITDA, adds the owner’s wage back without subtracting a replacement, parks losers in WIP at quoted margin, and calls overdue iron “buyer maintenance.” Percentage of completion and completed contract appear in the same paragraph with no map. That recast will not survive a QoE. It may not survive a competent lender. BizBuySell’s 223-day manufacturing median close is long enough for that fight to consume exclusivity (Source: BizBuySell).
The statements page exists so Recast H is recognized on the first read.
What should you ask the CPA and the controller next?
Ask which revenue method was filed. Ask how jobs are closed. Ask who can produce won-job versus actual by customer and by cell. Ask how WIP is measured and how often it is counted. Ask which add-backs have invoices. Ask whether the balance sheet classes raw, WIP, and finished. Ask whether customer-owned stock is on the books at all. Ask whether the next audit or quality-of-earnings team will be handed a binder or a story.
Educational information only. This page is not accounting advice, tax advice, or an appraisal. Confirm method, allocation, and lending questions with a CPA, M&A counsel, and the licensee the forum asked for. No earnings figure on a CIM is guaranteed.
Bring the three statements to a booked intro. After that conversation, email jason.taken@hedgestone.com. Jason Taken is a business broker with HedgeStone Business Advisors, not a CPA.