What should a buyer demand from a manufacturing quality of earnings?
Buying a manufacturing business without that discipline is how people buy a recast. Valuation still needs a size band after the QoE. A clean report on a 9% fabricator does not mint a search-fund median.
What a plant QoE has to open
A services-firm QoE that never leaves the P&L will miss the plant. Scope should include:
- Tax returns versus internal P&L versus the job-cost file
- Add-backs with documents under every line, especially owner compensation versus a real GM package
- Scrap, rework, warranty, and first-pass yield
- Customer-level and cell-level margin
- WIP and percentage-of-completion versus the peg
- Related-party customers and vendors
- Revenue cutoff and bill-and-hold
Job costing is the operations read. The QoE is the accountant's reconstruction. If the ERP is QuickBooks plus a whiteboard, budget more hours. That cost belongs in your model.
Expect a buyer-paid QoE once EBITDA is large enough that lenders or investors require it — commonly around ~$2 million+, sometimes lower if the add-back schedule is aggressive. On a Main Street file near BizBuySell's ~$268k median SDE at 2.75x (Source: BizBuySell via Iconic), a full PE-style report can be disproportionate — unless the add-backs are loud. Then pay for a focused recast or walk.
| QoE finding | What it usually means | What you do |
|---|---|---|
| Tax, P&L, job-cost diverge | Estimating or cutoff problem | Price chip or walk |
| Owner pay add-back > local GM | Recast is a brochure | Rebuild SDE / EBITDA |
| Scrap called "startup" every month | Process cost | Put it back in |
| Hero customer at 8% margin | Concentration plus bad work | Structure or walk |
| WIP at quoted margin | Peg fiction | Rewrite method; count |
| "One-time" legal every year | Recurring | Normalize in |
| Missing documents under add-backs | Unsupportable | Delete the add-back |
BLS OEWS May 2025: machinist median $58,750 (Source: BLS OEWS). If the recast assumes you will run the shop for less than a replacement package, the QoE should say so in plain numbers.
How the report talks to price, debt, and the peg
SBA 7(a) maximum is $5 million. SOP 50 10 8 wants a 10% injection (Source: SBA). The lender will model DSCR after a replacement wage they believe. A QoE that deletes a turn does not "get explained away" in a credit memo you have not seen.
Evaluation decides whether to order the report. The LOI decides who pays and what happens to price. Do not exclusive on a vague "we'll do diligence" if add-backs are the whole story.
The buyer peg explainer and the QoE should share a WIP religion. If the QoE says job-cost is unreliable, the peg cannot pretend inventory is precise.
ISM Prices 71.1 in August 2026 (Source: ISM) will show up as a margin gap if pass-through failed. A good QoE looks at jobs quoted before the move and shipped after it.
What a QoE will not do
It will not appraise machines. It will not sample soil. It will not recertify AS9100. It will not novate a government contract. It will not set Stanford's $16 million / 6.2x / ~25% margin median as your quote (Source: Stanford GSB). That median is a search-fund profile, never a job-shop multiple.
IBBA Q1 2026: 83% of deals over $5 million had three or more offers (Source: IBBA). In that band, a QoE is table stakes. Buyers who skip it to "win" buy the add-backs the others refused.
How to scope the firm without buying a generic memo
Tell the firm it is a manufacturer. Require a job-cost sample, a physical observation of the cage if the file is inventory-heavy, and a customer-margin table. A generic add-back memo that never leaves the trial balance is how you pay for comfort and still buy scrap. Ask whether they have reconstructed JobBOSS, E2, Epicor, or a whiteboard shop before. Brand names matter less than that reconstruction skill.
Field time is not optional on a plant. Someone should see whether travelers exist, whether the CMM log is current, and whether finished goods on hold sit in the eligible bucket. That is not an appraisal. It is earnings context. The machine-list page still owns OLV. The QoE owns whether maintenance was spent.
Materiality should be stated. On a file near $268k SDE, a $15,000 item can matter (Source: BizBuySell via Iconic). On a $2.5 million EBITDA search-fund-shaped file, the same item may not. Stanford's median ~$2.5 million EBITDA / ~25% margin / $16 million price is the study cell (Source: Stanford GSB). It is never a job-shop multiple. Do not use it to waive materiality on a small shop, or to invent materiality theater on a large one.
Related-party revenue — the seller's other company that "buys" at list — belongs in the first week of the report, not the last footnote. So do bill-and-hold shipments parked in a trailer. So do jobs that invoice on percentage-of-completion the tax return never saw.
Using the report in the room without turning it into a sermon
Share the report the way the LOI said you would. A full dump to a seller who has never seen a QoE can stall a Main Street file. A one-page bridge of adjustments with support behind it usually moves price. If the seller refuses any adjustment after a documented $400,000 EBITDA kill, you walk. IBBA's 83% three-offer statistic (Source: IBBA) does not mean you must absorb a bad recast to stay in the process. It means someone else may also have a report.
Lenders will ask for the report on files that need it. SBA 7(a) still caps at $5 million with a 10% injection (Source: SBA). A credit committee that receives a QoE showing unsustainable owner add-backs will not "get comfortable" because you are enthusiastic. Conventional and ABL files are blunter: DSCR and the borrowing base both assume the earnings are real.
ISM Prices 71.1 (Source: ISM) and Fed utilization 76.0% (Source: Fed G.17) belong in the report only as context for pass-through and capacity, not as substitutes for the target's jobs. Deloitte/MI's 3.8 million / 1.9 million workforce figures (Source: Deloitte/MI) belong in the replacement-wage discussion, next to OEWS medians, not as a synergy story.
Do not ask the QoE firm to bless a multiple. Valuation is a separate judgment labeled by size band and sub-sector. The report gives you an earnings base. You still have to pick a cell that matches the company you actually have.
A composite add-back bridge — how a turn disappears
Stated EBITDA $2.2 million. Owner salary $90,000 below a replacement GM in that metro. Spouse AP "salary" $70,000 that must be replaced. Personal truck, travel, and a condo $80,000 — some documented, $25,000 not. "One-time" legal $40,000 that appeared in two of three years. Scrap called startup $60,000 that hits every quarter. Related-party revenue $120,000 at list that will not recur.
A serious report might keep $90,000 of owner-pay add-back only if the GM hire is real, delete the undocumented $25,000, annualize the legal, put scrap back, and drop the related-party. You can lose $300,000–$400,000 of EBITDA without anyone lying in a cartoon way. At 6x that is $1.8–$2.4 million of EV. At Main Street 2.75x SDE the same dollars are a full turn on a ~$268k earnings file (Source: BizBuySell via Iconic).
SBA 7(a) DSCR inside a $5 million cap (Source: SBA) uses the number after this bridge. Stanford's 6.2x / ~25% margin / $16 million median (Source: Stanford GSB) uses a different animal. It is never a job-shop multiple. IBBA's 83% three-offer statistic (Source: IBBA) does not restore deleted add-backs.
BLS OEWS May 2025 machinist $58,750 (Source: BLS OEWS) belongs on the replacement-wage line. ISM Prices 71.1 (Source: ISM) belongs on the jobs quoted before the move. Fed utilization 76.0% (Source: Fed G.17) belongs on overtime you normalized out.
Jason Taken is a business broker, not a CPA and not the QoE firm. Educational information only. No outcome is guaranteed.
Questions the engagement letter should force the firm to accept
Will they reconstruct job-cost. Will they look at WIP religion. Will they test related-party revenue. Will they visit the cage if inventory is material. What is materiality on this size band. If the firm wants a generic add-back memo, hire a different firm.
Share a one-page bridge with support, not a 90-page dump that stalls a Main Street seller. If documented earnings fall a turn and the seller refuses any adjustment, walk. Lenders will use the after-bridge number. Do not ask the firm to bless a multiple. The report gives an earnings base. You still pick a size-band cell that matches the company you have.
Field time is not optional on a plant. Travelers either exist or they do not.
HedgeStone will tell the seller in the LOI who pays for the QoE and what happens if a turn disappears. Hope is not a price formula. A generic add-back memo that never leaves the trial balance is comfort you cannot take to a credit committee.
Use the report as an earnings base, not as a multiple picker
State materiality in the engagement letter so a $15,000 unsupported add-back is treated as a finding on a $268k SDE file and as a note on a $2.5 million EBITDA file. Do not let the firm guess. Share a one-page bridge with support, not a 90-page dump that stalls a Main Street seller. If documented earnings fall a turn and the seller refuses any adjustment, walk. Lenders will use the after-bridge number. Do not ask the firm to bless a multiple. The report gives an earnings base. You still pick a size-band cell that matches the company you have. Field time is not optional on a plant. Travelers either exist or they do not. A generic add-back memo that never leaves the trial balance is comfort you cannot take to a credit committee.