Analyzing customer concentration and backlog

How should a buyer measure customer concentration and backlog?

Buying a manufacturing business without those schedules is how people buy a single purchase order. Valuation already said documented backlog supports a multiple and concentration discounts it. This page is how you test that sentence.

Build the concentration file the way a lender will

Request:

• Sales by customer for the last 36 months, monthly or quarterly
• Related legal entities rolled up (three "customers" that share a treasury)
• Part-number concentration inside the top accounts
• Gross profit, not just revenue, by customer if the ERP can do it
• Scorecards, past-due dollars, and open claims

A one-year snapshot hides a program that already left. A revenue-only snapshot hides a customer who buys a lot at 8% margin and a customer who buys less at 32%. Job costing is the cousin of this analysis.

Broker practice treats top-customer share above ~25–30% as a 0.5–1.5x multiple discount or an earn-out / concentration escrow. That is a rule-of-thumb, not a statute. Below that band, still read the next three names. Five customers at 12% each is a different risk than one at 28% and a healthy tail.

SBA 7(a) still caps at $5 million with a 10% injection under SOP 50 10 8 (Source: SBA). A credit committee will not ignore a 40% customer because you like the multiple. Conventional and ABL files are harsher: eligibility and DSCR both assume the account stays.

Backlog is a legal category, not a slide

Firm. Purchase orders and releases you can ship under current terms.

LTA / blanket. Helpful if pricing, volume bands, and termination language are readable. Cancel-for-convenience still discounts the multiple. Price the convenience.

Forecast. A planning number. It is not collateral. ISM's August 2026 backlog index printed 51.8 with PMI 54.6 (Source: ISM). That is national climate. It does not make the target's forecast real.

Past-due. A pile of broken promises is not a blessing. On-time delivery versus promise date tells you whether the book is a gift or a hole.

Ask whether quoted lead times assume one shift or heroic overtime. Fed G.17 utilization was 76.0% in July 2026 (Source: Fed G.17). A shop claiming 95% utilization on overtime is telling you the backlog depends on people who may already be tired.

How structure should look in the LOI

If concentration is the open item, the letter of intent should freeze the measurement: named accounts, gross-profit definition, and the period. Earn-outs that say "EBITDA after buyer allocations" become fights. Earn-outs that say "gross profit on Account A and Account B" can be measured.

Do not model the earn-out as a lender repayment source. Most lenders ignore it until it is earned.

The working-capital peg intersects backlog when long jobs sit in unbilled WIP. You can have a beautiful book and still overpay if you buy incomplete jobs at full inventory value. Count the physical. Write the POC method before close.

Size band still governs the multiple you apply after the schedule

Main Street manufacturing in BizBuySell's 2021–2025 set cleared at 2.75x median SDE (Source: BizBuySell via Iconic). A concentrated SDE shop is still an SDE shop. Do not "fix" a 35% customer by pasting a lower-middle-market EBITDA cell onto it.

For EBITDA $1–3M precision / plastics, the site table sits at 5.0x–6.5x; AS9100 / ISO 13485 at 6.0x–8.0x — advisor-survey and reported-deal ranges, not an appraisal. Concentration moves you inside the cell or out of it. It does not move you into Stanford's $16 million / 6.2x / ~25% margin search-fund median (Source: Stanford GSB). That median is never a job-shop multiple.

IBBA Q1 2026: 83% of deals over $5 million attracted three or more offers (Source: IBBA). Competitive processes still produce concentration schedules. Buyers who skip them to "win" buy the risk the others priced.

A quality of earnings should test whether the top accounts' margins are real after job-cost, not only whether revenue landed. If the QoE shows the hero customer is the low-margin customer, your structure was pointed at the wrong number.

Customer calls belong late, on the seller's script, after paper survives. You are not there to announce a sale to a buyer who can re-source in a quarter.

Part-number and program risk inside a "fine" top-10

A top customer at 18% can still be one program at 16% and spare parts at 2%. If the program ends, you did not have diversification. Ask for sales by part family inside the top accounts. EMS and automotive files hide this. Aero LTAs hide it behind a logo. Medical device components hide it behind a SKU the OEM can dual-source.

Related-party customers — the seller's other company, a relative's distributor — are concentration wearing a costume. Roll them up. Then decide whether that demand survives the sale. If it does not, the QoE should treat it as non-recurring.

Credit quality of the top book belongs next to share. A 22% customer who pays at 75 days and disputes every third invoice is a working-capital problem. Put that aging next to the peg. ABL eligibility and the peg average both suffer.

How backlog talks to capacity and to the LOI

If firm backlog exceeds what one shift can ship before the promise dates, you are buying overtime or disappointment. Fed utilization 76.0% (Source: Fed G.17) is not your bay. Your roster is. If the seller's answer is "we always catch up," read on-time delivery. Past-due dollars are negative backlog.

The LOI should freeze how an earn-out measures named accounts and whether a lost program is a walk right or a chip. Do not leave "we'll know it when we see it" in a competitive process. IBBA's 83% three-offer statistic on deals over $5 million (Source: IBBA) means someone else will write a cleaner clause.

SBA 7(a) still caps at $5 million with a 10% injection (Source: SBA). A credit committee will model the top account leaving. If your DSCR dies in that scenario and you have no structure, you do not have a loan. Combined 7(a)+504 can reach $10 million (Source: SBA). It does not diversify the book.

ISM new orders 53.7 and backlog 51.8 in August 2026 (Source: ISM) describe national book-to-bill climate. They do not make the target's forecast firm. Stanford's $16 million / 6.2x / ~25% margin median (Source: Stanford GSB) often sits on stickier revenue. It is never a job-shop multiple and never a reason to skip the 36-month schedule on a 14-person shop.

A concentration exhibit that belongs in the LOI appendix

Table: rank, trailing-12 sales, trailing-36 sales, gross profit if available, firm backlog, forecast, scorecard color, change-of-control clause yes/no. Totals that foot to the P&L. Related entities rolled up. SBA 7(a) credit files inside $5 million (Source: SBA) will ask for a version of this. Combined 7(a)+504 can reach $10 million (Source: SBA). Same ask.

Main Street 2.75x SDE (Source: BizBuySell via Iconic) still deserves the exhibit. Stanford $16 million / 6.2x / ~25% margin (Source: Stanford GSB) still deserves it. The median is never a job-shop multiple. IBBA's 83% three-offer statistic (Source: IBBA) rewards the exhibit. ISM backlog 51.8 (Source: ISM) is climate. Fed utilization 76.0% (Source: Fed G.17) is climate. Your table is underwriting.

Jason Taken is a business broker. Educational information only. Contract interpretation is counsel's lane. No outcome is guaranteed.

Questions a concentration schedule should answer before exclusivity

Which legal entities are actually one buying group. Whether trailing-36 sales tell a different story than trailing-12. Whether gross profit on the hero account is the worst margin in the book. Whether firm backlog on that account covers the first two note payments if releases pause. Whether a change-of-control clause sits in the quality agreement. Whether the salesperson who holds the account is staying. Those answers belong in the buybox, not in a speech about loyalty.

If the schedule cannot be produced, you do not have a commercial file. You have a logo. Wait for the extract or walk. A book that is 40% one program with cancel-for-convenience language is a structure problem even when trailing-12 "looks diversified" because three related entities were listed separately. Roll them up. Then decide discount, earn-out, escrow, or exit.

HedgeStone will not send a seller into a process with a buyer who treats "they would never leave" as diligence. Write the exhibit. Put it behind the LOI. Then make the late, scripted customer call — not the week-one leak.

HedgeStone will treat a missing 36-month schedule as an incomplete commercial file, not as a paperwork delay. If you want the room, ask for the exhibit first. Related entities that share a treasury are one customer even when the CIM lists three logos. Roll them up before you decide the file is "diversified." The habit table — not a 2% per-point law — lives on the customer concentration discount calculator. How to pull on-time delivery, scrap, utilization, and a nameless concentration exhibit from the live file is the benchmarks guide.