What share starts to pressure the multiple?
The research file’s rule of thumb is blunt: top customer above 25–30% of sales → 0.5–1.5x multiple discount, or earn-out / concentration escrow. Label that as a broker rule, not a published dataset. A 26% customer on a five-year aerospace long-term agreement with pricing formulas is not the same fact as a 26% customer who emails releases on Thursdays.
Census CBP 2023’s 68.4% of establishments under 20 employees (Source: Census) is why Main Street CNC files so often have a “whale.” Small shops live on a few accounts. That does not make the whale free. It makes the CIM honest.
On a sde-under-1m general job shop at 2.0x–3.6x SDE, a 0.5x chip is visible money even when the dollars are small. On a ebitda-3-10m medical file at 6.5x–9.0x, a 1.0x chip is a different conversation. Concentration does not move you into another size band. It moves you inside the cell you already earned.
IBBA Pulse Q1 2026: 83% of deals over $5 million drew three or more offers (Source: IBBA). More offers help you discover whether the whale is a fear or a feature. They do not erase the whale.
How do contracts change the concentration chip?
Firm remaining term, volume commitments, and cancel-for-convenience language decide whether the 31% account is a contract or a habit. Backlog and contracts is the companion. A cancel-for-convenience clause with 30 days’ notice is not recurring revenue. An LTA that lags steel or resin is a margin risk attached to the same logo.
Switching costs help. A medical component with a validated process and ISO 13485 access is stickier than a commodity brake bracket. Quality certifications can make the customer a certificate customer: lose the audit, lose the account, even if five logos remain on the website.
Owner-held relationships cut the other way. If the VP of purchasing only takes the owner’s call, concentration and owner dependence are the same risk. Searchers notice. Stanford’s search-fund median sits on a professional-operator profile (about 6.2x EBITDA, ~25% margin; Source: Stanford GSB). A whale in the owner’s pocket fails that profile.
When is an earn-out the honest structure?
When the open item is whether a named account stays after close. Keep the earn-out measurable: gross profit on named accounts, not “EBITDA after buyer allocations.” Escrow can sit beside it. The research file’s habit on Main Street / lower LMM indemnity is 5–15% for 12–24 months — a rule of thumb, not a promise.
Selling to a strategic acquirer is the special case. Sometimes the strategic is the whale. That can clear a premium because the concentration disappears in the combination. It can also create a consent problem and a thinner buyer set if the process is exclusive too early.
Do not use an earn-out to paper over a 70% customer you refused to disclose. Buyers rebuild concentration from invoices. IBBA’s 67% of advisors seeing no material AI effect on value (Q1 2026) is a reminder that software stories do not replace a customer table.
| Concentration fact | Valuation move | Structure move |
|---|---|---|
| Top customer 15–20%, diversified rest | Stay in cell; tell the story | Usually none |
| Top customer 25–35%, firm LTA | Inside-cell discount or none if sticky | Consent; maybe small escrow |
| Top customer 25–35%, verbal releases | 0.5–1.5x rule-of-thumb pressure | Earn-out on named account |
| Top customer >40% | Cell pressure plus buyer-set shrink | Earn-out / walk / strategic-only |
| Certificate is the real customer | Treat as access concentration | Recert path; price the gap |
How should a CIM disclose the top five?
Show trailing-24-month sales by customer, percent of total, ship-to concentration if different from sold-to, remaining term, cancel rights, pricing formula, and the human relationship. Hide nothing that invoices will reveal. “Diversified industrial” is not a disclosure.
Program concentration in defense or auto can sit below 25% by logo and still be one platform. Automotive / IATF scorecards and aerospace program of record are the examples. Name the program.
Supplier-side concentration is the mirror image — see supplier concentration. A single resin or alloy source can turn a diversified customer list into a single-point failure. ISM Prices 71.1 in August 2026 (Source: ISM) is the climate for that conversation, not a multiple.
What is not a concentration problem?
A 22% customer with a transferable LTA, a second-chair salesperson, and low switching costs the other direction (they need you more than you need them) may sit at the high end of the cell. A shop with no customer above 8% and no certificate, no process scarcity, and no backlog quality may still sit at the low end because there is nothing to defend.
GF Data’s 7.2x manufacturing average (Q1 2026 recap) will not save a 44% whale on a commodity print file. BizBuySell’s 2.75x SDE median will not punish a 12% top account on a clean Main Street shop by itself.
Pick the sub-sector row. Then move inside it. Sellers: sell my manufacturing business. Buyers: buy a manufacturing business. Jason Taken is a business broker, not an appraiser. Educational information only. Counsel drafts the earn-out. No value is guaranteed.
How to rebuild concentration when the CIM is cheerful
Export invoices for twenty-four months. Map sold-to and ship-to. Roll programs that are one platform under many part numbers. Roll certificate-gated work into a certificate bucket. Compare the rebuild to the CIM’s pie chart. The delta is the finding.
A “top ten” that sums to 41% while the whale is hidden as three bill-tos is not diversification. A defense program with five primes and one platform is not five customers. Write the rebuild in the room before a buyer does it in exclusivity.
When a small whale is worse than a large one
On a $380,000-SDE shop, a 32% customer is often a relationship the owner can walk across the street. On a $7 million-EBITDA packaging converter, a 32% customer may have a procurement committee and a 180-day requalification. The rule-of-thumb band is the same. The switching-cost story is not.
Government mix can be a small logo and a large consent problem. FAR novation is not a commercial assignment. Price the clock.
Process language that belongs in the teaser
“Top customer 18%, next 11%, no certificate customer above 9%, two LTAs assignable.” That sentence is worth more than a paragraph about ‘sticky relationships.’ Backlog supplies the contract words. This page supplies the percents. Together they decide whether an earn-out is honest or decorative.
If you are selling into the whale, say so and run a process that still has an alternative. Exclusive LOI with the only customer is how price discovery dies. Strategics can be the right buyer. They should not be the only reader of the teaser unless you have already decided that.
How to write the teaser sentence that prevents a cheerful pie chart
Top customer 18 percent, next 11 percent, no certificate customer above 9 percent, two LTAs assignable. That sentence beats a paragraph about sticky relationships. Rebuild sold-to and ship-to for twenty-four months before you write it. Roll programs. Roll certificate-gated work. If the rebuild and the pie chart disagree, the rebuild wins.
A small shop whale the owner can walk to is not the same switching-cost story as a 7 million EBITDA converter with a 180-day requalification. The 25 to 30 percent rule of thumb still applies. The structure may not. If you are selling into the whale, keep an alternative reader in the process. Exclusive LOI with the only customer is how discovery dies.
A rebuild-before-pie-chart rule for every teaser
Export twenty-four months of invoices. Map sold-to and ship-to. Roll programs. Roll certificate-gated work. If the rebuild and the pie chart disagree, the rebuild wins. Write top customer, next, certificate customers, and assignable LTAs in one sentence. A 32 percent whale on a small SDE shop is not the same switching-cost story as a 32 percent whale on a large converter with a 180-day requalification. The 25 to 30 percent rule of thumb still applies. If you are selling into the whale, keep an alternative reader. Exclusive LOI with the only customer is how discovery dies.
A closer read on rebuild first
Export twenty-four months of invoices. Map sold-to and ship-to. Roll programs. Roll certificate-gated work. If the rebuild and the pie chart disagree, the rebuild wins. Write top customer, next, certificate customers, and assignable LTAs in one sentence. A 32 percent whale on a small SDE shop is not the same switching-cost story as a 32 percent whale on a large converter with a long requalification. The 25 to 30 percent rule of thumb still applies. If you are selling into the whale, keep an alternative reader. Exclusive LOI with the only customer is how discovery dies.
Why twenty-five to thirty percent is a conversation and sixty percent is a structure
A customer at about a quarter to a third of sales is a diligence topic: contract assignability, switching cost, and whether the relationship sits with the owner or the quality lead. A customer at sixty percent is usually a deal-structure topic: earn-out, holdback, or a walk. Those are different responses. Treating both as “we should diversify someday” is how files reach exclusivity with no plan.
The published cells already assume a book that can survive ordinary customer noise. They do not assume a single OEM who can zero the plant with one portal change. Sitting at the low end of the current size-band and sub-sector cell is the mild response. An earn-out tied to that customer’s trailing twelve is the structured response. Pretending the pie chart will look different next year without a second customer in backlog is not a response.
Match gated certificate sales to this page. An AS9100 customer who is also the whale is two risks in one name. A medical customer who owns the tooling is a third. Write the name once and list the risks. Do not split them across three CIM chapters so nobody adds them up.
BizBuySell’s manufacturing median close of 223 days in 2025 is already long. Concentration fights make it longer. Stage the top-ten sales by year, the contracts, and the owner’s role in each relationship before the first management meeting. Buyers who discover the whale in week five assume you hid it. That assumption is a chip even when you did not hide it.