The deal that kills you usually doesn't look like a disaster. It looks fine. The numbers go up and to the right, the seller is friendly, the broker is moving fast. The red flags when buying a business aren't loud — they're small things in the data room you talk yourself out of because you already want the deal to work.
I've watched smart people fall for a clean-looking business and inherit somebody else's problem. So this is the rundown I wish more buyers had in front of them: the real deal-killers, why each one matters, and what to actually do when you hit it. Not a fear list — a checklist for keeping your money.
Most of these won't show up as a number on the P&L. That's the point. The dangerous red flags when buying a business live in the gap between what the seller tells you and what the business actually is.
1. The business IS the owner
This is the one I'd put at the top of every list. If the owner personally holds every key customer relationship, makes every sale, knows every vendor by name, and is the reason the thing runs — you're not buying a business. You're buying a job that fires its best employee on closing day.
It's not a soft concern, it's a priced one. Owner-dependent businesses sell for 30 to 50% less than comparable owner-independent ones, and the discount is measured in full turns of EBITDA. What to do: map exactly what the owner does in a week, then ask who does it after they leave. If the honest answer is "nobody yet," you price that in, you structure a long transition with a holdback, or you walk.
2. Customer concentration
One customer is 40% of revenue. That customer has a relationship with the owner, not the business. The owner retires, the relationship walks out with them, and you just bought a company that lost almost half its revenue in month one.
The rough rule the pros use: any single customer over 10% of revenue is worth a hard look, top-five over 25% is a flag, and a single customer over 30% is where many lenders and PE firms pass entirely. What to do: pull revenue by customer for 3 years, check the contracts for auto-renewal and term, and find out whether the relationship is institutional or personal. If it's personal and concentrated, that's a case for a holdback — the seller gets paid the last slice only if the customer stays. (An earnout does the same job on a non-SBA deal; SBA 7(a) loans don't allow them.)
3. A decline dressed up as "flat"
"Revenue's been steady" is the most expensive sentence in a CIM. Steady can mean genuinely stable, or it can mean three years of slow bleed that nobody graphed because the graph would tell on them.
Pull the monthly revenue, not the annual. Annual hides the trend; monthly shows you the slope. A business that did $2M, $1.9M, $1.8M isn't flat — it's dying slowly, and you're being asked to pay a multiple as if it'll hold. What to do: chart it yourself, separate price increases from real volume growth, and if the trend is down, the multiple has to reflect a declining asset, not a stable one. This is half of what a careful read of the CIM is even for.
4. Books that don't reconcile
The P&L says one thing. The tax returns say another. Nobody can explain the gap. That's not an accounting quirk — that's the whole deal telling you something.
Tie the P&L to the tax returns to the bank statements. They should agree, or there should be a clean reason they don't. Be extra careful with a cash-heavy business — a bar, a laundromat, a restaurant — that shows suspiciously clean, perfectly consistent numbers. Real businesses are lumpy. Numbers that are too smooth are sometimes numbers that were made up. What to do: a quality-of-earnings analysis exists precisely for this. If the books won't reconcile and the seller can't explain it, that's not a discount — that's a walk.
5. Add-backs that don't pass the laugh test
Add-backs are legitimate — the owner's above-market salary, the truly one-time legal bill, the personal expenses run through the company. That's how seller's discretionary earnings gets calculated, and it's fair. The red flag is when the add-back list is a wish list.
When you see the owner's car, their spouse's no-show salary, "marketing we won't need," and a dozen other adjustments all stacking SDE higher, the seller isn't normalizing earnings — they're inventing them. Lenders know this game: one deal team reports applying a 15 to 50% discount to broker-reported SDE before they'll underwrite the debt. What to do: make the seller document every add-back with a receipt. The ones that survive are real. The ones that don't come straight out of the price.
6. A lease that won't transfer (or runs out)
For a location-dependent business — retail, restaurant, anything where the address is the asset — the lease can quietly be the whole deal. Two ways it bites: it has 14 months left and no renewal option, or assigning it to you requires landlord consent the landlord can use as leverage.
Most commercial leases require the landlord's sign-off to assign, and landlords routinely use that consent right to bump the rent, demand a personal guaranty, or grab a piece of the sale. What to do: read the lease early — remaining term, renewal options, escalators, the assignment clause — and start the landlord conversation 60 to 120 days before close, not the week of. A business you can't keep in its building is not the business you think you're buying.
7. Key employees who might walk
Sometimes the owner isn't the single point of failure — it's the master tech who's been there 18 years, or the salesperson who personally owns the top accounts. They didn't sign the deal, they may not even know it's happening, and they have no reason to stay for a new boss.
Lose the wrong two people in the first 90 days and the business you valued isn't the business you have. What to do: identify the people the business genuinely can't run without, find out their tenure and comp, and build retention into the deal — stay bonuses, new agreements signed at close. If you can't talk to them before closing (you usually can't, for confidentiality), at minimum stress-test what happens if they leave.
8. Undisclosed or pending litigation
A lawsuit the seller "forgot" to mention is two red flags in one: the liability itself, and the fact they hid it. If they're quiet about a pending claim, ask yourself what else didn't make it into the data room.
What to do: run litigation and lien searches, ask directly and in writing about pending, threatened, and recent claims, and put representations and warranties in the purchase agreement so a hidden liability that surfaces later is the seller's problem, not yours. An indemnity holdback — money parked in escrow for a year or two — is how you make those reps mean something.
9. A capex bomb hiding in plain sight
The numbers look great partly because the owner stopped spending on the business years ago. The roof, the HVAC, the production equipment, the trucks, the software nobody's updated since 2019 — it all still runs, until it doesn't, and the bill lands on you six months in.
Deferred maintenance is an owner inflating profit by not reinvesting, and it's invisible on a P&L. What to do: walk the place, get the equipment list with ages and condition, ask when the big assets were last replaced, and get the deferred capital cost estimated and subtracted from the price. A business that's been milked dry isn't cheap — it just looks cheap until the repair invoices show up.
10. A seller who stalls on documents
You ask for the bank statements. You get them in three weeks, partial. You ask for the customer list. "Let me get back to you." The pattern matters more than any single delay: a seller who drags on diligence is either disorganized or hiding something, and both cost you.
Clean businesses tend to have clean data rooms. The owner who can hand you reconciled books, contracts, and tax returns in a week is telling you something good about how the place is run. What to do: treat document stalls as data. Set deadlines, and if reasonable requests keep getting slow-walked, slow your own roll — that resistance usually means the answers aren't flattering.
More flags worth a hard look
Not every red flag needs its own section. A few more that quietly kill deals:
- An asking price disconnected from reality. For context, the median small business that actually sold in 2024 went for about $345,000 at roughly 2.6x cash flow. If a seller is quoting a wild multiple with no story to justify it, that's a signal about how the rest of the negotiation will go.
- The "one-time" expense that shows up every year. Legal fees, equipment repairs, owner bailouts — if a "non-recurring" cost recurs across all three years of financials, it's not one-time. It's an operating expense the seller is pretending isn't real to pad the earnings.
- A dying industry or a single platform dependency. A business that's 100% dependent on one Amazon account, one Google ranking, one regulatory carve-out, or one referral partner doesn't own its own demand — the platform does. One algorithm change or rule rewrite and the revenue is gone, and you had no vote.
What to actually do when you hit a red flag
Here's the part people get wrong: a red flag is not automatically a no. Most deals have at least one. The skill isn't avoiding flags — it's pricing them. You've got three honest moves, and the right buyer knows which to use.
Price it in. The flag is real but quantifiable — deferred capex, a soft add-back, a declining trend. Subtract the cost, lower the multiple, and if the seller won't move, you have your answer.
Structure around it. The risk is real but might not hit — customer concentration, owner-dependence, key-employee flight. Don't pay full price for a risk you're carrying. Shift it back onto the seller with a holdback, an escrow, a longer transition — or an earnout on a non-SBA deal (SBA 7(a) loans don't allow them). They get paid in full only if the thing they swear is fine turns out to be fine.
Walk. Some flags don't price and don't structure — books that won't reconcile, hidden litigation, a seller who lies. There's no discount that makes fraud a good deal. The best buyers I know are good at this not because they find perfect businesses, but because they're willing to walk from the wrong one. There's always another deal. There isn't always another bank account.
None of this works if you can't see the deal clearly in the first place, which is the whole reason to slow down and run a real due diligence checklist before you fall in love with a number. The deal you walk away from clear-eyed beats the one you talk yourself into — and spotting these flags is one piece of the whole process of buying a business.
DealStratum helps you find and source a business to buy — on-market and off. It's not a broker, a lender, or a financial advisor. Nothing here is investment or financial advice.
Sources
- Succession Thinking — the key-man discount on owner-dependent businesses
- Beancount.io — customer concentration thresholds and the 10% rule
- Regalis Capital — discounts applied to broker-reported SDE
- Bean, Kinney & Korman — assignment and consent standards in commercial leases
- BizBuySell 2024 Insight Report (via Small Business Trends) — median sale price and multiple
DealStratum · AI Deal Screening
Upload a CIM, get the numbers auto-extracted.
Drop in the PDF. DealStratum pulls SDE and the add-backs, charts the revenue trend, flags customer concentration, and hands you the implied multiple — with a plain-English read on what’s worth a second look.