The number of structural patterns that account for the majority of LBO post-close underperformance in the lower middle market. All five are detectable before close. None of them require proprietary information. All of them leave clear evidence in a well-constructed data room — if the person reading the data room is looking for them.
Dealithic deal analysis · 2025 · Based on post-close performance dataThe deals that underperform don't fail because of market conditions or bad luck. They fail because of structural patterns that were present in the data room before close and were not weighted correctly. Here is a field guide to the five that matter most.
The due diligence process is designed to build conviction. It is organized to answer the question: why should we do this deal? That framing — confirmatory, not adversarial — is the source of most LMM post-close underperformance. The right question, asked before any deal goes to IC, is the inverse: what would have to be true for this deal to fail? The five signals below are the most reliable answers to that question.
These are not novel observations. Every experienced PE practitioner knows customer concentration is dangerous, management depth matters, and EBITDA normalization is a source of post-close disappointment. What is less discussed is the specific threshold at which each pattern becomes a structural problem rather than a manageable risk — and the data room evidence that makes it detectable before you have spent $400,000 on diligence fees and a failed close.
A single customer representing 25%+ of revenue is not a concentration problem — it is a binary event risk. When that customer goes away, so does the EBITDA thesis. This signal appears benign in upcycle diligence because the customer relationship looks stable. It looks different in a downturn, a competitive shift, or the 12 months after ownership transition when management attention is divided and the customer relationship gets less care than it had under the founder.
Kill signal threshold: >25% single customer, >40% top-3 combinedThe org chart shows a CEO with four direct reports. The CEO answers every diligence question. Every customer reference mentions the CEO by name. The CFO has been in the role for 8 months. This is not a company — it is a principal who has organized resources around themselves. PE value creation requires management bandwidth to execute the 100-day plan while running the existing business. One-person-deep management teams fail this test almost universally at LMM transaction sizes.
Kill signal threshold: CEO in >3 revenue-critical relationships, no COO or VP-level benchThe CIM shows $2.4M in normalized EBITDA adjusted for $1.1M in add-backs. The largest add-back is $600K in 'owner compensation above market rate.' The second largest is $280K in 'non-recurring legal fees' from a dispute that settled 14 months ago. Ask one question: what was the legal fee for? If the answer is not clean — not a genuinely non-recurring operational expense with documentation — the EBITDA bridge has a problem that will be larger post-close than it appears in the CIM.
Kill signal threshold: add-backs >35% of stated EBITDA, any add-back without documentary supportThe business generates $2M in EBITDA and converts 85% to free cash flow. The maintenance capex figure in the model is $180K annually — 9% of EBITDA. But the company operates out of a 15-year-old facility with a fleet of equipment that hasn't been updated since 2018 and an ERP system running on a server in a closet. Deferred capital expenditure is the most reliable predictor of Year 1 cash flow disappointment in LMM transactions. The free cash flow that looks like investment capacity is actually backlog catching up.
Kill signal threshold: capex <12% of EBITDA for asset-intensive businesses, no recent major systems refreshThe trailing three-year CAGR is 22%. But 14% of that is in the most recent 12 months. Revenue growth that accelerated sharply immediately before a sale process is worth a diligence conversation, not a valuation premium. The explanation may be entirely benign: a new sales hire, a contract win, a market tailwind. It may also be: the seller knew the business was for sale and prioritized revenue at the expense of margin, customer quality, and long-term contract terms. Ask what changed. The answer tells you which one it is.
Kill signal threshold: growth rate in final year >2x the 3-year average, no clear structural explanationThese five signals should not be treated as a checklist. A checklist implies binary pass/fail. The right framework is a weighted risk register where each signal is scored by severity, evidentiality, and remediability.
Severity is the magnitude of the impact if the risk materializes — a 40% customer concentration in a B2B services business is more severe than the same concentration in a business with contractual revenue. Evidentiality is how clearly the signal appears in available data — a 22% EBITDA add-back with documentation is less worrying than a 14% add-back with no supporting memo. Remediability is whether the risk can be structurally addressed before or after close — management depth can sometimes be addressed by hiring before the wire transfers; customer concentration cannot.
“The IC memo answers ‘why should we do this deal?’ The adversarial diligence lens answers ‘what has to be true for this deal to fail?’ Every deal process needs both. Most only run one.”
“An AI-generated risk register does not have a career incentive to minimize signals that kill deals. It runs the adversarial lens without the employment relationship that makes human analysts systematically soften inconvenient conclusions. That is not a replacement for experienced judgment — it is the pre-processing layer that ensures experienced judgment is applied to accurate inputs.”
The experienced partner's judgment about whether a detected signal is a dealkiller or a negotiating lever is irreplaceable. The detection itself — the systematic application of adversarial analysis to a data room — should not require a process that is structurally biased toward confirmation.
Kill the right deals early. The $400K in diligence fees is not the expensive part. The expensive part is the 3-year hold on a business that misses its EBITDA plan from day one and returns capital at 0.8x. The five signals above are the filter that separates those deals from the pipeline before that cost is incurred.
Run the adversarial diligence lens on your next deal in 20 minutes.
Upload your CIM or enter deal parameters. The risk engine screens for all five structural signals, scores severity and evidentiality, and generates a weighted risk register alongside the IC memo. No career risk included in the analysis.
Analyze a Deal →“The best deal process is not the one that closes every deal faster. It is the one that kills the bad ones before they cost $400,000 and three years of a portfolio company's life. The five signals above are the pre-diligence screen that separates the ones worth pursuing from the ones worth passing on day one.”
Kill bad deals early. Run good ones harder.