Financial due diligence is thorough, legal review is meticulous, and the quality assessment — if it happens at all — consists of skimming a quality manual and confirming a certificate hangs in the lobby. Across two decades in automotive and aerospace quality, I have walked through enough acquisition targets to know the pattern holds everywhere. The liabilities that erode margins after closing rarely appear on the balance sheet. They sit in concession registers, warranty claim databases and the habits of engineers who quietly work around processes that never quite worked.
Quality debt behaves like technical debt in software. Every deviation someone signed off rather than fixed, every supplier whose capability was assumed rather than verified, every inspection added because the process cannot hold tolerance — these are deferred payments with interest. The seller knows about most of them. The data room will volunteer none of them.
The buyer's quality assessor has one job: find the debt before inheriting the repayments. What follows is where I have learned to dig, organised around the five places hidden liability concentrates — open concessions, warranty exposure, system maturity, the supplier layer, and traceability with its regulatory tail.
Open Concessions: The Register That Tells the Truth
Start with the concession or deviation register, and demand it in raw form — not the sanitised summary the corporate development team prepares. What matters is not the count of open concessions but their character. A hundred open deviations on cosmetic surface finish tells you the organisation tolerates noise. Twenty open deviations on dimensional characteristics with tolerance stack involvement tells you the product design is living on borrowed time.
Read the justification text closely. Phrases like "supplier to requalify next batch" repeated across months signal a supplier who failed a capability requirement and was never brought back into control. Then look at who approves. Concessions signed by the same quality engineer, for the same part number, on a monthly cadence, are permanent waivers wearing temporary clothing. Check expiry dates against closure evidence: a substantial share of "temporary" deviations in typical targets have lapsed and simply continue in production because nobody enforces the expiry.
Every lapsed deviation is a shipment of nonconforming product to a customer who never formally agreed to receive it — a contractual exposure, and under AS9100 flow-down requirements in aerospace, potentially a regulatory one you now own. Trace each concession into the customer base: a deviation tacitly accepted by one customer's receiving inspector is not the same as one covered by formal customer deviation authorisation. If the target cannot produce written customer approval for open deviations affecting safety, regulatory or interface characteristics, price that risk explicitly in the deal model. The day the receiving inspector changes or the customer audits the file, that exposure crystallises.

Warranty Exposure: Reading Claims Like a Failure Analysis
The warranty database is the most honest document in the data room, because it records what actually broke rather than what the process was supposed to produce. Insist on raw claim data by part number, failure mode, claim date and production date. The last field is the one sellers routinely omit, and the one that matters most: claims plotted against production month reveal whether failure rates are improving or deteriorating. A rising trend in the most recent production quarters, before the product has aged into its failure window, is the single most alarming signal in quality due diligence.
Segment the analysis by failure mode and distinguish design weakness from manufacturing scatter from supplier-induced defects. A target whose claims cluster around a single supplier's component has a negotiable problem: you can re-source. Claims spread thinly across many failure modes point to a manufacturing system that cannot hold its process — a far deeper hole. Capability indices in the data room mean nothing if the underlying distribution is unstable, and this pattern is how you detect instability without ever seeing the shop floor.
Watch for the claims that never got filed. Customers frequently absorb small failures rather than claim, especially in tight supply conditions where they fear allocation retaliation. Those suppressed claims return with interest the moment supply loosens. Always request customer scorecards and complaint correspondence alongside the formal warranty ledger; the gap between the two is your hidden exposure. Check claim latency too — many targets under-accrue because they book claims when paid rather than when reported, which flatters the reserve and understates the liability you are absorbing.
System Maturity: Auditing Capability, Not Certificates
Certificate walls mean little. I have audited certified plants that could not produce a coherent tolerance stack analysis, and uncertified job shops with immaculate first-article discipline. In due diligence, assess system maturity through artefacts, not attestations. Pull the internal audit schedule and findings from the last three cycles: are audits finding substantive process issues, or only documentation nits? An audit programme reporting nothing but clerical findings for years is not evidence of excellence — it is evidence of an audit function that has stopped looking.
Layered process audits, management review minutes, corrective action ageing and the ratio of preventive to reactive actions all tell you whether the quality organisation leads or follows. Check the corrective action system for repeat part numbers and repeat root causes. When the same failure mode recurs under different 8D report numbers, root cause analysis is cosmetic — the "root cause" being written is the one that closes the ticket fastest. That habit, multiplied across years, is the engine that generates the warranty claims you examined in the previous section.
Maturity also shows in measurement discipline, the foundation every Cpk figure rests on. Check how gauges are treated: are calibration labels current, is gauge discrimination adequate for the tolerances being checked, do operators know the reaction plan when a check fails? A shop floor where the answer to an out-of-tolerance reading is "run it again" has institutionalised doubt in its own measurements. You cannot assess capability honestly in such an environment, and every process capability figure in the data room becomes unverifiable.
What the certificate says versus what the artefacts show
What the data room shows
- ISO 9001 or AS9100 certificate in force
- Quality manual with policy statements
- Internal audit programme on schedule
- Management review held quarterly
What actually indicates maturity
- Audit findings that name process failures, not typos
- Corrective actions that do not repeat root causes
- Calibration current and gauge discrimination adequate
- Reaction plans operators can state without notes
The Supplier Layer: Inherited Debt You Cannot See
Any acquisition buys not just the target's quality system but its entire supply base, and supplier quality is where the worst surprises live. Request the approved supplier list with the last audit date and score for each critical supplier, then cross-reference it against the defect Pareto from incoming inspection. Suppliers who ship recurring nonconformances yet retain approved status, without a documented containment and improvement cycle, represent debt that transfers silently at closing.
Assess the target's supplier surveillance depth. Does it audit sub-tier processes for critical characteristics, or does it rely on certificates and incoming sampling? For castings, forgings and special processes — heat treatment, welding, plating, non-destructive testing — the answer determines whether you are inheriting validated processes or a chain of assumptions. Check special process approvals referencing Nadcap or equivalent accreditation for scope: a supplier accredited for one process does not cover the shop that actually ran your parts.
Flow-down requirements deserve their own review. In aerospace particularly, customer and regulatory flow-downs must reach the sub-tier contractually. Pull three critical supplier contracts and check whether key characteristics, notification-of-change clauses and right-of-access audit provisions actually exist in the terms. Where flow-down is missing, the target has been operating on goodwill, and the first supplier change or ownership transfer at a sub-tier can invalidate years of quiet stability without a single notification reaching you.
Suppliers who ship recurring nonconformances yet retain approved status are debt that transfers silently at closing.
Field Returns, Traceability and the Regulatory Tail
Beyond warranty economics sits recall exposure, and that question turns on traceability. Ask how the target isolates affected product when a defect is discovered: can it retrieve heat numbers, production records and shipment destinations for a given date range within hours, or does reconstruction take weeks of manual archaeology? Traceability capability is the difference between a targeted field action and a blanket recall, and the cost delta between those two outcomes can decide whether an acquisition pays back.
Regulatory posture compounds this. Review the history of customer notifications, agency reports and field actions over the past five years, including those closed quietly. Ask what reporting obligations attach to the product lines — automotive safety-critical reporting regimes and aviation authority requirements create tails that extend years beyond production end. Part 21 supply into aerospace carries continuing airworthiness obligations that survive the transaction; ensure the purchase agreement addresses who holds and services those approvals, because losing production organisation approval standing in a transfer is a self-inflicted wound.
Then there is the records dimension. Design records, process validations and acceptance test data held in obsolete formats or on departed engineers' laptops are debt of a subtler kind. When a customer demands archival evidence a decade old — and in aerospace, they will — the target's ability to produce it determines your liability. Test the archive during diligence with a random sample retrieval request. If it takes more than a day, you have found another liability nobody disclosed.
From diligence finding to deal term
- 01FindRaw registers, claim data, audit artefacts, supplier files, archive tests
- 02QuantifyCost of lapses, under-accrual, re-sourcing, remediation labour
- 03CategorisePrice adjustment, indemnity, or costed integration action
- 04ContractWrite it into the purchase agreement with owners and dates
Putting the Findings Into the Deal
Diligence findings only matter if they change price, structure or the first hundred days. Convert each major finding into one of three categories: price adjustment, indemnity, or integration action with a costed plan. Open concessions on customer-sensitive characteristics argue for indemnities. Warranty trends argue for a working-capital adjustment against the accrual. System immaturity argues for a priced remediation programme in your synergy model — not as a hope, but as a line item with owners and dates.
Insist on access to people, not just documents, in the final diligence phase. One structured interview with the target's senior quality engineer, away from the deal team, will surface more genuine risk than a month of data room review. Ask a single question: "What would you fix first if you had the authority?" The answer, in my experience, is the most accurate due diligence report you will ever receive — and it costs nothing.
Quality debt does not appear on the balance sheet, but it behaves like any other liability: it compounds, and eventually someone pays. The buyer who reads concession registers like a ledger, warranty data like a failure analysis and supplier files like inherited contracts will find most of it before signing. The buyer who trusts the certificate wall will spend the first two years after closing paying someone else's interest.
