Vendor due diligence: what it is, what a good VDD report contains, and how AI changes the first draft

A practitioner’s guide to sell-side due diligence: the report structure buyers expect, the EBITDA bridge, debt-like items and the working capital peg, the five-step red-flag framework — and where AI genuinely shortens the work.

What vendor due diligence is

Vendor due diligence (VDD) is a due diligence exercise commissioned by the seller of a business, performed by an independent advisor, and written so that it can be shared with prospective buyers. It covers the same ground a buyer’s own financial due diligence would cover — quality of earnings, net debt, working capital, the commercial and legal position — but it is done once, before the process starts, rather than by every bidder in parallel.

The purpose is not to produce a marketing document. A VDD report that reads like an information memorandum is worthless to a buyer and damaging to the seller, because the first inconsistency a bidder finds will reprice the whole deal. A good VDD report is deliberately neutral: it describes the facts, states the adjustments, quantifies the exposures and lets the buyer draw conclusions. The seller’s advantage comes from knowing everything first — fixing what can be fixed, explaining what cannot, and pricing in what remains — not from hiding anything.

Vendor DD versus buy-side DD

The analysis is the same; the audience and the posture are different. In a buy-side report the advisor works for the buyer, recommends a position (proceed, proceed with conditions, do not proceed) and proposes specific SPA protections for each finding. A vendor report is written for the seller but intended for buyers, so it:

• describes facts and adjustments neutrally, without advocating a price;
• replaces the buyer-side “recommended SPA mechanics” with a section on matters the seller should prepare for — the items a buyer will inevitably raise, and what the seller’s answer is;
• is released to bidders under a reliance letter or, more often in the first phase, a non-reliance letter.

Reliance matters more than most sellers appreciate. A non-reliance report lets bidders read the work but gives them no claim against the advisor; it is the normal basis for releasing the report at the indicative-offer stage. A reliance letter, typically issued to the preferred bidder (and sometimes its lenders) at the final stage, gives the buyer a contractual right to rely on the report, usually subject to a liability cap. Buyers who receive reliance can reduce their own confirmatory diligence to a “top-up” exercise, which is exactly what shortens the timetable. Agree with your advisor at the outset who will receive reliance, on what terms, and at what cap — it affects the scope and price of the engagement.

The ten-section structure of a VDD report

There is no statutory format, but good reports converge on the same skeleton. The structure we use follows the pattern that experienced buyers and their advisors expect to see:

Two habits separate a usable report from a decorative one: every statement is cited to a source document and page, and the report never guesses beyond the data — where a document is missing it says so, names it, and states the consequence.

Quality of earnings and adjusted EBITDA

The EBITDA bridge is the single most-read table in any due diligence report, because the headline price is almost always a multiple of adjusted EBITDA. The bridge starts from reported EBITDA and lists every adjustment with its amount, category and source. Quality-of-earnings adjustments are commonly grouped into four families — non-recurring items, non-cash items, items recorded in the wrong period, and pro forma adjustments — and that taxonomy is a good test of whether a bridge is complete. In our own work we expand it into seven groups:

1. Owner-related. Owner remuneration above or below market, relatives on the payroll without a role, private expenses run through the business, property owned by the shareholder and leased to the company below market, personal guarantees on company debt.
2. Non-recurring. One-off income (litigation settlements, insurance proceeds, grants, exceptional FX gains) deducted; one-off costs (restructuring, redundancy, transaction advisory fees, exceptional repairs) added back. Pandemic-era support deserves its own line, including claw-back risk.
3. Accounting policy. Changes in accounting policy restated for comparability; revenue recognition (over time versus point in time, principal versus agent, cut-off at period end); cost accruals and prepayments; depreciation and impairment; inventory valuation and obsolescence; receivables provisioning against the ageing.
4. Related parties. Sales to or purchases from related companies at non-market prices restated to arm’s length; intra-group loans at non-market interest; shared staff and services allocated properly.
5. Working capital and cash. Not EBITDA adjustments as such, but the analysis that produces the working capital peg and the maintenance-versus-growth capex split a buyer will use to convert EBITDA into cash.
6. Debt normalisation. Everything that feeds the net debt bridge below.
7. Pro forma and run-rate. Acquired or discontinued operations annualised or removed; stand-alone costs the business will have to carry after separation from a parent or owner (accounting, IT, insurance, management) deducted; cost savings included only where the action has already been taken and the saving is visible in the monthly numbers — everything else is shown separately, outside adjusted EBITDA.

Two discipline points. A seller-side bridge should include the adjustments that reduce EBITDA as diligently as those that increase it; a buyer’s advisor will find them anyway, and a one-directional bridge destroys the credibility of the whole report. And the bridge should cover three financial years plus the current year to date, with the multi-year average stated.

Net debt and debt-like items

If the price is agreed on a cash-free, debt-free basis, the definition of “debt” decides how much of the headline price the seller actually receives. Interest-bearing debt is the easy part: bank loans, bonds, overdrafts, finance leases, accrued interest. The negotiation happens over debt-like items — obligations that are not loans but which a buyer will argue it is effectively paying off on the seller’s behalf. A thorough VDD report walks through every category and records “not applicable” explicitly where that is the answer, so the buyer cannot claim an item was overlooked:

• Overdue trade payables beyond normal terms, and stretched creditors generally.
• Unpaid corporate tax for prior periods; overdue VAT, payroll taxes and social charges.
• Unfunded pension obligations and overdue wage claims. (Accrued but untaken holiday is normally treated as working capital, not debt — but state the treatment.)
• Customer prepayments and deferred revenue where the service has still to be delivered.
• Declared but unpaid dividends and other amounts owed to shareholders.
• Transaction bonuses, retention payments and severance triggered by the sale.
• Deferred consideration or earnouts still owed on the company’s own previous acquisitions.
• Grants that may become repayable, including research and pandemic-era support.
• Deferred maintenance capex and unpaid capital invoices.
• Provisions for litigation, claims, guarantees and environmental remediation.
• Leases that are financing in substance; deferred tax that crystallises on the transaction.

Off-balance-sheet items get their own list: factored receivables, guarantees given for related parties, bank guarantees, and committed purchase contracts. On the other side, cash-like items are cash above the minimum operating level the business needs to run, while restricted cash (deposits, collateral) is not cash-like. The chapter closes with a single line: net debt equals total debt and debt-like items less cash and cash-like items.

The working capital peg

The second price mechanism a buyer will insist on is a target net working capital (“peg”). The buyer expects to receive a business with a normal level of working capital; if closing working capital is below the peg, the price comes down by the difference, and vice versa. The definition is current assets less current liabilities, excluding cash, interest-bearing debt and corporate tax — all of which are dealt with in net debt.

Our house standard, in line with common practice, is a twelve-month average of monthly net working capital, adjusted for seasonality. The report should always show the sensitivity: a single month, the twelve-month average, and the same month a year earlier. The gap between them can be large — bonus and salary payment months, quarter-end billing cycles and inventory build-ups all move the number — and the party who understands the seasonality negotiates the peg. For the seller, that means monthly balance sheets for at least 24 months are not optional; they are the only way to prove that the peg should be set where you want it.

The red-flag framework

Findings are only useful if a reader can act on them. We require every red flag in a report to be written in the same five-step structure, the structure that makes a finding actionable:

1. Observation. What the documents show. One sentence, with a citation.
2. Implication. What it means for the transaction. One sentence.
3. Quantification. The effect on enterprise value, as a range if uncertain — in currency or as a multiple of EBITDA.
4. Recommended action. A specific remedy: a price adjustment, a specific indemnity, an escrow, a condition, a document to be produced by a date.
5. Priority. Deal-threatening, material, or disclosure-worthy.

The structure forbids the vague language that creeps into first drafts. “This could be a problem” becomes “this is an obligation of approximately X; a specific indemnity is recommended”. “Needs further review” becomes “cannot be assessed without the following document; if not provided within five working days, a retention of Y is recommended”. In a vendor report the recommended action is phrased as what the seller should prepare, but the discipline is the same.

In the SPA negotiation that follows, these findings map onto a small set of tools. Across European private-company deals, a purchase price adjustment or a locked box is the norm rather than the exception; escrows and retentions, earnouts, de minimis thresholds and baskets are all ordinary market practice, and warranty and indemnity insurance is increasingly common. The point for a seller is that most of the remedies a VDD report anticipates are standard tools, not exotic demands — and the seller who has quantified the issue first is the one who sets the terms.

Data gaps: what is not in the room

The question a good advisor asks before writing a single chapter is: what is not here that an experienced buyer’s diligence team would demand? Three years of audited accounts, management accounts reconciled to them, monthly balance sheets, debtor ageing, the top-ten customer contracts with their change-of-control clauses, all leases, tax filings for the reassessment window, related-party schedules, employment contracts for key people, insurance policies. Each missing item becomes a numbered data request with a deadline and a stated consequence.

There is a corollary we apply deliberately. A data room that contains no disputes, no customer losses, no regulatory correspondence and no bad debts over five years is not a clean company; it is an incomplete data room. No business of any age runs without friction, and the absence of problems in the documents is itself a finding — one that a buyer will address with a broad “no undisclosed litigation” warranty and a catch-all indemnity. A seller is better served by disclosing the small matters up front, in the report, with the resolution attached. Our 39-item sell-side due diligence checklist is the request list we use to make sure nothing is missing before the room opens.

Timeline and cost

A conventional VDD on an owner-managed company with a reasonably complete data room takes four to eight weeks from kick-off to a releasable report, and the fee is typically the largest advisory cost the seller incurs before the SPA lawyers. The time goes in a predictable order: one to two weeks collecting documents, two to three weeks of analysis and bridges, one to two weeks of drafting, then a fact-check with management. Elapsed time is driven far more by how complete the data room is on day one than by the size of the business.

Three things reliably compress it: having the request list delivered before the advisor starts; monthly management accounts that reconcile to the audited numbers without manual intervention; and a data room structured by workstream so that nobody is hunting for the lease in a folder called “Misc”.

How AI changes the first draft

The drafting stage of a VDD report is largely a reading and reconciliation exercise: find every contract with a change-of-control clause, trace every “exceptional” line in the accounts back to its invoice, check that the customer list in the CRM export matches the revenue by customer in the ledger, compare the lease term with the depreciation period of the fit-out. It is skilled work, but most of it is pattern matching across a few hundred documents — which is precisely what current language models do well when they are constrained to the documents in front of them and forced to cite.

This is how SellSide DD approaches it. The data room is uploaded as it is, with its folder structure preserved. Every document is indexed and summarised. The system then does the first pass: it extracts the financial tables, builds a draft EBITDA bridge from the adjustments it can evidence, lists the debt-like items it finds and the categories it could not evidence, consolidates contract clauses into risk themes (change of control, key person, leases, IP, exclusivity), flags inconsistencies between documents — the headcount in the HR list against the payroll summary, the revenue in the management accounts against the audited figures — and produces a draft report in the ten-section structure above, with every statement linked to its source document.

What it does not do is decide. Each finding is presented to the advisor, who can accept it, challenge it, or send it to the seller as a question. The published report reflects the advisor’s judgement, not the model’s; the model’s contribution is that the advisor starts from a cited draft and a gap list on day two instead of day twenty. Sensitive items — customer names, pricing, litigation — are flagged before anything reaches a bidder, and because the same index answers buyer questions in the Q&A phase, the seller’s story stays consistent from the first management presentation to the SPA disclosure letter. For what that requires of the room itself, see our guide to choosing a data room for sell-side M&A.