In a purchase agreement, the seller gives representations and warranties: the accounts are true, no litigation is pending, the material contracts are listed, the company is compliant for tax and employment purposes. If one of them turns out to be false after signing, the buyer has a claim against the seller. That claim is worth exactly what the seller is worth at the moment it is brought, and what remains reachable of it. This is where escrow, liability caps and survival periods come from: three ways of organizing a claim that nothing guarantees will be paid. A warranty and indemnity policy does not replace the warranty. It replaces the counterparty.
That sentence is the key to the product, and it explains why seller and buyer do not buy it for the same reason. The seller buys a clean exit. An investment fund selling a holding has a finite life, investors to repay and a distribution timetable: locking a slice of the price into escrow for two years delays the fund's closure and makes its return unreadable. The policy lets it walk away having collected everything. This is why the line grew with private equity rather than with industrial disposals, where the seller often remains solvent, identifiable and present ten years later.
The buyer is not buying the same thing, and this is the point sales presentations skip. It is not primarily buying money, since escrow already gave it that. It is buying the ability to claim without attacking anyone. In a sale where the management team sells and stays in charge, enforcing the warranties against the seller means suing the people who run the asset just bought, and many legally sound claims are never brought for that reason alone. The policy moves the claim to a third party with no role in the business, and that is what makes the recourse usable. To which an auction adds a less avowable and entirely real use: offering a deal with no escrow because it is insured is a concession that costs nothing on price and shows up immediately when bids are compared.
A structural consequence follows from those two motives. A policy can be taken out by the seller or by the buyer, and in practice almost all are taken out by the buyer, including where the seller started the process, obtained the first indications, then handed over. Two reasons compound. The first is mechanical: a buyer-side policy indemnifies the buyer directly, without it first having to establish the seller's liability, whereas a seller-side policy responds only once the seller has been found liable, which presupposes the very litigation one was trying to avoid. The second concerns fraud: nobody insures against their own fraudulent conduct, so a seller-side policy cannot by construction cover the seller's fraud, whereas a buyer-side policy can.
The policy does not start at zero. It carries a retention, expressed as a fraction of enterprise value, below which nothing is due, and a de minimis threshold below which a claim is not even counted. And the purchase agreement almost always keeps a symbolic seller liability, often one euro. That last item surprises people, and it has a precise function: a warranty behind which nobody stands any longer can be read as boilerplate, and the policy insures contractual warranties, not decorative statements. The symbol is therefore not a symbol, it is what keeps the insured object in existence.
What the policy does not replace is the work. Underwriting reads the buyer's due diligence, its legal, tax, financial and sometimes environmental reports, and questions the teams that carried them out. The principle is simple and it always surprises: what the review did not look at is not covered. A diligence exercise that skipped employment contracts does not produce a more expensive policy, it produces an employment law exclusion. Insurance therefore does not substitute for examination, it prices it, and a buyer in a hurry who thins out its review to hold a timetable discovers at claim time exactly what it had saved.
That leaves the nature of this underwriting, which is not that of the rest of the market. There is no law of large numbers: every transaction is unique, there is no portfolio of comparable sales across which to spread a hazard, and the underwriter is not estimating a frequency but the strength of a legal position. The craft is closer to writing an opinion than to rating a risk. It also explains the place of selection discipline: profitability in this line is decided by the files that are declined, and an insurer that accepts everything presented to it on a market without statistics has no way of noticing for several years.
A fund sells a services company to an industrial buyer. The management team sells its minority stake along the way and stays in place. The agreement provides for an escrow of 5 percent of the price for eighteen months. Four months after completion, the buyer discovers that a major client contract, presented as renewed, had in fact been terminated three weeks before signing. The fund has already distributed the proceeds to its investors. What does the presence or absence of a policy change in practice?
Without a policy the buyer has two routes and neither is comfortable. The escrow covers 5 percent of the price, which may or may not be enough depending on the weight of the lost contract, and it expires in fourteen months. Beyond that, the sellers must be pursued: the fund, which has distributed and whose structure is winding up, and the management team, which now runs the asset that was bought. That second route is legally sound and practically unusable, since bringing it means suing the people the value just paid for depends on. This is exactly the situation the policy resolves, by moving the claim to a third party with no role in the business. That leaves the question which actually decides the file, and it is not an insurance question: the termination predates signing. If it appeared in the data room or in the due diligence, it was known, therefore excluded, and no policy will change that. If it appeared nowhere, the warranty on material contracts is inaccurate and the cover responds. The whole matter therefore turns on what the buyer could have known at the date of signing, which is the subject of the next module.
- 01The policy does not replace the seller's warranty, it replaces the counterparty: what changes is who pays, not what was promised.
- 02The seller buys a clean exit, the buyer buys a usable claim against a third party. Two distinct motives for one contract.
- 03Almost all policies are taken out by the buyer, because they indemnify it directly and because a seller-side policy cannot cover the seller's fraud.
- 04What the due diligence did not look at becomes an exclusion: the policy prices the work of examination, it does not replace it.