Deal Structure as Risk Allocation

Key takeaways

• Every deal term other than price is an answer to one question: if this particular assumption is wrong, whose money fixes it?
• Earnouts pay about 21 cents on each maximum dollar across deals that use them, excluding life sciences (SRS Acquiom deal-terms research), so price the bridge at expected value, not headline.
• The peg is a second price: 93% of private-target deals carry a purchase-price-adjustment mechanism and 89% of those actually produce an adjustment (SRS Acquiom).
• In Chicago Bridge & Iron v. Westinghouse the headline price was zero dollars and the two sides' true-up calculations sat $2.578 billion apart. The adjustment WAS the deal.

Price is one number. The deal is a dozen.

A purchase agreement looks like a document about price. It is not. Price is the single number both sides agree on when they agree about everything else; the rest of the agreement exists because they do not. Each remaining clause takes one specific uncertainty (will the receivables collect, will the customers stay, were the financials prepared the way the seller said, is there a lawsuit nobody mentioned) and assigns it to a side. Read that way, an acquisition agreement is a risk-allocation schedule with a price stapled to the front.

This reframing changes what you argue about. Two buyers can offer the same headline number and be offering wildly different deals, because one is holding the seller responsible for eighteen months of unknown liabilities and the other is not. A seller who negotiates hard on price and signs the buyer's first draft of everything else has usually given back more than they won. The reverse is also true and less discussed: a buyer who wins a punishing indemnity package from a seller who cannot pay it has won nothing, because a covenant is only worth the balance sheet standing behind it.

The four instruments in this briefing do almost all of the work. The working-capital peg allocates the risk that the business arrives with less fuel in the tank than it normally runs on. The earnout allocates disagreement about the future. Holdbacks and escrows allocate the risk that a promise turns out to be false and the promiser has already spent the money. Reps and warranties define what was promised in the first place, and the survival clause decides how long the promise lives.

A useful discipline before signing anything: take every assumption your price depends on, write it in one line, and write next to it the name of the person who eats it if the assumption is wrong. Assumptions with no name next to them are the ones that will hurt, because they default to whoever holds the asset, which after closing is you.

The working-capital peg: the second price, negotiated last

Most operating businesses need a certain amount of cash tied up in receivables and inventory, minus payables, just to run. That number is net working capital. A buyer paying for a business paying for the earning power of that business is implicitly paying for a normal amount of working capital to come with it. So the agreement fixes a target, the peg, and trues up the price dollar for dollar against the actual balance at closing.

This is not a technicality. SRS Acquiom, which acts as shareholder representative on private-target deals and publishes aggregate deal terms, reports that 93% of private-target deals carry a purchase-price-adjustment mechanism and that 89% of deals with one actually produce an adjustment. The average adjustment across their sample ran to the buyer's benefit at roughly 0.9% of transaction value, and the median special-purpose escrow set aside just for that adjustment sits at about 1% of transaction value. Read those two numbers together and the market's own design becomes visible: the escrow is sized at roughly the average claim, which by construction means the above-average half of adjustments reaches past it.

Illustrative only: a distribution business is bought for $6,000,000 on a cash-free, debt-free basis. Its trailing-twelve-month average net working capital is $900,000, and that becomes the peg. At closing the actual figure is $740,000, so the price steps down by the $160,000 shortfall to $5,840,000. Now change one drafting choice. The business is seasonal, and its March 31 balance was $1,150,000. Had the parties pegged to that single month instead of the twelve-month average, a September closing at the same $740,000 would have produced a $410,000 reduction and a $5,590,000 price, a $250,000 swing created by nothing except which month became the benchmark. The $160,000 adjustment in the first version is 2.7% of the deal, well above the roughly 0.9% average SRS Acquiom observes; seasonal businesses generate the outliers.

The levers, in order of how much money they move: which period defines the peg (a trailing average beats any single date), which accounts are in and out of the definition (deferred revenue, accrued bonuses, and customer deposits are the usual fights), whether the closing statement is prepared on the target's historical accounting practice or on strict GAAP, and who breaks a tie. That last one is not a detail. In Chicago Bridge & Iron v. Westinghouse, Chicago Bridge sold its Stone and Webster subsidiary for zero dollars of upfront consideration against a closing net working capital target of $1.174 billion. Westinghouse's true-up calculation said Chicago Bridge owed it $2.15 billion; Chicago Bridge's calculation said the buyer owed Chicago Bridge $428 million. The parties' two versions of the same clause were $2.578 billion apart on a deal with a zero headline price.

What broke the tie was a different clause entirely. The agreement contained what the Delaware Supreme Court called a Liability Bar: none of the seller's representations survived closing, and the buyer's remedy for a breach was to refuse to close. On June 27, 2017 the Court reversed the Court of Chancery and held that the independent auditor appointed to resolve adjustment disputes had a brief confined to a discrete set of narrow disputes, and could not entertain arguments that the seller's historical financials failed to comply with GAAP, because that was the buyer routing a dead representation claim through the true-up. The risk was allocated in the survival clause; the true-up could not reallocate it.

Earnouts: buying a bridge that pays about 21 cents

An earnout exists because the two sides disagree about the future and neither will move. Rather than split the difference, they defer it: part of the price becomes contingent on the business hitting a defined result. In principle this is elegant risk allocation: the seller keeps the upside of their own forecast and the buyer stops paying for a forecast they do not believe.

In practice, the base rate is brutal and worth memorising. Across deals with earnouts, excluding life sciences, SRS Acquiom's deal-terms research finds that about 21 cents of each maximum earnout dollar is ultimately paid. Among deals that achieve any earnout at all, roughly half the maximum dollars are paid, which tells you the distribution is not a smooth curve but two clusters: nothing, or about half. Their 2024 sample also shows 68% of earnout deals using multiple metrics and no performance period longer than four years.

Illustrative only: a seller wants $8,000,000 and a buyer will pay $6,000,000. They bridge the gap with a $2,000,000 earnout over two years, $1,000,000 per year against gross profit targets of $2,400,000 and $2,650,000. Apply the observed base rate and the expected value of the bridge is $2,000,000 × 0.21 = $420,000. The eight-million-dollar deal is, on base rates, a $6,420,000 deal. And because achievers cluster near half the maximum, the realistic outcomes are approximately $0 or approximately $1,000,000, almost never the $2,000,000 the seller is mentally spending.

The structural trap is the AND-gate. Bristol Myers Squibb's 2019 acquisition of Celgene gave each Celgene share one BMS share, $50.00 in cash, and one tradeable contingent value right worth a one-time $9.00 in cash, payable only upon FDA approval of all three of ozanimod by December 31, 2020, liso-cel by December 31, 2020, and ide-cel by March 31, 2021. With 714.9 million Celgene shares outstanding, the face value was about $6.43 billion. The CVR first traded on November 21, 2019 at $2.30, and BMS recorded the shareholder CVRs at $1,644 million of fair value within $80,269 million of total consideration. The market's own price on the bundle was therefore about 26 cents on the dollar. It paid zero: the FDA did not approve liso-cel by December 31, 2020, and the CVR agreement terminated automatically in January 2021. Illustrative arithmetic on why AND-gates price so low: three independent milestones each 80% likely produce a joint probability of 0.8 × 0.8 × 0.8 = 51.2%, and each additional gate multiplies the discount again.

The second trap is control. A seller's earnout depends on a business the buyer now runs, and the buyer's ordinary commercial choices (reallocating salespeople, delaying a launch, prioritising a different product) can extinguish it without anyone acting in bad faith. Johnson & Johnson acquired Auris Health on April 1, 2019 for approximately $3.4 billion net of cash acquired, with additional contingent payments of up to $2.35 billion on predetermined regulatory milestones: 41% of the $5.75 billion maximum sitting in the contingent bucket. The dispute that followed ran for years. The Court of Chancery in 2024 awarded the former Auris holders over $1 billion, $300 million tied to the first earnout payment, $600 million to the remaining ones, and $61 million for fraud, plus pre-judgment interest, finding that J&J had breached its efforts obligations and had overstated the likelihood of one milestone. On January 12, 2026 the Delaware Supreme Court reversed the $300 million piece, holding that the implied covenant could not require pursuit of a De Novo approval pathway when the contract specified only 510(k) approval, while affirming the breach and fraud findings on the remainder and remanding for recalculation.

The levers follow directly. Prefer a metric high on the income statement, revenue or gross profit, because net income is downstream of every allocation decision the buyer now controls. Keep the period at two years or less; the further out the metric, the less of the result the seller caused. Write the efforts standard explicitly rather than relying on a court to supply one, and enumerate the specific actions the buyer must take. Add acceleration on a change of control, on the buyer discontinuing the product line, and on the seller's employment ending without cause. And run the whole structure through expected value before you agree to it, because 21 cents is the number the market actually pays.

Reps, warranties, and the survival clause that gives them teeth

A representation is a statement of fact about the business, made as of a moment, on which the other side is entitled to rely. The financial statements were prepared consistently with past practice. There is no undisclosed litigation. The company owns the intellectual property it uses. Taxes have been filed and paid. Each one looks like boilerplate and each one is a risk transfer: by making the statement, the seller agrees to be the person who pays if it is false.

The hard edge is not the representation but the survival clause. A representation that dies at closing is a statement the buyer may rely on only up to the moment of signing the cheque, after which its remedy is nothing. That is precisely the structure that decided Chicago Bridge: the reps did not survive, the buyer's only remedy was to refuse to close, and the Delaware Supreme Court would not let the true-up mechanism resurrect what the survival clause had killed. Meanwhile the market has been shortening survival generally. SRS Acquiom reports the median survival period holding at 12 months in recent years.

So the negotiation to actually have is not whether the seller will make a representation but three follow-on questions. How long does it live? What does it cost to make a claim, given the deductible or basket below which nothing is recoverable and the cap above which nothing more is? And which representations sit outside those limits as fundamental: title to the shares, authority to sell, and usually taxes, which typically survive longer and are capped higher or not at all.

The practical asymmetry is worth stating plainly. A seller's exposure under the reps is the single largest reason sellers want a clean, fast, low-survival deal and buyers want the opposite. Representation and warranty insurance exists to break that deadlock by moving the exposure to a third party, and SRS Acquiom's data shows the deal shape changes when it is present: walk-away structures, where the buyer's only recourse is the policy, appeared in 33% of deals with RWI in their 2024 sample against 18% without.

What breaks first, in practice, is disclosure. Most rep disputes are not lies; they are a seller who knew something, assumed it was obvious, and did not put it on the disclosure schedule. The schedules are where the actual risk allocation happens, they are always drafted last, they are always drafted by the most junior person available, and they are the document to read twice.

Holdbacks and escrows: the money that makes a promise real

A promise from a seller who has already been paid and distributed the proceeds to twelve shareholders is a promise you will collect by lawsuit or not at all. Holdbacks and escrows solve that by leaving part of the price where the buyer can reach it. The mechanics are simple; the design questions are not.

There are usually two distinct pots and confusing them is a common drafting error. The purchase-price-adjustment escrow exists only to settle the working-capital true-up, it releases within a few months of closing, and SRS Acquiom's data puts the median at about 1% of transaction value with over 75% of 2024 deals carrying one. The indemnity escrow backs the representations, releases when survival expires, and is sized against the risks the diligence actually surfaced. Nearly 30% of 2024 deals in their sample also carried a separate special-purpose escrow for a specific identified exposure, such as a pending tax matter or a live piece of litigation, which is the right answer whenever a known risk is large and lumpy: ring-fence it, size it to the exposure, and stop taxing the rest of the deal for it.

Illustrative only: on the same $6,000,000 distribution business, a market-shaped structure holds $60,000 (1%) in a PPA escrow releasing 90 days after closing, and an indemnity escrow sized to the deal's actual risk profile releasing at the survival date. If the true-up lands at the $160,000 shortfall from the earlier example, the $60,000 escrow covers 38% of it and the remaining $100,000 has to come from the sellers directly, which is exactly why the buyer will also want a right of set-off against any earnout still outstanding, and exactly why the seller should resist letting one instrument secure another.

The levers: size, term, release schedule (a tranched release, half at six months and half at survival, often unlocks a deal where a single long hold does not), whether the escrow is the buyer's exclusive remedy or merely its first stop, and who pays the escrow agent. Exclusive-remedy language is the one to read carefully: it converts every representation into a claim capped at the escrow balance, which is either the seller's best win in the document or the buyer's worst loss, depending on which side of it you sit.

What breaks first is the seller's cash planning. Sellers routinely model the gross price, commit it, and then discover that a fifth of it is unavailable for eighteen months, that the true-up moved it again, and that tax was due on the whole amount at closing. Model the net, on the release schedule, before you sign.

Assembling a structure: allocate, then price

The sequence that produces good deals is the reverse of the one most people use. Do not start from a price and then argue about protections. Start by listing the things that could be untrue, decide who carries each one, and let the price fall out of that allocation.

Illustrative composite, drawing the pieces together on the $6,000,000 distribution business used above. Diligence surfaces four live risks: seasonality in working capital, revenue concentration in two customers, an unfiled state sales-tax exposure of uncertain size, and a founder whose relationships drive half the accounts. Each gets an instrument rather than a discount. Seasonality goes to a peg set on the trailing-twelve-month average of $900,000 with a $60,000 PPA escrow. Customer concentration goes to a two-year, $2,000,000 gross-profit earnout, worth about $420,000 at the 21-cent base rate, which both sides should say out loud, with acceleration if the buyer discontinues either account. The tax exposure goes to a ring-fenced special-purpose escrow sized to a real estimate, not to a round number. The founder goes to a transition agreement with a defined scope and an exit date, not to a vague consulting arrangement nobody enforces.

Now price it. A buyer holding all four risks unallocated should pay meaningfully less than $6,000,000 for the same business; a buyer who has allocated them is paying for the earning power rather than for the uncertainty around it. That is the whole argument for structure: it lets the price be about the business.

The honest counterweight is that every instrument costs something real. Earnouts create two years of arguments about accounting between people who now work in the same company. Escrows freeze the seller's capital. Long survival periods keep sellers awake. Complex structures need lawyers on both sides and take weeks that a seller with another buyer may not give you. Structure is not free, and past a certain density it becomes the reason a good deal dies rather than the reason it closes.

So apply the instruments where the risk is genuinely large and genuinely uncertain, and take the rest on price. A structure with two well-drafted mechanisms usually beats one with six, because the failure mode of deal documents is not insufficient protection. It is a document so intricate that neither side, two years later, can agree on what it says. General education, not legal or financial advice: the specific terms of any transaction belong with your own counsel.

Put it to work

Before you sign, list every price-moving assumption and name which side eats it if wrong. Set the working-capital peg from a trailing twelve-month average, not one month. Cap the earnout period at two years and tie it to a metric the seller still controls. Write the efforts standard explicitly. Run the working-capital-peg tool on the target's own monthly balances first.

Sources & references

Linked entries open the named source directly. Entries without a link say exactly what kind of reference they are — and how to check them yourself.

• SRS Acquiom — key deal-terms findings (earnouts, escrows, purchase price adjustments, RWI, survival) — Aggregate private-target deal-terms data from a shareholder representative; the 21-cents earnout payout, the 0.9% average adjustment, the 12-month median survival and the RWI walk-away figures are taken from this page.
• SRS Acquiom — M&A escrow and purchase-price-adjustment statistics — Source for the 93% of deals carrying a purchase-price-adjustment mechanism and the 89% of those producing an actual adjustment.
• Bristol-Myers Squibb, Form 10-K for fiscal 2020 (filed February 10, 2021) — Primary source for the Celgene CVR terms ($9.00, three milestones with dates), the 714.9 million share count, the $2.30 first-trade price, the $1,644 million CVR fair value, the $80,269 million total consideration, and the January 2021 termination.
• Johnson & Johnson, Form 10-K for fiscal 2019 (filed February 18, 2020) — Primary source for the Auris Health acquisition: approximately $3.4 billion net of cash acquired on April 1, 2019, plus contingent payments of up to $2.35 billion.
• Harvard Law School Forum on Corporate Governance — drafting guidance from the Delaware Supreme Court on earnouts, efforts obligations and fraud (February 10, 2026) — Source for the Chancery damages breakdown ($300m / $600m / $61m) and the January 12, 2026 Delaware Supreme Court holding in Johnson & Johnson v. Fortis Advisors.
• Harvard Law School Forum on Corporate Governance — Delaware Supreme Court reverses the Chancellor in Chicago Bridge & Iron v. Westinghouse (September 6, 2017) — Source for the zero-dollar upfront consideration, the $1.174 billion net working capital target, the $2.15 billion versus $428 million true-up positions, the Liability Bar, and the June 27, 2017 reversal.
• Arnold & Porter — lessons from Johnson & Johnson v. Fortis — Corroborates the $5.75 billion total including the $2.35 billion maximum earnout, and the efforts-standard drafting points.
• Worked examples in this briefing — The $6,000,000 distribution business and every figure derived from it are illustrative arithmetic written for this library, not measurements of a real transaction.