We’ve written about what living inside an earnout feels like. This two-part series goes underneath that post to the machinery: today, how earnout targets actually get built (the metric, the period, the math), and in part two, the specific ways they go wrong. Because the first lesson of the term sheet is that “20% of the price as an earnout” is not a term. It’s a category. The real terms are which number, measured how, over what window, with what slope, and every one of those choices moves real dollars from likely to unlikely or back.
Choosing the metric: the seller’s ranked list
Earnout metrics come in three families, and they are not equal.
Revenue-based targets pay you for the top line of the acquired business hitting thresholds. For sellers, this is the good family: revenue is observable, hard to manipulate downward by accident, and largely insulated from the buyer’s cost decisions. If the buyer loads your P&L with corporate allocations and platform fees, your revenue earnout doesn’t care. The buyer’s objection is symmetrical: you could hit revenue targets with unprofitable growth, which is why revenue earnouts often arrive with a margin guardrail attached. Fine. A guardrail is negotiable arithmetic.
EBITDA-based targets pay on the profit of a business unit inside someone else’s company, computed by someone else’s accountants, after integration decisions you don’t control. This is the family that generates the disputes, because post-close EBITDA contains a dozen judgment calls (overhead allocation, shared services, migration costs, who pays for the new sales hire), and each one defaults toward the ledger’s owner. If you must accept an EBITDA earnout, the definitions section becomes the most valuable pages in your purchase agreement: named accounting standards, listed exclusions, and a worked example with actual numbers.
Retention-based targets pay for clients (or revenue from a defined client list) still being there at measurement dates. When the earnout exists because of client concentration, this is the honest version: it measures precisely the risk the buyer priced, nothing else rides along, and both sides can count logos without an accountant. Its weakness is that retention is partly the buyer’s performance (their service quality drives churn too), which argues for the operating covenants part two will cover.
The seller’s ranking, then: retention or revenue where you can get it, EBITDA only with definitions bolted down, and never a metric you couldn’t compute yourself from data you’ll have access to. That last clause is its own term: audit rights and data access belong in the agreement.
The measurement period: shorter, and in your era
Time is risk. A 12-month earnout measures a business still recognizably yours: your team mostly intact, your clients mid-transition, your influence real. A 36-month earnout measures the buyer’s company, three reorganizations later, with your fingerprints fading and integration decisions compounding. The market norm for MSP deals runs 12 to 24 months, and every month past that is a month you’re betting on someone else’s execution.
Watch the calendar mechanics too. When does the clock start (close, or the first full quarter, which matters if closing lands mid-December)? Are targets annual, cumulative, or measured at each anniversary? Cumulative structures let a strong year two rescue a soft year one; annual cliffs make each period its own coin flip. And insist the measurement periods end with defined true-up dates and payment deadlines, because “within a reasonable time after determination” is how an earned payment becomes a negotiation.
Caps, floors, and the shape of the slope
The last piece of machinery is the payout curve, and its shape matters more than its endpoints. All-or-nothing cliffs (“full $500k at $2.0M EBITDA, zero at $1.99M”) are dispute factories: they concentrate the entire earnout into an argument about the last dollar, and they hand whoever controls the accounting a reason to find that dollar or lose it. Sliding scales (linear payout between a floor and a cap, say 80% to 120% of target) pay proportionally for partial performance, keep both sides’ incentives smooth, and remove the knife-edge. Add a floor that pays something for a near-miss, and a cap you can live with, since buyers will insist on one.
Two refinements worth asking for: a catch-up provision (an under-target year one can be recovered by an over-target year two) and acceleration triggers (the earnout pays out, at target or better, if the buyer sells the company again, terminates your employment without cause, or materially breaches the operating covenants). Acceleration is the clause nobody needs until the platform gets sold in month nine, and then it’s the only clause that matters.
The negotiating posture
Underneath all the machinery, keep the strategic frame from price versus terms: an earnout is contingent money, and its design determines the contingency. Set targets off a realistic base (the trailing numbers diligence verified, not the buyer’s synergy model), keep the metric in the family you can influence and audit, keep the window short, shape the slope, and price the whole instrument honestly when comparing offers. And keep the best alternative on the table at every step: smaller earnout, more cash at close. Every dollar you move across that line is a dollar that never needs a definitions section.
Part two covers where even well-built earnouts go wrong, and the contract language that prevents it.