Earn-outs in technology M&A: engineering metrics you can live with

Delaware's courts have spent two years telling the market what a good earn-out looks like, at nine-figure scale. Most engineering metrics fail the test. The question for a technology deal is not "revenue or EBITDA?" but "which metric survives integration, is measured by a system neither side controls, and can be adjudicated by an expert in weeks?" Here is a way to choose.

An earn-out is a bet that the parties can agree, today, on how to measure something that will happen after one of them has lost control of it. In technology deals the temptation is to measure the thing the buyer is actually paying for — the product, the platform, the engineering team — rather than a financial proxy. That instinct is right. The execution is usually wrong, because the metrics engineering teams use to run themselves were never designed to survive a hostile reading by the other side's counsel.

Why earn-outs are back in the argument

Earn-outs never went away; they swing with the bid-ask gap. SRS Acquiom's 2026 deal-terms data show that "24% of private-target (non-life sciences) deals in 2025 report an earnout, up from 19% in 2014",¹ while the American Bar Association's 2025 Private Target Deal Points Study, covering 139 acquisition agreements, records that "use of earnouts decreased from 26% during the period covered by the 2023 Study to 18%".² The two studies sample different deal populations; what they share is that roughly one deal in five now carries contingent consideration. What actually gets paid is the more sobering number. Across non-life-sciences deals, SRS Acquiom finds that "earnouts pay about 21 cents on the U.S. dollar", that where anything is paid "about half the maximum earnout dollars are paid", that 68% of deals now stack multiple metrics, and that no deal closing in 2024 had an earn-out period longer than four years.³ Litigation has tracked the money: court dockets mentioning earn-outs "almost doubled" between the first quarters of 2022 and 2023.⁴

The case that changed the conversation is Fortis Advisors v. Johnson & Johnson. In September 2024 the Delaware Court of Chancery entered a judgment of just over $1 billion against J&J over milestones in its acquisition of the surgical-robotics company Auris Health; in January 2026 the Delaware Supreme Court affirmed the breach-of-contract and fraud findings — including the $600 million awarded for the remaining regulatory milestones and the $60.9 million fraud award — vacated the $300 million attributable to the first milestone, and remanded for recalculation of the judgment; the Court of Chancery entered a stipulated judgment of about $811 million on 26 January 2026.⁵ In June 2025 the Court of Chancery had awarded more than $180 million in expectation damages against Alexion for breaching an efforts obligation in the Syntimmune deal, on top of a $130 million milestone already found payable.⁶ Buyers who assumed that "commercially reasonable efforts" was a phrase without a price have been corrected.

What the courts actually enforce

Three lessons from the Delaware decisions govern metric design, and all three cut both ways.

The efforts standard is whatever you wrote. In Auris, the parties had defined commercially reasonable efforts by reference to J&J's own conduct — efforts "consistent with the usual practice of [J&J] and its Affiliates with respect to priority medical device products of similar commercial potential at a similar stage" — and the court held J&J to its own priority-product standard, finding among other things that "more than 80 Auris personnel were diverted from the development roadmap" to a competing project.⁵ In Alexion the standard was outward-facing — a hypothetical comparable company — and the buyer's idiosyncratic corporate initiative was held to be "not a typical factor" such a company would weigh.⁶ In Himawan v. Cephalon the standard was also outward-facing — the efforts a company with substantially the same resources and expertise would use, with due regard to the cost and nature of the undertaking — and the buyer's decision to cease development, weighing development cost, the likelihood of approval and the prospects of profitable commercialisation, was held commercially reasonable even without contractually specified exit criteria — the opposite outcome on a similar standard; the Supreme Court affirmed in January 2025.⁷ Parties who want more certainty can write the permitted reasons to stop into the contract: in Menn v. ConMed the buyer's "commercially best efforts" obligation was subject to "one of several preapproved reasons" for ceasing development, including a determination that the device posed a risk of injury to patients or surgeons.⁷ And in Fortis Advisors v. Medtronic MiniMed, "sole and absolute discretion" coupled with a covenant not to act "for the primary purpose of frustrating" the milestones was held to impose minimal obligations.⁷

The implied covenant of good faith is not a safety net. Lazard Technology Partners v. Qinetiq settled in 2015 that where the contract prohibits only conduct taken "with the intent" of reducing the earn-out, the seller must prove intent, and the implied covenant cannot lower the bar for a party that "lost at the bargaining table".⁸ The 2026 Auris opinion applied the same logic in the other direction: because the parties had tied every milestone to one regulatory pathway "and only to that pathway", there was "no genuine contractual gap in the Merger Agreement for the implied covenant to fill", which is why the $300 million first-milestone award fell.⁵

Discretion clauses are enforced, and so are anti-reliance clauses — if drafted properly. The ABA's 2023 study recorded that express buyer discretion to operate the business post-closing appeared in 50% of earn-out deals (up from 33%), an express disclaimer of buyer fiduciary duty in 25% (up from 8%), and an express statement that the earn-out does not accelerate on a change of control of the target in 75% (up from 42%).⁹ In Auris the fraud claim also succeeded, because the anti-reliance provision bound only one side; practitioners now warn buyers to be careful "when commenting on the likelihood of achievement of milestones, as overstating it can lead to fraud charges".¹⁰

The design consequence is that an earn-out metric is only as good as the operating covenant, the discretion clause and the dispute mechanism around it. A metric that cannot be tied to a defined efforts standard, protected by a covenant the buyer can actually live with, and determined by a named expert is a metric that will end in Wilmington.

Four tests for a defensible metric

We apply four tests before a metric goes into a term sheet. Each corresponds to a discipline that has to sign off.

Legal: can the metric be defined before signing, without reference to judgement calls the buyer will later control, and adjudicated by a named expert under a "final and binding" clause? Ambiguity is expensive: in Fortis Advisors v. Dematic the phrase "Company Products" — and whether it captured the seller's software once integrated into the buyer's products — generated litigation that a clearer definition would have avoided.⁴
Technical: is the metric measured by a system, or a third party, that the party being measured cannot reconfigure after closing? Was it instrumented before signing, so that there is a baseline neither side can dispute?
Commercial: does the metric correlate with what the buyer paid for, and does it survive integration — rebranding, bundling, cost allocation, sales-force reorganisation — without either side being able to move it by an ordinary business decision?
Tax and employment: can the metric be paid to sellers whether or not they remain employed, without personal performance targets, so that it stays capital consideration rather than disguised remuneration?

The fourth test is the one technology deals fail most often, because the metrics that feel most natural — retention of the engineering team, delivery of the roadmap — are the ones that look most like employment income.

Engineering and product metrics that pass

Four families of technical metric survive the tests, with conditions.

Contracted service-level attainment for named services. Uptime or latency against the thresholds the target already promises its customers, measured by the monitoring stack that existed at signing, with a closed list of exclusions (planned maintenance windows, upstream provider outages, force majeure). It is objective, it is already customer-facing, and it is hard to inflate without delivering real reliability. It fails if the buyer migrates the service to its own platform without agreeing how the measurement carries over.

Gross or logo retention on a frozen customer cohort. Take the customer list at closing, schedule it, and measure how many of those customers — or how much of their recurring revenue — is still there at each test date. Gross retention "cannot exceed 100%", which removes the upsell-attribution argument; net revenue retention is acceptable only if the parties define how expansion revenue is attributed once the buyer's sales team is selling into the cohort.¹¹ Retention on a closing-date cohort survives the buyer restructuring its new-business motion, which new-logo bookings do not. For calibration, SaaS Capital's 2025 benchmarks put median net revenue retention for private B2B SaaS companies with annual contract values of $25,000–50,000 at 102%, with a top quartile of 111%.¹¹

Binary certification and product milestones with pre-agreed acceptance criteria. A SOC 2 Type II report issued with an unqualified opinion; an ISO/IEC 27001 certificate for a defined scope; a named regulatory clearance; general availability of a version against a written acceptance test. Binary milestones are cheap to adjudicate. The Auris lesson is to write the alternative pathway into the milestone: J&J's milestones were tied to a single FDA route, and when the strategy changed the seller had no contractual answer.⁵ For technical certifications, specify which certifying bodies and which scope satisfy the condition, and what happens if the standard itself is revised during the earn-out period.

Integration milestones with a written test plan and third-party acceptance. Migration of a named set of customers to the combined platform, API parity against a checklist, decommissioning of a legacy system — each verified by an agreed third party. These are the milestones most exposed to the buyer's post-closing conduct, and they are defensible only if the buyer's resourcing obligations are spelled out. Both Auris (personnel diverted to another project) and Alexion (a company-wide initiative used to justify stopping) were, at bottom, resource-allocation cases.⁵⁶

Metrics that fail

The metrics engineering leaders use internally are the ones to keep out of the purchase agreement.

The DORA four keys. Deployment frequency, lead time, change failure rate and recovery time are excellent operating metrics and terrible earn-out triggers. DORA's own guidance says the metrics "are meant to be applied at the application or service level", that "comparing metrics between vastly different applications … can be misleading", and — in an explicit warning that invokes Goodhart's law — that setting broad targets "increases the likelihood that teams will try to game the metrics".¹² Deployment frequency doubles if you split each deployment in two.

Activity metrics. Story points, velocity, commits and lines of code. The authors of the SPACE framework are categorical: "activity metrics alone do not reveal which of these is the case, so they should never be used in isolation either to reward or to penalize developers".¹³ Goodhart's law applies with unusual force to people who write the scripts that count.

Engineer headcount retention. Retaining the team is often what the buyer is paying for, and it is the metric most likely to convert capital consideration into employment income. HMRC's guidance lists, among the indicators that an earn-out is sale consideration, that "where the vendor continues to be employed, there are no personal performance targets incorporated in the earn-out" and that "non-employees or former employees receive the earn-out on the same terms as employees remaining".¹⁴ A metric that pays founders more if their engineers stay is a personal performance target in all but name. Use retention bonuses for retention; keep the earn-out for the business.

EBITDA on a buyer-controlled cost base, and new-logo bookings once the buyer owns the sales motion. The buyer can allocate overhead, add management charges, restructure compensation and redeploy the sales team; none of that is bad faith under Medtronic or Lazard, and all of it moves the number.

Drafting the operating envelope

A defensible metric still needs an envelope. Post-Auris drafting guidance is unusually specific: earn-outs "should be drafted as bespoke provisions, tailored to the specific product, business, company, industry, science, regulators"; comparator products used in an inward-facing standard should be identified by name; and the agreement should be explicit "as to whether the discretionary factors are subject to, or instead independent of, the efforts standard".¹⁰ Sellers should ask for the Auris seller's clause prohibiting decisions "based on taking into account the cost of making any Earnout Payment(s)"; buyers should ask for ConMed-style off-ramps and, where they want them, the Medtronic discretion language. Both sides should decide acceleration on a change of control deliberately — three-quarters of the earn-out deals in the ABA's 2023 sample expressly excluded it⁹ — and should cap any right of set-off against indemnity claims.

Formalities matter more than they seem. In Hughes v CSC Computer Sciences the English Commercial Court held in 2025 that earn-out determinations served by email were invalid because the notices clause "did not contemplate service of notices or other communications by email".¹⁵ Records matter too: the courts read board decks and effort logs, so, as Jones Day advises, "buyers should maintain records of their commercial reasons for taking actions relating to acquired programs that are subject to earnout payments".¹⁶

Dispute design

Choose the forum for the metric, not the metric for the forum. Financial metrics go to an independent accountant acting as an expert rather than an arbitrator, with scope limited to items in the objection notice and no independent investigation; the Delaware Superior Court confirmed in AM Buyer v. Argosy that such determinations, marked final and binding, are enforceable absent fraud or "clear and manifest error".¹⁶ Technical milestones go to a named technical expert with the same scope limits and a written acceptance protocol. Efforts disputes — whether the buyer did enough — cannot be delegated to an accountant and will end in arbitration or court, which is why the metric should be chosen to minimise them. In every case: monthly metric statements from the buyer, seller access to the underlying data, defined objection windows and explicit fee-shifting if intended.

Tax and employment guardrails in Europe

The three jurisdictions we work in most reach similar places by different routes. In the United Kingdom, HMRC's Employment Related Securities Manual sets out the indicators of genuine sale consideration — including that the earn-out is not conditional on continued employment beyond "a reasonable requirement to stay to protect the value" — and warns that an earn-out partly rewarding services will be apportioned.¹⁴ In Sweden, a tilläggsköpeskilling is by default capital income where the selling shareholder-employee receives a market salary in addition to it and there is no clear link between the payment and the profit generation of continued employment; payments offered only to those who stay, or tied to post-sale value creation by the individual, risk reclassification as salary.¹⁷ In Spain, an earn-out received by an individual seller is a capital gain, but the seller must estimate the total consideration including the contingent element in the year of sale and rectify later, or elect instalment treatment; corporate sellers recognise the income when it can be estimated with reasonable reliability.¹⁸ The drafting rule is the same in all three: sellers who leave and sellers who stay get the earn-out on identical terms, with no personal targets, and retention is bought separately.

The checklist

Before an engineering metric goes into a term sheet, it should pass the four tests, be measured by a system frozen at signing, sit inside a named efforts standard with either off-ramps or a no-cost-consideration clause, be determined by a named expert on a final-and-binding basis, and pay leavers and stayers alike. Part 2 of this series explains how diligence findings become the delivery risks an earn-out is meant to carry; Part 5 deals with the related problem of founders who remain after the sale.

Disclaimer

This article is general commentary for professional readers. It is not legal, tax or financial advice, does not address the circumstances of any particular transaction or taxpayer, and does not create an advisor–client relationship with Cabo Advisory SL. Case law, deal-terms data and tax guidance are stated as at the date of writing and change; the tax treatment of contingent consideration depends on the facts and the jurisdiction, and specialist advice should be taken before terms are agreed.

Notes and sources

Kip Wallen, "Earnout and Milestone Trends in Private-Target M&A Deals" (drawing on the 2026 SRS Acquiom M&A Deal Terms Study), SRS Acquiom, 2026. https://www.srsacquiom.com/our-insights/ma-earnout-milestone-trends/ ↩
Jessica Cabral Pearlman and Tatjana Paterno, "Announcing the ABA's 2025 Private Target Mergers & Acquisitions Deal Points Study", ABA Business Law Today, 16 December 2025. https://www.americanbar.org/groups/business_law/resources/business-law-today/2025-december/aba-2025-private-target-mergers-acquisitions-deal-points-study/ ↩
SRS Acquiom, "M&A Deals: Key Trends from the 2025 Deal Terms Study" (2,200+ private-target acquisitions closed 2019–2024), 2025. https://www.srsacquiom.com/our-insights/ma-deals/ ↩
F. Dario de Martino, Clare O'Brien, Mara Goodman, Sarah Ference and Michael Zinzi (A&O Shearman), "The Art and Science of Earn-Outs in M&A", Harvard Law School Forum on Corporate Governance, 11 July 2025. https://corpgov.law.harvard.edu/2025/07/11/the-art-and-science-of-earn-outs-in-ma ↩ ↩²
Johnson & Johnson v. Fortis Advisors LLC, No. 490, 2024 (Del. 12 January 2026), affirming in part and reversing in part Fortis Advisors LLC v. Johnson & Johnson (Del. Ch. September 2024). Opinion: https://courts.delaware.gov/opinions/download.aspx?id=389980 ; summary: https://law.justia.com/cases/delaware/supreme-court/2026/490-2024.html ; on the remand judgment, Selendy Gay, "Selendy Gay Obtains Trial and Appellate Victories to Secure Largest Earnout Award in Delaware History", 30 January 2026, https://www.selendygay.com/news/general/2026-01-30-selendy-gay-obtains-trial-and-appellate-victories-to-secure-largest-earnout-award-in-delaware-history ↩ ↩² ↩³ ↩⁴ ↩⁵
John A. Barker et al. (Goodwin), "Delaware Court Awards More Than $180 Million for Breach of Commercially Reasonable Efforts Obligation in Drug Development Milestone Dispute" (Shareholder Representative Services LLC v. Alexion Pharmaceuticals, Inc., Del. Ch., liability 5 September 2024, damages 11 June 2025), 18 June 2025. https://www.goodwinlaw.com/en/insights/publications/2025/06/alerts-lifesciences-cldr-delaware-court-awards-more-than-180-million ↩ ↩² ↩³
Ted Powers III and Jonn R. Beeson (Jones Day), "Earnouts in M&A Transactions: Recent Decisions From Delaware", 2 April 2025, https://www.jonesday.com/en/insights/2025/04/earnouts-in-ma-transactions-recent-decisions-from-delaware ; Tom Reems and Madeline Maersk-Moller (Squire Patton Boggs), "Recent Court Decisions Provide Guidance on Earn-outs", April 2025, https://www.squirepattonboggs.com/insights/publications/corporate-insights-recent-court-decisions-provide-guidance-on-earn-outs/ (discussing Himawan v. Cephalon, Inc., Del. Ch. 30 April 2024, aff'd Del. 24 January 2025; Menn v. ConMed Corp., C.A. No. 2017-0137-KSJM, Del. Ch. 30 June 2022; and Fortis Advisors LLC v. Medtronic MiniMed, Inc., Del. Ch. 29 July 2024). ↩ ↩² ↩³
Lazard Technology Partners, LLC v. Qinetiq North America Operations LLC, No. 464, 2014 (Del. 23 April 2015). https://law.justia.com/cases/delaware/supreme-court/2015/464-2014.html ↩
Fasken, "What's Market in Private M&A? Highlights From the ABA's 2023 Private M&A Deal Points Study (US Deals)", 15 February 2024. https://www.fasken.com/en/knowledge/2024/02/highlights-from-the-aba-2023-private-ma-deal-point-study-us-deals ↩ ↩²
Fried Frank, "Drafting Guidance from Delaware Supreme Court on Earnouts, Efforts Obligations, and Fraud", Harvard Law School Forum on Corporate Governance, 10 February 2026, https://corpgov.law.harvard.edu/2026/02/10/drafting-guidance-from-delaware-supreme-court-on-earnouts-efforts-obligations-and-fraud/ ; see also Mayer Brown, "Delaware Law Alert: New Perspectives on Earnouts", 5 February 2026, https://www.mayerbrown.com/en/insights/publications/2026/02/delaware-law-alert-new-perspectives-on-earnouts ↩ ↩²
Nick Perry, "What is a Good Retention Rate for a Private SaaS Company in 2025?", SaaS Capital, 18 September 2025. https://www.saas-capital.com/blog-posts/what-is-a-good-retention-rate-for-a-private-saas-company/ ↩ ↩²
DORA (Google Cloud), "DORA's software delivery performance metrics", dora.dev, updated 5 January 2026. https://dora.dev/guides/dora-metrics-four-keys/ ↩
Nicole Forsgren, Margaret-Anne Storey, Chandra Maddila, Thomas Zimmermann, Brian Houck and Jenna Butler, "The SPACE of Developer Productivity", ACM Queue 19(1), 2021. https://queue.acm.org/detail.cfm?id=3454124 ↩
HM Revenue & Customs, Employment Related Securities Manual, ERSM110940 "Securities Options: earn-outs: key indicators of earn-out being sale consideration", updated 5 May 2026. https://www.gov.uk/hmrc-internal-manuals/employment-related-securities/ersm110940 ↩ ↩²
Richard Burrows and Dominic Sedghi (Macfarlanes), "Corporate law update: 8–14 March 2025" (discussing Hughes v CSC Computer Sciences Ltd [2025] EWHC 302 (Comm)), 14 March 2025. https://www.macfarlanes.com/insights/102loe9/corporate-law-update-8-14-march-2025/ ↩
Jones Day, "Earnouts in M&A Transactions: Recent Decisions From Delaware", 2 April 2025 (discussing AM Buyer v. Argosy, Del. Super. 3 September 2024). https://www.jonesday.com/en/insights/2025/04/earnouts-in-ma-transactions-recent-decisions-from-delaware ↩ ↩²
Hannes Næss, "Ska tilläggsköpeskilling beskattas som lön?", Morris Law, 25 August 2022. https://www.morrislaw.se/aktuellt/ska-tillaggskopeskilling-beskattas-som-lon ↩
Jorge Borrajo Dios, "¿Qué es el earn out en la compraventa de empresas y cómo se tributa?", Kreston Iberaudit, 22 July 2025. https://kreston.es/blog/earn-out-compraventa-empresas/ ↩