Skip to content
Perspectives

StandpointInternational Business Law

Why contracts fail on the statement of work

Contracts rarely fail on the fine print. They fail when both sides believe they have agreed on the performance. Why nobody feels responsible for the statement of work, and how to get it right.

Peter Poleacov (updated )15 min

Clean room at Kennedy Space Center: technicians check the fit between the Mars Climate Orbiter and the third stage of the Delta II launch vehicle.

At 09:04 universal time on 23 September 1999, radio contact with the Mars Climate Orbiter breaks off, 49 seconds earlier than calculated. The probe has nine months of flight and 669 million kilometers behind it and is due to swing into orbit that morning. Instead, it dips into the atmosphere 57 kilometers above the Martian surface. Below 80 kilometers, it cannot survive. The signal never returns.1

NASA’s investigation board found the cause in a single file on the ground. The manufacturer’s software delivered the thruster data in pound-seconds. The navigation software at the Jet Propulsion Laboratory calculated in newton-seconds, exactly as the interface specification required. A factor of 4.45, for nine months, on every single course correction. Both teams were among the best in the world. Both worked flawlessly within their own frame. Nobody caught the error, because each side believed the other had understood. That illusion cost NASA the probe. NASA had put the program at 235.9 million dollars, launches not included.2

“The great enemy of communication, we find, is the illusion of it.” The line dates from 1950 and belongs to the journalist William H. Whyte. To this day it mostly circulates as a George Bernard Shaw quote. Of all sentences, the most famous line about misunderstanding has been misattributed for seven decades.3

Contracts fail the same way. After twenty years of negotiating and drafting them, I am convinced that the most expensive contract problem is not in the fine print. It arises before the first clause is drafted, when the parties believe they have agreed on the performance and have in fact been talking past each other. Everyone works on the wording of the legal clauses down to the last detail. The description of the performance, which is the actual deal, stays vague. And because the topic belongs to no discipline, almost nobody writes about it.

Negotiations happen in the wrong place

Watch an international contract negotiation and you will see precision work in the wrong place. Days are spent discussing whether it must be “best efforts,” “reasonable best efforts,” or “commercially reasonable efforts.” In the liability exclusion, the drafters list synonym after synonym: “indirect, incidental, special, or consequential damages.” I have sat through rounds in which such formulas were argued over with devotion. Meanwhile, the statement of work appeared in every draft only as “Exhibit 1: to follow.” The software world calls this pattern bikeshedding, after Parkinson’s law of triviality. The debate takes place where everyone can have a say, not where the risk lies. That this exact pattern appears as an instruction in a 1944 sabotage manual is no accident: see “How to Sabotage an Organization”.

Kenneth Adams, author of the leading style manual for English contract language, has positioned his work explicitly as an alternative to the dysfunction of traditional contract drafting.4 He is right. But the real dysfunction lies one level higher. It is not in how something is worded, but in whether it is clear what is owed at all. Most contracts fail not on what they say but on what both sides took for granted.

How much money can be involved became clear in 2021, in the dispute between the European Commission and AstraZeneca over vaccine deliveries. Of all things, it concerned an efforts formula, the kind whose wording is otherwise negotiated for days. Before the Brussels civil court, in a contract worth 870 million euros, the case turned on the statement of work. The question was how much AstraZeneca actually had to deliver under its “Best Reasonable Efforts” promise.5 The formula had become the statement of work. The parties had even defined it in the contract, as the effort a comparable company would make in view of the pandemic. Quantities and delivery dates were in the contract, but subject to the “Best Reasonable Efforts” standard. Whether they were owed did not follow from the definition either.

The attention goes to the legal clauses. The description of what is actually owed stays a placeholder.

Why nobody feels responsible for the statement of work

Circle of competence

That is what Warren Buffett and Charlie Munger call it: know where the boundary of your own ability runs, and act inside it. In contract negotiations I regularly watch the inversion. The engineers debate warranty and liability with enthusiasm while the technical parameters, their own field, remain incomplete. The lawyers treat the statement of work as a technical annex: “Exhibit, comes from the business side.” Each discipline leaves its circle of competence precisely where it is most expensive. And nobody feels responsible for the interface.

The teams do not talk to each other

Sales promises, engineering calculates, legal drafts, often in sequence rather than together, and without a shared knowledge system. What the customer actually needs gets lost between the departments. A sales director once put it to me like this: “We are dysfunctional. But the others are more dysfunctional.” He meant it as consolation. It is a precise description of most contract processes. The winner is not the side that communicates well, but the side that communicates less badly.

Same words, different languages

International contracts are negotiated in English by people whose native language is rarely English. “Acceptance,” “commissioning,” and “availability” mean something different to each side, shaped by legal system, industry, and culture. In “The Culture Map,” Erin Meyer shows that Germany and the United States are among the cultures that make communication most explicit, while many business partners around the world are used to reading the essentials between the lines. Her rule for multicultural teams fits in one line: “Multicultural teams need low-context processes.”6 Low-context means nothing is left between the lines; everything that matters is said out loud. A contract is exactly that, the low-context process of a deal. It works only if the parties spell out in it what one of the two sides takes for granted.

The result is that the statement of work belongs to nobody. The engineers consider it a matter for the lawyers, the lawyers a matter for engineering, management a matter of detail. On liability clauses, there are countless publications. On the craft of describing a performance clearly, there is next to nothing. That is why it is the most important contract topic nobody writes about.

What applies in a dispute

The illusion of agreement is stable as long as nothing happens. It holds until the first real test. The machine runs, but at a different throughput than the buyer expected. The software works, but the interface to the legacy system was “obviously included.” From that moment, what governs is no longer the shared understanding that never existed, but the text of the contract.

And that text is interpreted under rules that surprise many. Before a German court, what counts is not what you meant but how an objective recipient was entitled to understand the declaration (Sections 133 and 157 of the German Civil Code, Bürgerliches Gesetzbuch, BGB). This applies where the two sides meant different things. The statement of work also defines what a defect is (Section 434 BGB). Without an agreed specification, you will later argue about what is “usual” and what the buyer “may expect.”

The more expensive mistake is the reverse one, the assumption that your description displaces the standard. By itself, it does not. Goods must meet the objective requirements alongside the agreed ones, “unless effectively agreed otherwise” (Section 434(1) and (3) BGB); between businesses that deviation is permitted, and whoever wants to deviate from the standard should agree on it expressly. The same applies in works contracts through the generally accepted rules of technology, which the contractor promises tacitly. A craftsman had installed a wooden staircase with stringers 40 mm thick, exactly as the contract specified; the applicable body of rules generally requires 50 mm or proof of structural stability. The BGH found a defect. Agreeing on 40 mm does not count as a deliberate departure from the minimum standard. It counts as one only if this is expressly pointed out or the customer knows of the shortfall for other reasons, for instance from their own expertise. Whether the staircase was in fact stable made no difference.7

Conversely, what nobody wrote down counts too. The use presupposed under the contract is the one both sides assumed, provided the buyer made it known to the seller at the latest when the contract was concluded, and suitability is already lacking where it is merely impaired – for instance because that use carries serious health risks or the risk of major economic damage.8

If the contract is governed by English or US law, it gets stricter. The parol evidence rule largely cuts off recourse to earlier understandings, emails, and drafts, and the customary entire agreement clause has the same effect by contract; under German law it generally does not. See the article “No oral side agreements exist. But they do!”. The CISG is more generous and expressly allows negotiations and practices to be used in interpretation (Article 8(3) CISG). Do not rely on it. Whoever invokes something that is not in the contract is in a weak position.

There is a twist that even many lawyers have never seen up close. The legal clauses of standard contracts face strict judicial review under German standard-terms law, and German courts strike down unreasonable clauses routinely. The actual statement of work, by contrast, escapes that strict review of content, but only in its narrow core.9 What directly sets the performance owed, no court will test for reasonableness. A court does review clauses that restrict or modify the promised performance. And every clause must be clear and comprehensible. This freedom from review in the core is opportunity and risk at once. Opportunity, because within that core the statement of work is the most powerful drafting tool that content review leaves untouched. What is not owed cannot be breached, and a precise statement of work often limits liability more effectively than the liability clause that fails under Section 307 BGB. This applies as long as the statement of work is clear and defines the performance rather than restricting it. Risk, because conversely nobody will correct this part of the contract for you. What a vague statement of work costs once the contract is running is set out in the article “Contract management decides your margin”: unclear scope is the most frequently named cause of contract value erosion worldwide.

What helps

First, the sequence. I keep being instructed to “go ahead and draft the contract” while the essential content, above all the statement of work, is still open. That is the Mars Climate Orbiter sequence: launch first, talk about units later. The efficient path runs the other way, content first, then wording; every draft before that is a draft of the misunderstanding. Four steps have proven themselves.

Understand the context before you draft

What is the performance supposed to deliver in the customer’s daily operations? How is success measured, by whom, with which method? The Mars test is banal and hard at once. Are the units defined? Metric, measuring method, reference conditions, tolerances. The orbiter was not lost to physics, nor to a missing agreement. The units were in the interface specification. It was lost because for nine months nobody checked whether the data delivered matched it. A specification nobody verifies is as good as none.

Get the disciplines around one table

Engineering, sales or procurement, and legal discuss the statement of work together, not in sequence. One shared hour before the first draft saves weeks of negotiation rounds and, in the worst case, years in court. To make that repeatable you need a system: defined terms, maintained templates, working knowledge management.

Concentrate on what matters

Do not regulate every eventuality; regulate the parameters the deal economically depends on: quantities, qualities, dates, delivery terms (Incoterms), interfaces, duties to cooperate, acceptance criteria. A statement of work that clearly covers the important things is better than one that covers everything.

Clear, complete, no room for interpretation

Every technical term is defined or made measurable. The test is whether a third party who knows none of the preliminary talks would understand this description the same way you do. If two readings are possible, the one argued in a dispute will be the one that hurts you.

One distinction helps here that negotiations rarely spell out. In a technically specified statement of work, the parties set out what will be delivered: values, tolerances, procedures. In a functional one, they set out what the delivered item has to be able to do. The difference is an allocation of risk. A supplier who owes values has generally performed once the values are met, as long as no further function has been agreed or is presupposed under the contract. That holds even if the plant does not do in the customer’s operation what the customer had in mind. A supplier who owes a function carries precisely that risk. For the buyer, the functional description is therefore the right choice; for the supplier, the dangerous one. In practice it comes down to the combination: defined performance values plus the function they are meant to deliver, both in the same document.

The clearest example is the popular formula “fit for purpose.” It sounds precise and is the opposite. “Purpose” can reach beyond the agreed performance values to anything the customer mentioned as an intended use in the course of the negotiations; under English law, fitness for purpose may even be read into the contract as an implied term where the buyer makes the particular purpose known and relies on the seller’s skill or judgment (Section 14(3) Sale of Goods Act 1979). If you cannot avoid the formula, tie “purpose” expressly to the specifications defined in the contract.

German law has the same risk under a different name. A contractor who promises a yard surface that lorries of up to 60 tonnes can drive on does not owe the planned method of execution but a work that withstands all foreseeable loads of that operation, including the edges nobody had in mind while drafting. If the intended execution turns out to be insufficient, the contractor remains bound to the result and the price and bears the additional cost.10

And one term is not as good as another. “Generally accepted rules of technology,” “state of the art,” and “state of science and technology” denote three different levels. The first asks what has prevailed among the experts and proven itself in practice; the second moves the yardstick to the frontier of technical progress, precisely without requiring general acceptance.11 Replace one term with the other and you shift the standard owed, usually without noticing.

And the lawyer? His place is not the legal column but the interface. Lawyers are the one discipline at the table whose craft is making things explicit; that is what they should be used for, well before the choice of law comes up. Whether a technical specification is complete, I cannot guarantee. Only your engineers can. I can ask the questions that expose the gaps, translate between the disciplines, and make sure the answers end up in the contract rather than in the minutes of a conference call.

Conclusion

The Mars Climate Orbiter did not fail on bad work. Both teams delivered, each flawless by its own standards. The probe was lost at the interface, to the illusion of having understood each other. Contracts fail the same way – rarely on the clause everyone fights over, almost always on the performance nobody talked about. By all means work on the wording of your next contract. But start with Exhibit 1.

Cover image: the Mars Climate Orbiter being mated to the third stage of its Delta II launch vehicle, Kennedy Space Center, November 1998. Image: NASA (KSC-98pc1736).

Notes

  1. NASA, Mars Climate Orbiter Mishap Investigation Board, Phase I Report, 10 November 1999: loss of signal on 23 September 1999 at 09:04:52 UTC, 49 seconds earlier than predicted; reconstructed periapsis of 57 km against a survival limit of roughly 80 km. ↩

  2. Root cause per the Phase I report: thruster data in the ground file “Small Forces” in pound-seconds instead of the “specified and expected units” of newton-seconds, off by a factor of 4.45. Cost per NASA: 193.1 million USD for development and 42.8 million USD for mission operations, 235.9 million USD in total for the twin program of Mars Climate Orbiter and Mars Polar Lander, launch vehicles not included (JPL press kit, September 1999; likewise the 1998 Mars Missions press kit). ↩

  3. William H. Whyte, “Is Anybody Listening?”, Fortune, September 1950; on the widespread misattribution to Shaw: Quote Investigator, 31 August 2014. ↩

  4. Kenneth A. Adams, A Manual of Style for Contract Drafting, 5th ed. 2023; overview at adamsdrafting.com. ↩

  5. Tribunal de première instance francophone de Bruxelles, order of 18 June 2021 – 2021/48/C, CISG-online 5607; at issue was the “Best Reasonable Efforts” clause of the vaccine supply agreement between the European Commission and AstraZeneca AB (Advance Purchase Agreement of 27 August 2020, sections 1.9 and 5.1). ↩

  6. Erin Meyer, The Culture Map, PublicAffairs 2014, pp. 34 et seq. (low-context ranking) and p. 55 (rule for multicultural teams). ↩

  7. BGH, judgment of 7 March 2013 – VII ZR 134/12, paras. 9, 12 f., 15. ↩

  8. BGH, judgment of 26 April 2017 – VIII ZR 80/16; on the version of Section 434(2) sentence 1 No. 2 BGB effective from 1 January 2022, BT-Drs. 19/27424, p. 23. ↩

  9. Section 307(3) BGB; settled case law, for example BGH, judgment of 5 October 2017 – III ZR 56/17, paras. 15 f., 27. ↩

  10. OLG Karlsruhe, judgment of 27 March 2018 – 8 U 51/15. ↩

  11. BVerfG, order of 8 August 1978 – 2 BvL 8/77 (Kalkar), reasons B II 3 b; on the distinction between the technical standards and the three-tier theory, see Mark Seibel, Abgrenzung der “allgemein anerkannten Regeln der Technik” vom “Stand der Technik”, NJW 2013, 3000. ↩

Reference: Poleacov, P. (2026). Why contracts fail on the statement of work. INN.LAW. https://inn.law/en/perspectives/statement-of-work/