Standard terms: template, generator, or lawyer?

On 4 June 1996, the first Ariane 5 lifts off from Kourou: Europe’s new launcher, ten years in development, four research satellites on board. The control software is proven: the alignment module comes from Ariane 4, where it worked flawlessly for years. Why rewrite what works?

For 36 seconds, the flight is nominal. Then the inherited module reports a velocity it never saw on Ariane 4, because the new rocket is much faster in this phase of flight. A number overflows. The backup system does not help. It runs the same software with the same flaw and fails for the same reason. The nozzles swing to full deflection, and at second 39 the launcher breaks apart and self-destructs. The loss is estimated at around 370 million dollars.

The remarkable part is in the ESA inquiry report. The software was not defective. After lift-off, the module simply kept running without any purpose (“as soon as the launcher lifts off, this function serves no purpose”), and it did so only because Ariane 4 had needed it that way. Proven code, wrong context.

This is exactly how generic standard terms fail, not immediately, not visibly, but under the first real load, and then completely. The fact that a set of clauses worked elsewhere for years says nothing about whether it carries your business. Why German standard-terms law reaches every pre-formulated clause, and how strictly it reviews them, is set out in the entry article Standard terms in B2B: freedom of contract is the exception.

Do I need standard terms at all?

Yes, for three reasons. First, legal certainty. Without your own terms, the statute governs, and the statute knows neither a liability cap in your favor, nor an extended retention of title, nor your payment and delivery terms. Second, efficiency. Standard terms standardize day-to-day business instead of negotiating every single order.

The third reason, defense, is the least known. If only the other side puts forward standard terms and incorporates them validly, they apply in full, with every clause working against you. Only your own terms create the collision in which conflicting clauses cancel each other out and the neutral statute takes their place (on the “battle of forms,” see “Incorporating standard terms in international business”). Having no standard terms is therefore not a neutral position. In case of doubt, it means the other side’s terms apply.

Four sources, three fail predictably

Most companies get their standard terms from one of four sources: a template, a generator, a lawyer – external or their own in-house counsel – or the internet.

outline = your business

templatethe average: too much and too little

generator/AIa random cut: gaps remain

lawyermade to measure

The fourth source is quickly dealt with. It means copying another company’s terms, preferably from the same industry. That does not save money; that is Russian roulette. You adopt, sight unseen, the flaws, the risks, and the legal vintage of someone else’s business, plus a copyright risk on top. Might work out, might not. That leaves three serious sources. Two of them cost little and fail predictably. Because the choice is not “cheap and fast versus expensive and thorough.” The choice is between a document that reflects the average and one that fits your specific case and survives judicial content review.

The three sources, one by one.

The template: built for the average

A template has to fit thousands of companies at once. So it governs what they all have in common, and that is the average. Your industry, your customer and supplier structure, your liability profile, and your international business do not appear in it. That is no criticism of the publishers; templates, as a rule, state in the text itself that they are no substitute for advice on the individual case. It is in the nature of the thing.

Templates are also static. They reflect the state at the time of their publication, while the case law – national and international – and the statutes keep moving. The template never learns about it, and neither does its user.

Legally, there is no grandfathering. An adopted template contains your own standard terms from the first use and is subject to full content review. “It came from the trade association” is not an argument in court. The template is the Ariane 4 software – proven, cheap, and built for a different system.

The generator: no judgment, no responsibility

Generators, these days mostly AI-based, sound like the contemporary answer. You answer a few questions and download a finished set of terms. Three properties disqualify them for anything beyond standard business.

First, AI-based tools are not deterministic. The same input can produce one text today and a different one tomorrow. For a document whose entire value lies in its reliability, that is a design flaw, not a detail.

Second, there is no judgment on the facts. The generator asks for what you already know. It does not investigate how your contracts are actually concluded, what can go wrong in your supply chain, or which clause carries your specific risk. In my experience, the most expensive gaps are the ones nobody asked about.

Third, nobody stands behind the result. If the clause falls in court, the generator is not liable. The responsibility stays entirely with you.

This is not an objection to the technology; I work with AI every day. But a text produced without looking at the specific facts, and without anyone standing behind it, is not legal drafting. It is a bet.

Why generic text falls in court

That template and generator terms fail is not lawyer’s rhetoric; it follows from three rules of German law on standard terms. And a fourth risk comes from outside.

Content review applies between businesses too, and it is stricter than almost everyone believes

An empirical study commissioned by the Federal Ministry of Justice showed that companies systematically underestimate the reach and the strictness of this review.¹ At its core, only the description of the goods and services escapes the review. Practically every other pre-formulated clause is on the test bench – in my assessment, 95 percent of all contract clauses. And hardly anyone has on their radar that clauses in court mostly fail simply because they are intransparent (Section 307(1) BGB), not drafted in a clear and intelligible way. Whether they also unreasonably disadvantage the contract partner does not even come into play anymore.

That this strictness is politically contested does not help you in a dispute. Reform of the law on standard terms in business-to-business dealings has been demanded for years and is being debated again, among other things as a brake on the new commercial courts; nothing has been enacted.² You have to draft under the law as it stands today.

What is prohibited toward consumers is an indicator between businesses

If a clause falls under the prohibited-clause catalog of Section 309 BGB in consumer business (B2C), that indicates in business between companies (B2B) as well that it unreasonably disadvantages the contract partner and is therefore invalid.³ The case itself involved a classic from the forms trade, a sale “excluding any and all warranty.” The clause fell in its entirety, and the seller was liable as if no exclusion had ever been agreed.

If the clause falls, it falls completely

German courts do not rescue an overreaching clause by trimming it back to what would just barely be permissible; there is no validity-preserving reduction. The statute takes the clause’s place (Section 306(2) BGB). The limitation of liability that covers one case too many ends up limiting nothing at all.

Invalid clauses can draw warning letters from competitors

The dispute with your contract partner is not the only risk. The Federal Court of Justice has treated the use of invalid standard terms as a breach of professional diligence and classified the prohibitions in Sections 307, 308 No. 1, and 309 No. 7(a) BGB as market-conduct rules (Section 4 No. 11 UWG in its former version, today Section 3a UWG). Anyone using such clauses can therefore be served a warning letter (Abmahnung) by competitors and qualified institutions, without any dispute at all, simply because the terms are published, for instance in an online shop.⁴ This is settled for consumer business; for purely business-to-business dealings the Court left the question open. In my view, much speaks for its applying there too: Section 307 BGB applies directly between businesses, and professional diligence binds there as it does elsewhere. Betting on invalid clauses being harmless in B2B is a risk. In the decided case, a trader had given a cease-and-desist undertaking after a warning letter and kept using the clause; that cost him EUR 10,200 in contractual penalties.

How predictably the very clauses found in templates and generator output fall is documented across this series:

• “As far as legally permissible”: the popular safety tail makes the clause intransparent and therefore invalid, in B2B too.
• “No individual agreement”: the pre-formulated confirmation that the parties negotiated individually is itself an invalid standard term.
• “No oral side agreements exist”: the entire-agreement clause does not hold; the individual agreement always prevails.
• All-caps clauses: capitalization in the US style is no substitute for valid drafting under German law.

And even the perfect text does not apply by itself

Suppose the template were flawless: it would still be worthless as long as it never becomes part of the contract. Standard terms have to be incorporated into every single contract, and in international business that means transmitting them rather than linking to them, in the right language, with an answer to the other side’s conflicting terms (in detail: “Incorporating standard terms in international business”).

How hard this stage is shows in a case from overseas trade. A carrier had squeezed its terms onto the back of the form so tightly that they were readable only with a magnifying glass, and even then only with effort. They never became part of the contract, and the jurisdiction clause buried in them fell with them – against a merchant, and with a clause customary in the trade.⁵ The clause was not surprising, but it was illegible. More on this in “All Cap Clauses”.

Incorporation is not a text; it is a business process. It means implementation in the existing routines of procurement and sales, and integration into the other legal documents – from the purchase order and the order confirmation to contracts and the code of conduct. That is precisely why no template and no generator can solve it. The download ends where the real work begins.

What the lawyer does differently

Standard terms from a lawyer cost more because something different stands behind them: not a text, but judgment on your specific case, and a system that carries the text. Three levels.

Built for your specific case

Terms of sale and terms of purchase follow your industry, your customer and supplier structure, and your corporate structure, not the average. A distributor needs the extended retention of title, a plant manufacturer needs a different liability architecture than a software house, and a company selling through commercial agents faces different incorporation questions than the operator of a webshop. The clause that is right there is wrong here.

International validity and enforceability

This is where generic text fails most silently and most expensively. The usual jurisdiction clause reads confidently and, against a contract partner outside the EU, often delivers nothing but a judgment nobody there will enforce, because with many states that are among Germany’s most important trading partners, there is no treaty on the recognition of court judgments. In my webinars I put it this way: a judgment I cannot enforce against my contract partner abroad is not enforcement, it is decoration. The clause only becomes effective once it fits the market – outside the EU, often as an arbitration clause, whose award is enforceable in more than 170 states under the New York Convention.

The arbitration clause is not a blind spot here. It is itself a standard-terms clause and is reviewed as such, under the law of the seat of arbitration. What it does not deliver along the way is an escape from the law on standard terms. A contractor sought a declaration that arbitration was inadmissible because the contract expressly waived Sections 305 to 310 BGB, exposing it to a possibly excessive contractual penalty. Without success: the validity of the arbitration agreement is independent of the validity of the parties’ agreements on the procedure, and the arbitration clause stood.⁶ Whether the waiver itself holds, the BGH did not decide; that is for the arbitral tribunal. In the literature it is considered valid, provided it was individually agreed or at least does not take the contract partner by surprise.⁷ Choosing the seat and the governing law therefore also decides how far the review of standard terms reaches. That belongs at the start of the drafting, not in the boilerplate.

There are only three ways out of the law on standard terms, and each has its price. The first is genuine negotiation: a clause that is individually agreed leaves content review behind – but the requirements are demanding, and a pre-formulated confirmation that the parties negotiated is precisely what does not help (see “No individual agreement”). The second is choosing another law, often Swiss law. It only works where the case genuinely reaches abroad; in purely domestic business the parties remain bound by mandatory German law under Article 3(3) Rome I, and that includes the law on standard terms.⁸ The third is the arbitration agreement, with the uncertainty described above. None of these is a clause you copy out of a template.

The rest of the international business needs the same precision: a deliberate decision on the CISG instead of the reflexive opt-out, the Incoterms clause that matches the actual supply chain, a tactical but validly drafted choice of law, and incorporation as a practiced process.

The system after the document

Standard terms are not a project with an end date. The lawyer’s work includes what comes after delivery: monitoring the case law and updating the terms, training procurement and sales – because that is where incorporation is decided in day-to-day business – reflecting compliance requirements on both the purchasing and the sales side, from supplier codes of conduct to sanctions clauses and pass-through of supply-chain duties, and a maintained, documented clause library instead of a PDF in a folder.

Does it pay off? As a matter of experience, not statistics: a single silent clause defect costs more in a dispute than the entire project: in money, and in the business relationship, which rarely survives the fight.

The decision at a glance

When a template is enough

So that this piece does not end as advertising in my own cause, the honest cross-check: there are settings in which a carefully read, current template is defensible. Five conditions must come together:

• purely domestic business,
• standard services,
• small volumes,
• manageable liability exposure,
• no compliance requirements from contract partners.

Anyone using a template there should follow three rules:

• Choose a current, serious source. A good template states its as-of date; no date, or older than two years: hands off.
• Understand every clause instead of copying it.
• Delete the known failure clauses, above all “as far as legally permissible,” the entire-agreement clause, and the blanket exclusion of liability.

Above that threshold is anyone who sells or buys internationally, carries substantial liability risks, or genuinely depends on their clauses when it matters. If even one of the five falls away, the template is no longer thrift; it is a postponed loss event.

Conclusion

Ariane 5 did not fail because of bad software; it failed because of proven software in the wrong context. Templates and generators give you exactly that: text that worked elsewhere and knows nothing about your system. Whether it carries your business is something you find out in flight.

Frequently asked questions

Are standard terms from a generator legally reliable?

Nobody stands behind them. A generator knows neither your business model nor your supply chain, and AI-based tools are not deterministic: the same input can produce different texts. Nobody tests whether a clause survives judicial content review on your specific facts, and the responsibility stays entirely with you. For a standard transaction with low risk, the result may be adequate; you should not rely on it.

May I copy another company’s standard terms?

I advise against it. Standard terms can be protected by copyright, so copying them can trigger cease-and-desist claims. Above all, someone else’s terms govern someone else’s business: their products, their risks, their processes. Once adopted, they are your own standard terms from the first use (Section 305(1) BGB) and are subject to full content review, with every flaw of the original.

What does it cost to have a lawyer draft standard terms?

I work on fixed fees: the price is set before the work begins and depends on the value for your company, the complexity of the business model, the degree of internationality, and the topics to be covered. For perspective: a single silent clause defect, say an invalid limitation of liability, regularly costs a multiple of the entire project once a dispute arises.

How often should standard terms be reviewed?

On triggering events, and at regular intervals. A trigger is any change to the business model, such as new products, new markets, or a new sales channel, and any relevant court decision; the courts keep striking down clauses that passed as safe only yesterday. Beyond that, a fixed review rhythm keeps the terms from quietly going stale; my recommendation: every two years.

Notes

1. Leuschner, NJW 2016, 1222. ↩

2. Berger/Reiss, NJW 2025, 1311; skeptical: Kraner, RInPrax 2025, 150. ↩

3. BGH, default judgment of 19 September 2007 – VIII ZR 141/06, the so-called “Gleichschritt” (lockstep) case law. ↩

4. BGH, judgment of 31 May 2012 – I ZR 45/11. ↩

5. BGH, judgment of 30 May 1983 – II ZR 135/82. ↩

6. BGH, order of 9 January 2025 – I ZB 48/24. ↩

7. Pfeiffer, NJW 2025, 866. ↩

8. Pfeiffer, NJW 2012, 1169. ↩

Reference: Poleacov, P. (2026). Standard terms: template, generator, or lawyer?. INN.LAW. https://inn.law/en/perspectives/standard-terms-template-generator-lawyer/

---