Last updated:
Contract Clause Character Design - Word Counts, Sentence Length, and Legal Clarity
Nobody reads the contracts they agree to, and the reason is arithmetic rather than laziness. A 2008 Carnegie Mellon study of privacy policies found that a single policy averaged roughly 2,500 words and took about 10 minutes to read. One document at that size is manageable; the burden appears when the number of documents a person accepts is multiplied by that reading time, because every service added to a routine adds another policy and another set of terms. Length in legal documents is therefore not only an inconvenience to the reader - it is a design decision that determines whether the obligations in a clause can actually be located and understood. Understanding the relationship between word count, sentence length, and legal effectiveness is essential for anyone who drafts or reviews contracts.
What Determines the Length of a Legal Document
The length of a contract is set by the number of matters that have to be agreed, not by the drafter's verbosity. That distinction explains why document types cluster at consistent sizes, and why the sizes cannot be reduced by editing alone.
A non-disclosure agreement is short because its skeleton is fixed and narrow: define what counts as confidential information, state the obligation not to disclose or use it, set the term, and add the boilerplate that makes the whole thing operable (governing law, dispute resolution, severability). There is very little else to decide, so NDAs converge on a compact form that decades of practice have already refined. An acquisition agreement sits at the opposite extreme for the same structural reason. Representations, warranties, and indemnities exist to allocate risk over the specific circumstances of the target company, so every unusual lease, pending claim, or key employee arrangement is transcribed into another clause. The document is long because the deal has that many moving parts.
Terms of Service occupy the middle and grow over time, which is why they are the type most often complained about. Each new feature introduces its own issues - user-generated content needs a licence, payments need refund and chargeback handling, messaging needs conduct rules - and the terms accumulate accordingly. Estimating reading time is a useful way to see the cost that accumulation imposes on the reader, but the estimate is a symptom, not a cause.
This is where length reduction is usually attempted in the wrong place. What can safely be cut is duplication and the "just in case" provision that restates a default rule the governing law already supplies. What cannot be cut is a substantive issue, because deleting an issue does not shorten the contract so much as move it into the category of things the parties never resolved. A contract that is short because it left questions open has not gained clarity; it has converted a drafting problem into a legal risk that surfaces only when the parties disagree.
Sentence Length and Interpretive Ambiguity
Long sentences in contracts are not merely hard to read - they generate ambiguity that a court must eventually resolve, often against the drafter's intent. The damage is done by specific structures rather than by length as such, and there are three that account for most of it.
The first is distance between the subject and its verb. When a clause names the obligated party and then travels through several qualifications before stating what that party must do, the reader has to carry the subject forward across the whole span, and a second candidate subject introduced along the way can quietly capture the verb. The second is stacked conditionals. Once a sentence contains more than one condition, the scope of any exception becomes uncertain: it may attach to the nearest condition, or to all of them, and the sentence itself offers no way to tell. The third is a modifier that can attach in two directions. A phrase such as "suppliers and delivery locations designated by the Purchaser" can be read as designating both items or only the first, and each reading assigns the designation right to a different party.
A restriction clause of the kind found in software licences shows the second structure at work: "The Licensee shall not, and shall not permit any third party to, directly or indirectly, reverse engineer, decompile, disassemble, or otherwise attempt to derive the source code, underlying ideas, algorithms, structure, or organization of the Software, except to the extent that such restriction is expressly prohibited by applicable law." That is 49 words in one sentence, and the closing exception is the problem. Does "except to the extent" qualify only the attempt to derive source code, which it immediately follows, or every prohibition in the list? Both readings are available on the face of the text, and the difference decides whether a licensee in a jurisdiction that protects interoperability may decompile at all.
The remedy is structural rather than numerical: one obligation per sentence, with conditions and exceptions moved out into numbered paragraphs or sub-paragraphs where each one visibly attaches to the item above it. Splitting a clause this way often makes the total text longer while making the meaning single-valued, which is the trade worth taking. Word count is useful here as a way to measure the result, not as the target - a clause can be short, score well on a readability formula, and still be dangerous, because no formula can tell whether an exception has one reading or two.
The Plain Language Movement
The push for comprehensible official and consumer documents has produced concrete obligations in some places and a published standard in others.
In the United States, the Plain Writing Act was signed in October 2010. Its scope is narrower than it is often described: it applies to the covered documents that federal executive agencies issue to the public, not to private contracts. Internationally, the field acquired a formal reference point with ISO 24495-1, published in 2023, which sets out plain language principles for authors of any kind of document.
The practical guidance that agencies publish is worth reading closely, because of what it does and does not contain. The United States federal plain language guide for writers is built on four principles: prefer shorter words, keep sections short, write in the active voice, and use the present tense. Notably, it sets no numeric threshold at all - no target word count for a sentence, no character limit for a section. The instruction is about the shape of the sentence, and the shape is what improves comprehension.
Of those four, converting passive constructions to active voice is the one that pays off most directly in contract drafting, and for a reason that has nothing to do with style. Write "the work shall be completed promptly" and the sentence is grammatical, complete, and silent about who must complete anything. Rewriting it in the active voice forces the drafter to name a party, and the clauses where no party can be named are exactly the clauses where the obligation was never allocated. The active version is often the longer one, and for a contract that is the right outcome: making the obligated party explicit matters more than shortening the sentence. For more on how legal documents handle length constraints in digital contexts, see Legal Document Length.
Readability Metrics for Legal Text
Standard readability formulas can be applied to a contract, but it helps to know what they actually measure before treating a score as a verdict.
Formulas in the Flesch family work from two surface quantities: how long the sentences are, and how long the words are. Both are proxies for effort, and legal text scores poorly on them for structural reasons rather than accidental ones. Clauses are long because conditions and exceptions are packed into single sentences, and legal vocabulary is long because it is heavily Latinate. Legal drafting reaches for "notwithstanding" where general prose would use "despite," "indemnification" where it would use "compensation," and "hereinafter" where it would use "below." Each of those choices adds characters without adding meaning, and the effect compounds across a document, since the same handful of terms recurs in every clause.
The limit of these formulas is that they were never designed for legal text and cannot see the property that matters most in it. A score is computed from sentence and word lengths alone, so a clause with a dangling exception and a clause whose scope is unmistakable can receive an identical score. Readability measurement is therefore a screening tool: a poor score is a reliable signal that something is worth rewriting, while a good score is not evidence that a clause is unambiguous. For techniques on reducing text length while preserving meaning, see Error Message Design, which covers similar principles in a software context.
Practical Guidelines for Contract Drafting
The most counterintuitive principle in contract drafting is that a shorter contract is not a less protective one. Courts interpret ambiguous terms against the party that drafted them under the contra proferentem rule, which means every ambiguity a drafter introduces operates as a liability for the drafter's own side. A long contract assembled from long sentences and stacked exceptions accumulates those liabilities faster than it accumulates protections, so brevity that comes from resolving issues cleanly is a legal strategy rather than a stylistic preference.
Defined terms deserve the same scrutiny, because each one asks the reader to hold a meaning that differs from ordinary usage. "Material" in a contract does not mean fabric; it means significant enough to affect the agreement. That substitution is cheap once and expensive in bulk. When a document defines more terms than a reader can reasonably track, the defined meanings stop being retrieved and the words get read in their ordinary senses instead, which is how a carefully constructed definition section turns into a source of disputes. Cross-references behave similarly: each one interrupts reading to send the reader elsewhere, and a clause that cannot be understood without three lookups is effectively unread.
Taken together, these point at the same discipline. Decide every issue the deal actually raises, give each obligation its own sentence with a named party, push conditions and exceptions into a visible hierarchy, and define only the terms that genuinely carry a special meaning. The character count of a contract is ultimately a proxy for the cognitive load it imposes, and managing that load is the core challenge of legal drafting.