The CREAC method in legal writing is a five-part structure — Conclusion, Rule, Explanation, Application, Conclusion — that organizes an analysis so it opens and closes with the answer, with the reasoning stacked between. The defining feature is the commitment up front: the writer states the predicted outcome in the first sentence, then walks the reader through the law and the facts that support it. Most American law schools teach it as the default framework for legal memoranda, and supervising attorneys and judges expect the same architecture in practice.
The Five Parts and What Each One Does
The elements form a chain. Each depends on the one before it, and skipping a link shows.
Conclusion. One or two sentences answering the legal question directly. Not a summary of the analysis. A prediction. For a predictive memo, that reads something like: “A court will likely find that the employer violated the statute because the termination occurred within 30 days of the employee’s complaint.”
Rule. The governing legal standard, drawn from statutes, regulations, or controlling case law. State the law in the abstract, without referencing the client’s facts. If the claim has multiple elements, list them all here so the reader sees the full test before any analysis begins.
Explanation. A narrative showing how courts have applied the rule in earlier cases. Naming cases is not enough. Describe the facts those courts treated as significant and the reasoning they used to reach a result. This section builds the reader’s sense of what the rule actually means in practice.
Application. A direct comparison between the client’s facts and the precedents laid out in the Explanation. This is where the legal reasoning lives. Draw parallels to favorable cases, distinguish unfavorable ones, and show why the client’s situation should come out the way the opening Conclusion predicted.
Conclusion. A restatement of the answer, now reinforced by the analysis. A sentence or two that ties the Application back to the opening prediction and gives the reader a clean landing.
How CREAC Differs From IRAC
The gap between CREAC and its older cousin IRAC is smaller than it first looks. IRAC opens with the Issue, phrased as a question. CREAC opens with the Conclusion, phrased as an answer. CREAC also splits what IRAC calls the Rule into two distinct stages: the abstract Rule and the case-law Explanation. Separating them forces the writer to think through each step and gives the reader a clearer path.
Many practitioners treat the two acronyms as interchangeable shorthand for the same core skill. The fundamentals are identical. CREAC just makes the architecture more explicit, and the front-loaded conclusion suits how supervising attorneys and judges read: they want the answer first.
Where Counterarguments Go
Competent legal analysis addresses the other side’s best argument. CREAC has a specific place for it: inside the Application, after the main analysis is complete. Sequence matters. Presenting the affirmative case first lets the reader absorb the writer’s reasoning before running into the opposing view. Dropping a counterargument into the middle of the main analysis breaks the reader’s understanding of both.
The technique is straightforward. Identify the strongest argument the other side could make, state it fairly, then explain why it fails or carries less weight than the writer’s position. A memo that ignores obvious weaknesses signals inexperience or dishonesty, and neither helps the attorney who has to assess real risk. One structural rule: any case cited in a counterargument must already appear in the Explanation. Introducing a new case in the Application, without groundwork, breaks the framework’s logic.
Building the Rule From Multiple Cases
Most legal issues are not governed by one clean statute. The rule usually comes from piecing together holdings across several cases, each of which refined or narrowed the standard. Rule synthesis is the skill of merging those holdings into a single coherent statement that can then be applied to the client’s facts.
The goal is something that reads like a checklist. Start with the broadest formulation of the rule from the most authoritative case, then layer in refinements from later decisions. Some cases broaden the rule. Others narrow it, carve out exceptions, or add factors. Each case plays a specific role in the final synthesis, and the writer should know what that role is before drafting the Explanation.
When a synthesized rule becomes unwieldy, look for natural dividing lines. If different standards apply to different parties or to different categories of conduct, break the rule into separate statements rather than cramming everything into one paragraph. A reader who has to re-read a rule statement three times has already lost the thread.
Organizing an Analysis With More Than One Issue
Real legal problems rarely involve a single issue. A breach-of-contract dispute might require separate analyses of formation, breach, and damages. Each discrete issue gets its own complete CREAC block, with its own opening conclusion, rule, explanation, application, and closing conclusion. Subissues get separate subheadings.
Before the first CREAC block, the memorandum needs an umbrella paragraph. That paragraph sits under the main heading and orients the reader to the full scope of the analysis. It states the broad governing rule, identifies which elements are contested and which are undisputed, and previews the conclusion for each sub-issue. It does not define elements or dive into case law. That work belongs in the individual CREAC sections below. Think of the umbrella paragraph as a table of contents in prose: a reader should be able to glance at it and know what is coming and how it turns out.
For sub-issues that themselves break into further sub-parts, each level gets its own umbrella. The structure nests: a main umbrella covers three elements, each element gets its own CREAC block, and any element with contested sub-factors gets another umbrella before those sub-blocks begin.
Predictive Versus Persuasive Use
CREAC works for both objective memoranda predicting how a court would rule and persuasive briefs arguing for a specific outcome. The skeleton stays the same. Tone and strategy shift.
Predictive Writing
A predictive memo is advisory. The audience is usually a supervising attorney who needs an honest assessment of the client’s chances. The Conclusion states the most likely outcome regardless of whether it favors the client. The Explanation presents favorable and unfavorable precedents with equal weight. Hiding bad facts in a predictive memo is a serious mistake. The attorney relying on it needs to know the weaknesses before deciding how to proceed.
Persuasive Writing
A persuasive brief argues for a result. The audience is a judge. The Conclusion becomes an affirmative assertion rather than a prediction: “The court should grant summary judgment” instead of “The court will likely grant summary judgment.” Lead with the strongest argument. Place the most compelling facts at the beginning and end of each section, where they have the most psychological impact. Frame unfavorable facts by surrounding them with favorable context rather than ignoring them.
Point headings do heavy work in a persuasive brief. Neutral labels give way to argumentative conclusions. Instead of “Whether the Fourth Amendment Was Violated,” a persuasive heading states the conclusion and incorporates the key fact: “The Officers Violated the Fourth Amendment When They Continued Searching After the Defendant Revoked Consent.” If a judge reads nothing but the headings, the structure and result of the argument should still come through. And when addressing the court, “the court should” tends to land better than “the court must.” No judge wants to be told what to do.
Preparing the Underlying Research
Drafting a CREAC analysis without thorough research is where most writers get into trouble. Pin down every element of the claim or defense before writing a word.
Start with the governing standard. If the issue involves a federal statute, find the statute’s text and the elements courts have identified for a claim under it. Federal district courts, for instance, have jurisdiction over civil actions arising under federal law under the general federal-question jurisdiction statute.1Office of the Law Revision Counsel. 28 USC 1331 – Federal Question If the issue turns on common law, the rule comes from case synthesis rather than a single statutory provision.
Then gather cases from the highest available courts that applied the rule to facts resembling the client’s. For each case, note three things: the relevant facts, the court’s holding, and the reasoning connecting the two. Cases from higher courts carry more authority, but a factually close lower-court decision can be more useful in the Explanation than a Supreme Court case with distant facts. The point is precedents that mirror the client’s circumstances closely enough to support a persuasive comparison in the Application.
Verifying AI-Generated Citations
Generative AI tools have become common in legal research and drafting, and pulling AI output into a CREAC analysis carries real ethical risk if the citations are not checked. The most visible cautionary example came in 2023, when attorneys in a personal injury case submitted a brief containing six fabricated case citations generated by ChatGPT. The court imposed a $5,000 penalty under Rule 11 and required the attorneys to send copies of the sanctions order to every judge whose name appeared on a fake opinion.2Justia Law. Mata v Avianca Inc, No 1:2022cv01461 – Document 54
The underlying ethical rule predates AI. Lawyers have a duty of candor toward any tribunal, meaning they cannot make false statements of law or fact and must disclose directly adverse legal authority in the controlling jurisdiction even if the other side does not raise it.3American Bar Association. Rule 3.3 – Candor Toward the Tribunal AI-generated text does not change that obligation. It just creates new ways to violate it by accident.
A growing number of federal courts now require affirmative disclosure when AI tools are used to prepare filings. Some require lawyers to name the specific tool, identify which portions of the filing involved AI-generated content, and certify that every legal citation was independently verified. The District of Kansas, for example, issued a standing order in January 2026 giving the court discretion to require sworn statements from filers regarding AI use, including identification of the tool and certification that the filer personally verified every cited authority.4United States District Court District of Kansas. Standing Order 26-01 – Use of Artificial Intelligence Similar requirements exist in federal districts across Pennsylvania, New Jersey, North Carolina, California, and elsewhere, though the specifics vary by court and sometimes by individual judge.
The practical lesson for anyone drafting CREAC analyses with AI assistance is simple. Treat every citation, every case holding, and every statutory reference as unverified until you have pulled up the source yourself and confirmed it says what the AI claims. That is the minimum standard for competent practice, and Rule 11 backs it up: every legal contention in a filing must be warranted by existing law or a good-faith argument for changing it, and sanctions run to whatever a court finds sufficient to deter repetition.5Legal Information Institute. Federal Rules of Civil Procedure Rule 11 – Signing Pleadings, Motions, and Other Papers; Representations to the Court; Sanctions