Copilot Prompt Patterns for In-House Counsel Contract Review

In-house counsel often spend hours reviewing contracts for risk, compliance, and consistency. Copilot can speed up this process when you use the right prompt patterns. The key is structuring prompts to get precise, clause-specific responses instead of vague summaries. This article covers prompt patterns for contract review, including risk identification, obligation extraction, and compliance checks.

Why Prompt Patterns Matter for Contract Review

Copilot processes natural language, but vague prompts produce generic answers. For contract review, you need Copilot to focus on specific clauses, parties, and legal risks. Prompt patterns are reusable templates that force Copilot to return structured, actionable output. Without a pattern, you might get a paragraph that misses key obligations or risks. With a pattern, you get a table, a list, or a direct comparison. This saves time and reduces the chance of missing a critical term.

Before using these patterns, ensure Copilot has access to the contract document. In Microsoft Word, open the contract file. In the Copilot pane, select the document as the source. For Microsoft 365 Chat, attach the file or paste the text. Copilot works best with documents under 50 pages. For longer contracts, break the review into sections.

Prerequisites for Using Copilot for Contract Review

You need a Copilot for Microsoft 365 license. The contract document must be in a supported format: .docx, .pdf, or .txt. For .pdf files, ensure the text is selectable, not scanned images. Copilot cannot read scanned PDFs without OCR. If your contract is an image PDF, use a PDF-to-text tool first. Also, Copilot respects Microsoft 365 sensitivity labels. If the contract is labeled as confidential, Copilot may block access. Adjust the label or grant yourself explicit permission.

Pattern 1: Risk Identification Prompts

Risk identification prompts ask Copilot to scan the contract for high-risk clauses. The pattern is: Identify [number] [type] clauses in this contract and explain why each is [risk level]. Replace the bracketed terms with your specific needs. For example: “Identify three high-risk indemnification clauses in this contract and explain why each is risky.” Copilot returns a numbered list with clause location and risk explanation. This pattern works best when you specify the clause type. Without it, Copilot may pick general clauses like termination or confidentiality, which might not be your priority.

Example Risk Identification Prompt

Prompt: “Identify five clauses that could result in financial liability for the buyer in this contract. For each, state the clause number, the potential liability amount, and the condition that triggers the liability.” Copilot generates a table with columns: Clause Number, Liability Amount, Trigger Condition. This gives you a quick financial risk map. If Copilot cannot find dollar amounts, it will state “not specified” in the amount column. This still helps because you know the clause exists but lacks a cap.

Pattern 2: Obligation Extraction Prompts

Obligation extraction prompts force Copilot to list duties for each party. The pattern is: List all obligations of [party] in this contract. Include deadlines, deliverables, and remedies for breach. For example: “List all obligations of the vendor in this contract. Include delivery deadlines, acceptance criteria, and remedies for late delivery.” Copilot outputs a table with columns: Obligation, Deadline, Deliverable, Remedy. This pattern is useful for comparing obligations across multiple contracts. You can copy the table into Excel and build a compliance tracker.

Advanced Obligation Extraction

For complex contracts, add a condition: “Only list obligations that have a specific performance standard, such as ‘commercially reasonable efforts’ or ‘best efforts.'” This filters out vague duties and highlights enforceable standards. Another variation: “List all obligations that survive termination of the contract.” Copilot will scan for survival clauses and list obligations like confidentiality, indemnification, and audit rights. This helps you prepare for post-termination compliance.

Pattern 3: Compliance Check Prompts

Compliance check prompts verify that the contract meets a specific legal standard. The pattern is: Check if this contract complies with [regulation] [article or section] requirements for [topic]. For example: “Check if this contract complies with GDPR Article 28 requirements for data processing agreements.” Copilot compares the contract text to the regulation and lists compliant and non-compliant clauses. If Copilot lacks the regulation text in its training data, it may not perform the check. In that case, paste the regulation text before the contract text in the prompt. For example: “Here is GDPR Article 28 text: [paste text]. Now check if this contract complies with those requirements.”

Example Compliance Check Prompt

Prompt: “Check if this contract complies with California Consumer Privacy Act requirements for consumer data rights. List any missing clauses and suggest language to add.” Copilot generates a compliance gap table with columns: Requirement, Status, Suggested Language. This gives you a draft for negotiation. Always verify Copilot’s suggestions with a legal source, as Copilot may hallucinate case law or outdated regulations.

Pattern 4: Redline and Comparison Prompts

When you have two versions of a contract, use a comparison prompt. The pattern is: Compare this contract to the attached revised version. List all changes in clause structure, language, and risk level. Copilot outputs a table with columns: Clause, Original Text, Revised Text, Risk Change. This pattern is faster than manual redline review. For best results, ensure both documents are in the same format and have clear clause numbering. If clauses are unnumbered, Copilot may misalign the comparison. In that case, add: “Use the first five words of each clause as the identifier.”

Example Comparison Prompt

Prompt: “Compare the attached original contract to the revised version. Identify any clauses where the revised version increases liability for the buyer. For each, state the original risk level and the new risk level.” Copilot returns a risk change table. This helps you focus negotiation on high-risk changes without reading every word.

Common Issues and How to Adjust Prompts

Copilot Returns Generic Summary Instead of Structured Output

This happens when the prompt lacks output format instructions. Add a format directive: “Return the results as a table with columns: Clause Number, Risk Description, Suggested Mitigation.” Copilot respects explicit format requests. If it still returns a paragraph, rephrase the prompt with “table” or “list” in the first sentence.

Copilot Misses Clauses or Obligations

Copilot may miss clauses if the contract is long or uses non-standard language. Break the contract into sections and run a separate prompt for each section. For example: “Review only Section 5 (Indemnification) of this contract. List all obligations of the indemnifying party.” This reduces context length and improves accuracy. Also, avoid ambiguous terms like “review the contract” without specifying what to look for.

Copilot Hallucinates Legal Requirements

Copilot may invent regulations or case law. Always verify compliance suggestions with a legal database. To reduce hallucination, provide the exact regulation text in the prompt. For example: “Here is the text of GDPR Article 28: [paste]. Based on this text, check if the contract complies.” This grounds Copilot in the source material.

Prompt Pattern	Best Use Case	Expected Output Format
Risk identification	Scanning for high-risk clauses	Numbered list with clause location and risk explanation
Obligation extraction	Listing duties for a specific party	Table with Obligation, Deadline, Deliverable, Remedy
Compliance check	Verifying against a regulation	Table with Requirement, Status, Suggested Language
Redline and comparison	Comparing two contract versions	Table with Clause, Original Text, Revised Text, Risk Change

You now have four reusable prompt patterns for contract review. Start with risk identification for a quick scan. Then use obligation extraction to build a compliance tracker. For regulated industries, run a compliance check prompt with the regulation text pasted. Save your best prompts as templates in a Word document or OneNote for reuse. As a next step, try the comparison pattern on two versions of a non-disclosure agreement to see the risk change table in action. For advanced use, combine patterns in a single prompt: “Identify high-risk clauses, then list buyer obligations, then check compliance with GDPR Article 28.” Copilot can handle multi-step prompts if you separate each request with a clear delimiter like “Next.”