ChatGPT for GEO: The Essential Prompts

What it does: This prompt gives you a clear score (1-25) across five GEO dimensions—so you know exactly where your content stands and what to fix first. Think of it as a diagnostic report, not vague advice.

Your input: One complete content piece. Works for blog posts, landing pages, product pages, or any content you want AI to cite.

Expected output: A structured scorecard with five dimension ratings, the specific issue in each area, one concrete fix per dimension, and your overall GEO readiness score.

What to do next: If you score below 15, focus on the fixes before investing in promotion. Score 15-20 means you're in good shape but have room to grow. Above 20? Your content is optimized—move on to tracking.

What it does: AI engines quote specific passages, not entire articles. This prompt finds your existing quotable passages and—more importantly—reveals where you're missing them entirely. It's like having an editor highlight exactly what AI can extract.

Your input: Content with clear section headings. The more sections, the more useful the analysis. Works best with 3+ distinct sections.

Expected output: A section-by-section breakdown showing your existing capsules (quoted exactly), quality ratings (Strong/Weak/Missing), and rewritten versions for anything that needs improvement.

What to do next: For any section rated "Missing," use the improved version the AI suggests as your new opening. For "Weak" sections, edit to remove context dependencies—make each answer stand alone.

What it does: Your content might be excellent—but the opening buries the answer. This targeted prompt rewrites just the first paragraph to lead with what AI can quote. Surgical fix, not a full overhaul.

Your input: Your current opening paragraph and a brief description of what the article covers. The AI needs topic context to write a relevant answer-first opening.

Expected output: A single rewritten paragraph (30-50 words) that answers the main question directly, includes one supporting fact, and stands alone as a quotable passage.

What to do next: Replace your current opening with the new version. Then check if the transition to your second paragraph still flows—you may need to adjust one connecting sentence.

What it does: Your competitor keeps appearing in AI answers. Why? This prompt reverse-engineers their content to reveal exactly what makes it citation-worthy—structure, authority signals, quotable passages—so you can apply the same patterns.

Your input: Content from a competitor that AI currently cites for your topic. Copy the full text from their page—the AI needs to see what's actually working.

Expected output: A breakdown of their structural patterns, authority signals, answer density, and—critically—the gaps they're not covering that you could exploit.

What to do next: Focus on the "Gaps I Can Exploit" section. Create content that covers what they miss with better structure. You're not copying—you're outflanking.

What it does: You can't optimize for queries you don't know exist. This prompt generates the actual questions people type into ChatGPT, Perplexity, and Claude about your topic—organized by intent type so you know which ones to target first.

Your input: Your topic and target audience. Be specific—"B2B SaaS onboarding" gives better results than just "software."

Expected output: 15 queries split into informational, comparative, and actionable categories. Each query includes the search intent and whether your current content covers it.

What to do next: Actually test these queries in ChatGPT and Perplexity. Note who gets cited. Use Prompt 5 to analyze their content. This turns theoretical optimization into real competitive intelligence.

What it does: Generic bios ("John is passionate about marketing") don't build AI trust. This prompt creates a bio designed for entity recognition—specific credentials, measurable achievements, and connections to verifiable entities that AI can cross-reference.

Your input: Your name, current role, years of experience, credentials/certifications, measurable achievements (with numbers), company affiliations, and any publications or speaking engagements.

Expected output: A 100-150 word bio in third person, factual tone. Heavy on specifics, light on fluff. Includes "sameAs-worthy" mentions—entities AI can verify independently.

What to do next: Use this bio consistently across all platforms: your website, LinkedIn, guest posts, podcast appearances. Every repetition reinforces your entity signals.

What it does: Sporadic testing tells you nothing. This prompt creates a structured protocol—specific queries to test, platforms to check, and a schedule to follow—so you can track AI visibility systematically over time.

Your input: Your brand or topic name and a description of your target audience. The more specific your audience description, the more relevant the test queries.

Expected output: 10 test queries (5 informational, 5 comparative), a testing template for ChatGPT/Perplexity/Claude, and a weekly tracking schedule—all formatted as a markdown table you can copy to a spreadsheet.

What to do next: Run the first round of tests immediately to establish a baseline. Then follow the weekly schedule. Patterns emerge after 4-6 weeks of consistent tracking.

What it does: Your citations dropped. Or spiked. Either way, you need to understand why. This prompt analyzes before/after data to identify the most likely cause—so you can fix problems or replicate wins.

Your input: Your before and after citation data from tracking: which queries you were cited for, how often, and in what context. The more detail you provide, the sharper the analysis.

Expected output: A root cause analysis considering content changes, competitor activity, AI model updates, authority signal shifts, and query pattern changes. Plus a recommended action based on whether visibility went up or down.

What to do next: If visibility dropped, implement the recommended fix and re-test in 2 weeks. If visibility increased, document what worked and apply the same pattern to other content.