Most AI prompts for SEO produce generic output because they ask for generic things. A prompt like “write an SEO article about content marketing” hands the model no context, no constraints, and no output format. What comes back is a 700-word Wikipedia summary you’d delete on read-through. The difference between prompts that save you 3 hours a week and prompts that waste 20 minutes is specificity, and there are 7 templates I use daily on real client work that consistently produce usable drafts.
You’ll get each template with its exact structure, the variables to swap in, and what the output should look like. Every template below has been tested against ChatGPT-5, Claude Opus 4.7, and Gemini 2.5 Pro. The framework works across all three with minor tuning.
Why Most AI Prompts for SEO Fail the First Try
The failure pattern in AI prompts for SEO is consistent. Prompts without a defined persona, output format, or constraint list default to what LLMs were trained to produce, averaged, safe, generic content. OpenAI’s own 2025 prompting guide showed that prompts with explicit output structure generated responses rated 47% higher by human reviewers than unstructured prompts on the same task.
Good AI prompts for SEO do four things at once. They assign a specific role (“You are a technical SEO strategist”), they provide context (“The site is a B2B SaaS with 180 blog posts and a DA of 32”), they specify output format (“Return a markdown table with columns X, Y, Z”), and they set hard constraints (“No more than 60 characters per meta title, no em-dashes, no hedging phrases”). Skip any one of the four and output quality drops measurably.
The templates below all follow this pattern. Use them as skeletons. The specifics matter more than the wording.
Template 1: The Content Brief Generator
This one replaces 45 minutes of manual brief writing. Feed it a target keyword and competitor URLs, and it returns a writer-ready brief with angle, outline, and semantic coverage list.
Prompt structure: “You are a senior SEO content strategist. I need a content brief for the keyword [KEYWORD]. Top 3 ranking pages are [URL1], [URL2], [URL3]. The target reader is [AUDIENCE]. My site’s existing coverage includes [LIST OF 5 RELATED POSTS]. Return: (1) recommended angle that differentiates from top 3, (2) suggested H2 outline with 5 to 7 sections, (3) must-include entities and terms, (4) suggested word count based on competitor depth, (5) 2 internal link opportunities from my existing posts. No generic advice.”
The output quality jumps significantly when you paste the actual top 3 page text into the prompt. Without it, the AI invents what competitors might have written. With the real text included, you get accurate gap analysis.
Template 2: The Meta Title and Description Batch
One prompt, 20 meta tag sets, all compliant with length limits. This is the highest-impact template for sites refreshing 50+ pages at once.
Prompt structure: “You are an SEO copywriter. For each of these 20 URLs, write a meta title (50-60 characters, includes focus keyword) and meta description (130-155 characters, includes focus keyword, ends with an action phrase). Return as a markdown table with columns: URL, Current Title, New Title, Current Description, New Description. Do not use colons unless necessary. Do not use the word ‘ultimate’ or ‘comprehensive’. Here’s the data: [PASTE ROWS].”
Run this against a CSV export of your current meta tags from Screaming Frog. You’ll get back a ready-to-import sheet. I tested this on a 200-URL refresh project and it cut the work from 2 days to 3 hours, with the only manual work being final review for brand voice.
Template 3: The Internal Link Opportunity Scan
Every content manager has asked “what should I link to from this new post?” and then spent 30 minutes scrolling through their archive. This template does the scroll for you.
Prompt structure: “You are an information architect. I’m publishing a new post titled [TITLE] on [TOPIC]. Here’s my existing post archive as title and URL pairs: [PASTE 50-150 PAIRS]. Return 5 internal link opportunities from my new post to existing posts, with (1) suggested anchor text, (2) which paragraph or section to add the link in, (3) one-sentence reason why the link helps the reader. Skip any that feel forced.”
The output accuracy depends on title clarity. Posts with vague titles like “Tips and Tricks” get overlooked. Posts with specific titles like “How to Fix Core Web Vitals in WordPress” get selected appropriately. Clean up your title inventory before running this at scale.
Template 4: The SERP Intent Classifier
Before committing to a content angle, you need to know what SERP type dominates the query. This prompt saves the 15 minutes of manual SERP analysis per keyword when you’re processing a batch.
Prompt structure: “You are an SEO researcher. For each keyword below, identify (1) primary intent (informational, commercial, navigational, transactional), (2) secondary intent if present, (3) dominant SERP format (listicle, how-to, comparison, definition, video), (4) whether featured snippets, People Also Ask, or AI Overview likely appear, (5) suggested content format to match. Here are the keywords: [LIST]. Return as a table.”
For highest accuracy, paste actual SERP screenshots or top-result snippets alongside each keyword. Without that grounding, AI models guess based on keyword patterns, which works for obvious cases and fails on ambiguous ones like “amazon reviews” or “best bank 2026.”
Template 5: The Content Gap Finder
Traditional gap tools cost $100+ per month. This template surfaces most of what they find for the cost of a prompt.
Prompt structure: “You are a content strategist. Here are the headlines and H2s from my top 3 competitors on the topic of [TOPIC]: [PASTE]. Here are the headlines and H2s of my existing content on the same topic: [PASTE]. Identify: (1) topics my competitors cover that I don’t, (2) topics I cover that they don’t, (3) topics where I cover them shallowly and they go deep, (4) 3 prioritized content ideas that would fill the biggest gaps, with suggested word count for each.”
The output isn’t perfect. Expect to verify 2 or 3 of the suggested topics against actual SERP data before committing. But as a starting point for a quarterly content planning session, it saves 4 to 6 hours of manual comparison work.
Template 6: The Rewrite for E-E-A-T
This one takes an existing thin post and strengthens it against quality rater criteria without requiring a full rewrite.
Prompt structure: “You are a senior editor reviewing content against Google’s E-E-A-T quality rater guidelines. Here’s an existing post: [PASTE]. The author is [NAME] with expertise in [SPECIALTY]. Return: (1) 3 places where first-hand experience language would strengthen credibility (with suggested rewrites), (2) 2 claims that need source citations (with suggested sources), (3) the opening paragraph rewritten to show expertise in the first 2 sentences, (4) 1 sidebar or callout that would signal author expertise. Keep all changes consistent with the author’s voice.”
Results are surprisingly strong on posts with clear author attribution and a named specialty. Posts with generic “team” bylines get weaker suggestions because the AI has nothing specific to anchor to.
Template 7: The Technical SEO Audit Summary
When a 400-issue Screaming Frog export lands in your inbox, this template prioritizes the 20 things that actually matter.
Prompt structure: “You are a technical SEO consultant. Here’s an audit export with every issue flagged: [PASTE OR CSV]. Group issues by (1) impact on crawl budget, (2) impact on rendering, (3) impact on index coverage, (4) impact on Core Web Vitals. Return the top 20 issues in priority order, with: issue name, affected URL count, estimated traffic impact, and specific remediation step. Skip low-impact issues like short meta descriptions unless they affect more than 50% of pages.”
Pair this with real GSC crawl stats data in the prompt context. Without crawl data, AI can’t accurately estimate impact, it defaults to treating every issue as equally important, which defeats the purpose.