25 ChatGPT Prompts for SEO That Actually Save Time

Prompt 1 — Keyword clustering by intent

Below is a list of keywords related to [your topic]. Group them into clusters based on shared search intent. For each cluster: name the cluster, identify the primary keyword, and list the supporting variants. Return the result as a JSON object.

[paste your keyword list here, one per line]

Prompt 2 — Topic cluster map from a seed keyword

I'm building a topic cluster for the keyword "[seed keyword]". Generate:
1. One pillar page topic (broad, high-intent)
2. Eight to ten cluster page topics (specific, supporting the pillar)
3. For each cluster page, one example H2 section heading

Format as a numbered list with sub-bullets. Do not include any search metrics.

Prompt 3 — SERP intent classification from page titles

Here are the titles and URLs of the top 10 results for the keyword "[target keyword]":

[paste titles and URLs here]

For each result, classify the search intent as: informational, navigational, commercial investigation, or transactional. Then identify the dominant intent across the full set and the most common content format (guide, listicle, product page, etc.).

Prompt 4 — Content gap analysis between two pages

I'm comparing two articles targeting the keyword "[keyword]". Here is Article A (mine):

[paste article A content or H2 list]

Here is Article B (competitor):

[paste article B content or H2 list]

List the topics that Article B covers that Article A does not. Rank them by likely importance to a user searching for "[keyword]".

Prompt 5 — Question keyword generation (People Also Ask style)

Generate 20 question-format keywords a person might search when researching "[topic]". Focus on questions that imply they want a direct, factual answer — the kind that might appear in a "People Also Ask" box. Group by: Definition questions, How-to questions, Comparison questions, and Best-practice questions.

Prompt 6 — Long-tail variant generation from a head term

Given the head keyword "[head keyword]", generate 15 long-tail variants that add specificity in one of these ways: by industry, by use case, by buyer stage, or by format. Return each variant with a one-sentence explanation of the intent it represents.

Prompt 7 — Title tag variants (with character constraints)

Write 5 title tag options for a page targeting the keyword "[target keyword]". Each title must:
- Be under 60 characters
- Include the exact keyword phrase
- Use a different angle: question, benefit, number, how-to, and comparison

Return in a table: Title | Character count | Angle used

Prompt 8 — Meta description variants

Write 3 meta description options for a page titled "[page title]" targeting "[keyword]". Each must:
- Be between 150 and 160 characters
- Include a benefit or call to action
- Not repeat the title tag word-for-word

Return in a table: Description | Character count

Prompt 9 — H2 section outline for a target keyword

Generate a H2 section outline for an article targeting "[keyword]" with a [informational / how-to / comparison] intent. The article should cover the topic thoroughly for a [beginner / intermediate / expert] audience.

Format: numbered H2 list, each with a one-sentence description of what the section should cover. Also suggest which H2 should contain the primary answer block (40–60 words, answer-first).

Prompt 10 — Internal linking opportunities in a draft

Below is a list of my site's pages and their topics:

[paste page inventory: URL — Page topic]

Now here is a draft article I am writing:

[paste article draft]

Identify 5 to 8 places in the draft where I should add an internal link. For each, suggest: the anchor text, the target URL from my page inventory, and a one-sentence reason why this link is relevant at that point.

Prompt 11 — Alt text generation from image descriptions

Write SEO-optimized alt text for the following images on a page about "[page topic]". For each image, I'll describe what it shows:

1. [describe image 1]
2. [describe image 2]
3. [describe image 3]

Each alt text should: be under 125 characters, describe the image accurately, and include a relevant keyword where it fits naturally (do not force it).

Prompt 12 — Readability simplification for a dense paragraph

Rewrite the following paragraph for a general business audience (assume no prior SEO knowledge). Keep all the information intact but reduce the average sentence length, replace jargon with plain language, and aim for a Flesch Reading Ease score above 60. Do not add new claims.

[paste paragraph]

Prompt 13 — FAQPage schema from a Q&A list

Generate a valid JSON-LD FAQPage schema object for the following Q&A pairs. Use the exact text provided — do not paraphrase or summarize. Output only the JSON-LD, wrapped in a script tag with type="application/ld+json".

Q: [Question 1]
A: [Answer 1]

Q: [Question 2]
A: [Answer 2]

[continue for all Q&A pairs]

Prompt 14 — HowTo schema from numbered steps

Generate a valid JSON-LD HowTo schema for the following process. Include: name, description, and a step array where each step has a @type of HowToStep, a name, and a text field using the exact wording provided.

Process name: [name of the how-to]
Process description: [1–2 sentence description]

Steps:
1. [Step name]: [Step description]
2. [Step name]: [Step description]
[continue]

Prompt 15 — BlogPosting schema from article metadata

Generate a valid JSON-LD BlogPosting schema for the following article. Include all standard BlogPosting properties plus a speakable property pointing to CSS selectors ["h1", ".answer p"].

Headline: [article title]
URL: [canonical URL]
datePublished: [YYYY-MM-DD]
dateModified: [YYYY-MM-DD]
Author and publisher: Icarus Works (Organization @id: https://icarusworks.ai/#org)
Description: [meta description text]

Prompt 16 — BreadcrumbList schema from a URL path

Generate a valid JSON-LD BreadcrumbList schema for the following URL hierarchy:

Level 1: Home — https://[domain]/
Level 2: [Section name] — https://[domain]/[section]/
Level 3: [Page name] — https://[domain]/[section]/[slug]/

Output only the JSON-LD object, no explanation.

Prompt 17 — Organization schema from brand facts

Generate a valid JSON-LD Organization schema using the following brand information. Include sameAs links to all social profiles listed.

Name: [brand name]
URL: [website URL]
Description: [1–2 sentence brand description]
LinkedIn: [URL]
Twitter/X: [URL]
[any other verified profiles]

Use @id: [website URL]/#org

Prompt 18 — Full content brief for a target keyword

Create a detailed content brief for a [blog post / hub page / landing page] targeting the keyword "[keyword]" with [informational / commercial] intent for a [describe audience] audience.

Brand context: [describe your brand positioning in 1–2 sentences]
Primary differentiator: [what makes your take on this topic unique]
Competitor gap: [one topic competitors cover poorly that you'll cover well]

Include in the brief:
- Recommended title tag (under 60 chars) and meta description (150–160 chars)
- H1
- TL;DR answer block (40–60 words, answer-first)
- H2 section list with 1-sentence descriptions
- Required subtopics (topics that must appear based on SERP analysis)
- FAQ section with 4 planned Q&A pairs
- Internal links to suggest
- Schema types to implement
- Estimated word count

Prompt 19 — Competitive differentiation angle

I'm writing an article about "[topic]". The top-ranking results take the following approaches:

[paste summaries or H2 lists of 3–4 competitor articles]

Suggest 3 differentiation angles I could take that would make my article more useful than the existing results. For each angle, explain: what it covers that competitors miss, who would find it most valuable, and what type of evidence or examples would support it.

Prompt 20 — Featured snippet targeting: 40-55 word answer block

Write a direct, factual answer to the question "[target question]" suitable for a featured snippet or an AI answer engine quotation. Requirements:
- Between 40 and 55 words
- Start with the subject of the question (not "Yes" or "It depends")
- Use plain declarative sentences
- Include the key term "[keyword]" naturally
- No marketing language, no hedging, no citations needed

Write two versions with different opening phrases.

Prompt 21 — Listicle structure with reasoning

I'm writing a "[number]-best [topic]" listicle for the keyword "[keyword]". Generate a structured outline with:
- The list items (do not fill in specific products/tools — I will supply those)
- For each item slot: a suggested subheading format, the evaluation criteria that slot should address, and 1–2 questions a reader would want answered in that section

Also suggest: an ideal intro paragraph structure, a comparison table format, and a closing recommendation section format.

Prompt 22 — robots.txt audit for AI crawler access

Here is my current robots.txt file:

[paste robots.txt contents]

Identify any rules that would block the following AI crawlers: GPTBot (OpenAI), PerplexityBot (Perplexity), Google-Extended (Google AI training), ClaudeBot (Anthropic), and Googlebot-News.

For each blocked crawler found: show the blocking rule, explain the consequence, and suggest the corrected robots.txt line. Then provide a revised robots.txt with all recommended changes applied.

Prompt 23 — Title tag audit across a page list

Here is a list of pages from my site with their current title tags:

[paste: URL | Title tag]

Audit each title tag for:
1. Length (flag any over 60 characters)
2. Keyword presence (flag any missing the page's primary topic)
3. Duplicate titles (identify any repeated across pages)
4. Formatting issues (ALL CAPS, missing brand name where needed, etc.)

Return a table: URL | Current Title | Issue | Suggested Revision

Prompt 24 — Redirect mapping for a site migration

I am migrating a website. Here are the old URLs and their closest equivalent new URLs:

Old URL → New URL:
[old URL 1] → [new URL 1]
[old URL 2] → [new URL 2]
[continue]

Generate an Apache .htaccess redirect block (301 redirects) for each mapping. Also flag any old URLs that have no clear equivalent and should be redirected to the nearest category page or the homepage. Explain each flagged redirect decision in a comment.

Prompt 25 — Structured data debugging from rendered HTML

Here is the JSON-LD from a page's <head> element:

[paste JSON-LD]

And here is the relevant visible HTML that the schema should match (FAQ section, HowTo steps, or article metadata):

[paste HTML section]

Identify any mismatches between the schema and the visible HTML — text that is paraphrased rather than copied verbatim, missing @type declarations, or properties that don't match the page content. Return a corrected version of the JSON-LD with all discrepancies resolved.

Tasks where prompts reliably fail:

• Statistics research: Do not ask ChatGPT to provide statistics, cite studies, or confirm data points. It will hallucinate plausible figures. Statistics must come from primary sources and should be verified before inclusion in any published content.
• Competitor-specific information: ChatGPT's knowledge of specific company products, pricing, and positioning has a training cutoff and is frequently out of date or fabricated. Use it for frameworks, not for competitive intelligence.
• Live SERP state: Without browsing enabled, ChatGPT cannot tell you what currently ranks for a keyword, what featured snippets look like today, or whether a page you wrote is indexed. Use real tools for these checks.
• Final-draft B2B or technical copy: Expertise-heavy content — case studies, technical documentation, or category-defining thought leadership — requires genuine subject matter knowledge that ChatGPT cannot supply. Use it to draft structure; keep humans responsible for substance.
• Regulatory or legal language: In any context where accuracy has legal consequences, ChatGPT output requires attorney review regardless of how confident it sounds.

The practical rule: if you could not verify the output is correct by applying your own knowledge or checking a primary source, do not publish it. These 25 prompts are designed for tasks where the quality of the output depends on applying a framework — not on knowing facts that ChatGPT might invent.

For more on the right and wrong uses of AI in SEO, see how to use ChatGPT for SEO. For the GEO connection, see GEO best practices.