15 ChatGPT Prompts for SEO & Marketing (Copy-Paste Library)
A library of copy-paste ChatGPT prompts for SEO and marketing: keyword research, content briefs, on-page, meta tags, and auditing your AI visibility.
Updated · Published
Frequently asked questions
What are the best ChatGPT prompts for SEO?
The most useful ChatGPT prompts for SEO cover keyword research, content briefs, on-page optimization, meta tags, content refreshes, and competitor research. The best ones are specific: they give ChatGPT your topic, audience, and goal rather than a vague request.
Can ChatGPT do keyword research?
ChatGPT can brainstorm keywords, cluster them by intent, and suggest related questions and subtopics, which speeds up research. It doesn't have live search-volume data, so pair its output with a dedicated keyword tool for accurate metrics.
Which SEO tasks should you never use ChatGPT for?
Anything requiring a fact it has to recall. Search volumes, keyword difficulty, competitor rankings, backlink counts, current SERP features and supporting citations are all invented with total confidence. The reliable uses are transformations of data you supply — clustering an export, rewriting titles, drafting schema, summarizing a crawl. If the prompt asks ChatGPT to know something rather than reshape something, pair it with a tool that has the data.
How do I write good SEO prompts?
Give ChatGPT clear context (topic, audience, intent, format), specify the output you want, and constrain it (length, tone, structure). Iterate: refine the prompt based on the first response rather than expecting a perfect answer on the first try.
ChatGPT speeds up the repetitive parts of SEO (keyword brainstorming, briefs, meta tags, content refreshes, competitor research) when you give it specific, well-structured prompts. This page is a copy-paste library of prompts for the most common SEO and marketing tasks, plus a set for auditing your own visibility in AI search. Treat ChatGPT's output as a fast first draft to refine, not a finished deliverable.
Key takeaways
ChatGPT is best for ideation, drafting, and analysis, not live search-volume data.
Specific prompts (topic, audience, intent, format) beat vague ones every time.
Always edit and fact-check the output; quality still decides SEO results.
The most strategic prompts below audit how AI describes your brand and category.
Copy any prompt, replace the bracketed placeholders with your specifics, and iterate on the response. For the bigger picture of how AI is reshaping search, see our pillar on AI for SEO.
Before the library, the one distinction that predicts every good and bad result here.
Transformations work. Recalls do not. Give ChatGPT data and ask it to reshape that data, and it is excellent. Ask it to retrieve a fact from memory, and it will produce a confident, plausible, wrong answer.
Clustering 500 keywords you exported is a transformation. Asking for their search volumes is a recall, and the numbers will be fabricated. Rewriting 80 title tags against a template is a transformation. Asking which competitors rank for a term is a recall. Summarizing a page you paste in is a transformation. Asking for a study that supports your argument is a recall, and it will invent a citation that looks entirely real.
Every prompt below is written as a transformation. Where one needs outside data, it says so.
ChatGPT won't give you accurate search volume, but it's excellent for expanding a seed topic, clustering by intent, and surfacing the questions real users ask.
Act as an SEO strategist. For the topic "[topic]" and an audience of"[audience]", generate 30 keyword ideas grouped by search intent(informational, commercial, transactional, navigational). For each group,note the likely funnel stage.
List 20 questions people ask about "[topic]", phrased the way a real userwould type them into a search engine or ChatGPT. Group them into themes Icould turn into content clusters.
Turn a target keyword into a structured brief a writer can execute.
Create a content brief for an article targeting "[keyword]". Include: aworking title, the search intent, a suggested H1, 6-10 H2/H3 headings, keyquestions to answer, entities and subtopics to cover, and 3 internal-linkideas. Audience: "[audience]". Goal: "[goal]".
Use ChatGPT to pressure-test an existing page against its target query.
Here is the content of a page targeting "[keyword]": [paste content].Assess how well it satisfies the search intent, list what's missing versuswhat a top result would cover, and suggest specific improvements to headings,structure, and direct answers.
Fast, on-brand title and description options that respect length limits.
Write 5 SEO meta titles (≤60 characters) and 5 meta descriptions (≤155characters) for a page about "[topic]" targeting "[keyword]". Make themcompelling enough to earn a click, and include the primary keyword naturally.
Here is an older article: [paste content]. Identify outdated information,gaps versus current best practices on "[topic]", and 5 specific additions orupdates that would make it more useful and more likely to be cited.
Synthesize positioning and content gaps from material you provide.
Compare my page [paste or summarize] with a competitor's page [paste orsummarize] for the query "[keyword]". Where does the competitor cover thetopic more thoroughly, and what specific sections or angles should I add towin the comparison?
Technical work is mostly rule-following against a published spec, which is where models are strongest. Paste in the export rather than describing it.
Here is a crawl export as CSV. Group the issues by type, then rank thegroups by likely impact on organic traffic for a [site type] site.For each group, give the fix and roughly how long it takes.Ignore anything cosmetic.[paste CSV]
Here are 200 lines of server log. Summarize crawl behaviour by directory:which sections are crawled most, which have not been visited, and anypattern in the user agents. Flag anything that looks like wasted crawl budget.[paste log]
Generate JSON-LD for a [page type] page with these fields: [fields].Use schema.org types only, no custom properties. Output valid JSON-LDin a single script tag and nothing else.
Validate the schema output before shipping it. Models produce syntactically perfect JSON-LD with invented property names often enough that unvalidated output is a real risk.
Here is a list of my published URLs with their titles and target keywords.Suggest internal links I am missing: for each suggestion, give the sourcepage, the destination page, and the anchor text. Prioritize links thatpoint at pages with commercial intent.[paste list]
Here is the body text of one page and a list of my other URLs with titles.Identify 5 places in this page where an internal link would genuinely helpthe reader, and write the sentence it should sit in.[paste page and list]
This is the set most SEOs overlook. Because buyers increasingly ask ChatGPT before they search, it's worth asking ChatGPT what it thinks about your brand and category, then reading the answers critically.
What do you know about [your brand]? Describe what we do, who we're for, ourpricing if you know it, and how we compare to alternatives. Note anythingyou're unsure about.
What are the best tools for "[your category]"? List the options you'drecommend and why.
Compare [your brand] and [competitor]. Which would you recommend for"[use case]", and what are each one's strengths and weaknesses?
Run these in both ChatGPT's base mode and its search mode, and repeat them, since answers vary. If your brand is missing, described inaccurately, or framed worse than a competitor, that's a concrete gap to fix. Platform-specific tactics live in how to appear in ChatGPT.
Asking ChatGPT about your brand by hand is a useful spot check, but it doesn't scale and it's hard to track over time. The systematic version is prompt tracking: monitoring a defined set of these questions across AI engines on a schedule, and recording how your mentions, citations, and sentiment trend.
Elmo is an open-source tool that automates exactly that, running your audit prompts across ChatGPT, Perplexity, Gemini, and Google AI Overviews so you can manage your AI visibility instead of spot-checking. If the audit prompts above surfaced a problem, that's the way to measure and fix it. For the strategy behind it, start with answer engine optimization.
Give context. State the topic, audience, intent, and goal. ChatGPT can't read your mind, and specificity is the single biggest quality lever.
Specify the output. Ask for a format (a table, a numbered list, a brief) and constrain length and tone.
Iterate. Treat the first response as a draft, and refine the prompt or ask follow-ups rather than expecting perfection in one shot.
Verify everything. ChatGPT can be confidently wrong, and it has no live volume data, so fact-check claims and pair it with dedicated SEO tools for metrics.
Keep a brand voice note. Paste a short style guide so drafts come back sounding like you.
Ask for the reasoning. "Explain why you grouped them this way" surfaces bad assumptions before you act on the output.
Prefer one job per prompt. Chained instructions produce output that is mediocre at all of them.
Worth stating plainly, because a lot of prompt libraries imply otherwise.
It has no search volume, keyword difficulty, backlink or ranking data. Any number it gives you for these is invented. It cannot see your Search Console, your analytics, or the current SERP unless you paste them in. It cannot tell you whether a page is ranking, and its account of what is on a live URL is unreliable unless browsing is on and you check what it actually retrieved.
It also cannot tell you whether AI engines mention your brand. It has no visibility into what other models answer, and asking ChatGPT about your own brand tells you about one sampled response from one engine on one day — not a measurement. That job needs a tool that runs prompts across engines repeatedly and records the results.