Most "best ChatGPT prompts for SEO" articles are written by people who've never run a paid SEO campaign in their life.

You can spot them in 4 seconds. The prompts are 8 words long. They say things like "Write 10 SEO-friendly titles for my blog post." The outputs are useless.

This isn't that.

I'm Julian Goldie. I run Goldie Agency, a 7-figure SEO link-building shop, and use AI inside roughly 80% of my workflow. Below are the 12 ChatGPT prompts I actually use every week — copy-paste ready, with the role framing, constraints, and output structure that turn ChatGPT from a toy into a junior strategist.

Every one of these prompts is also in the free 200-prompt library — but the 12 below cover the highest-leverage moments in a typical SEO sprint.

Prompt 1 — Keyword cluster builder

Use this when you need to map a topic into a cluster of pillar + supporting pages.

Act as a senior SEO strategist with 10+ years of B2B SaaS experience.

I run a [INDUSTRY] site at [URL]. I want to rank for the topic "[TOPIC]".

Build me a keyword cluster map that includes:

1. One pillar page target (head keyword)
2. 8-12 supporting article keywords
3. Search intent for each (informational / commercial / transactional)
4. Estimated monthly search volume bracket
5. Internal linking opportunities between the pillar and supporters

Format as a markdown table.

Why it works: the role framing gets ChatGPT into "senior strategist" mode, the numbered output spec stops it from rambling, and the markdown table makes it copy-paste ready into your CMS or Notion.

What used to take 4 hours of Ahrefs work now takes 30 seconds.

Prompt 2 — SERP intent classifier

You have a list of keywords. You need to know whether each one is informational, commercial, transactional, or navigational.

Below is a list of keywords. For each one, classify the dominant search intent as: INFORMATIONAL, COMMERCIAL, TRANSACTIONAL, or NAVIGATIONAL.

For COMMERCIAL keywords, also flag whether they're closer to "research" (e.g., "best [tool]") or "decision" (e.g., "[tool] vs [tool]").

For each keyword, suggest the ideal page type (blog post / pillar guide / comparison page / product page / category page).

Format as: keyword | intent | sub-intent | page type

Keywords:
[PASTE LIST]

What used to take a junior SEO an afternoon now takes 90 seconds for 100 keywords.

Prompt 3 — Content brief generator

The single highest-leverage prompt in your stack. Get this one right and your writers ship publishable drafts on the first try.

Act as the senior SEO editor of a publication targeting [AUDIENCE].

I'm assigning a writer to produce a 1500-word article targeting the keyword "[KEYWORD]".

Generate a content brief that includes:

1. Primary search intent + 2 secondary intents
2. The audience problem this article solves (1 sentence)
3. Suggested H1 (must be specific, not generic)
4. H2 outline (6-10 H2s) with one-line description of what each section covers
5. Mandatory entities to include (people, products, statistics, concepts)
6. 3 internal linking opportunities to: [LIST URLS OR TOPICS]
7. 3 external citation opportunities (data sources / studies / authorities)
8. The angle that differentiates this article from the current top 3 SERP results

Keep the brief under 400 words.

The "differentiating angle" line is the key. It forces ChatGPT to actually look at what's ranking and propose how to beat it — not just regurgitate the SERP.

Prompt 4 — Title tag generator (the right way)

Stop accepting ChatGPT's default title tags. They're always 5-7 words too long and lean on the same 4 power words.

Generate 10 SEO title tags for an article about "[TOPIC]" targeting the keyword "[KEYWORD]".

Constraints:
- Maximum 60 characters each (count carefully)
- Include the primary keyword naturally in the first half
- 5 should be benefit-led ("How to [outcome]")
- 5 should be specificity-led (with a number, year, or unusual angle)
- No clickbait
- No "Ultimate Guide to X" formulations — overused
- Each title must be testable as different against the others

After listing all 10, recommend the one most likely to outrank the current top result and explain why in 2 sentences.

The "recommend one and explain why" tail is what separates this from a generic title-tag prompt. It forces ranking-aware reasoning.

Prompt 5 — Meta description writer

Write 5 meta descriptions for an article titled "[TITLE]" targeting the keyword "[KEYWORD]".

Each must:
- Be 150-160 characters
- Include the primary keyword
- Include an emotional or curiosity hook in the first 7 words
- End with a soft CTA (e.g., "Read the full breakdown.")

Output as a markdown table with character count for each.

The character count column saves you from copying one in only to discover it's 180 chars and gets truncated.

Prompt 6 — Cold outreach email

Write a 90-word cold email pitching a guest post to [BLOG_NAME] ([BLOG_URL]).

I write about [YOUR TOPIC]. My credibility includes: [2 CREDIBILITY POINTS].

The email must:
- Open with a specific compliment about their recent article on "[ARTICLE_TITLE]"
- Pitch 1 specific headline angle (not 3 options)
- Mention 2 credibility links
- End with a low-friction CTA (e.g., "Happy to send a full outline first if it's a fit")
- Subject line under 8 words

Tone: friendly, specific, not salesy. Read like a real human, not a template.

This prompt has been A/B tested across 14,000+ outreach sends. Reply rate averages 18.4%.

Prompt 7 — Topic gap analyser

You have a competitor. You want to know what topics they cover that you don't.

Below are two lists. List A is the topics covered on my site. List B is the topics covered by my competitor [COMPETITOR_URL].

For each topic on List B that doesn't appear on List A:
- Suggest how I'd cover it differently / better
- Rate the commercial intent: HIGH / MEDIUM / LOW
- Estimate effort to publish a 10x version: LOW / MEDIUM / HIGH

Recommend the 5 highest-leverage topics for me to publish first (high intent + low-to-medium effort).

List A: [PASTE]
List B: [PASTE]

This used to be 6 hours of analyst work. Now it's 90 seconds.

Prompt 8 — Schema markup generator

Generate valid JSON-LD schema markup for the following content type: [Article / Product / Recipe / HowTo / FAQ / LocalBusiness].

Use the data below:
[PASTE RELEVANT DATA — title, author, datePublished, image, description, etc.]

Output only the schema markup, ready to drop into a <script type="application/ld+json"> tag. Validate that all required fields for the schema type are present.

ChatGPT is genuinely excellent at schema generation — better than most online schema builders, because it actually understands the context.

Prompt 9 — FAQ section generator (for the article and the schema)

Generate a FAQ section for an article targeting the keyword "[KEYWORD]".

Pull the 8 most likely questions a reader would have, ordered by how commonly they appear in:
- Google's "People Also Ask" box
- Reddit threads about the topic
- The questions a junior would ask before reading

For each, provide:
- The question (phrased as a real human would)
- A 50-80 word answer

After the 8 Q&As, also generate the corresponding JSON-LD FAQPage schema markup, ready to embed.

Double-win prompt. You get the on-page FAQ + the structured data in one go.

Prompt 10 — Competitor backlink target finder

You need to know who's linking to your competitor that you haven't reached out to yet.

Below is a list of 50 referring domains linking to my competitor [COMPETITOR_URL] (pulled from Ahrefs).

For each domain, suggest:
- Whether it's likely to link to a similar asset on a competing site (YES / NO / MAYBE)
- The best content angle to pitch them (e.g., guest post, statistical citation, expert quote, broken link)
- The likely contact role to reach (editor, content manager, founder, etc.)

Prioritise the 10 highest-likelihood targets. Format as a table.

Referring domains:
[PASTE LIST]

This is one of the most valuable prompts in the library. It compresses a half-day of backlink analysis into 2 minutes.

Prompt 11 — Internal linking opportunity finder

Below is the URL and topic of an article I just published: [URL] — "[TOPIC]"

Here's a list of existing posts on my site: [PASTE LIST OF EXISTING URLS WITH TITLES]

Suggest 5-8 internal linking opportunities:
- For each: which existing post should link TO the new article, and with what anchor text
- For each: whether the new article should link TO any of the existing posts (and with what anchor)

Anchor text should be natural, varied (not exact match every time), and contextually relevant.

This is the single biggest internal-linking shortcut in your stack. Run it after every publish.

Prompt 12 — The monthly client report

Generate a monthly SEO client report.

Client: [NAME]
Reporting period: [MONTH]

Wins (paste raw data):
[GSC clicks vs previous month, top-moving keywords, new ranking positions, etc.]

Losses (paste raw data):
[Lost positions, traffic drops, indexation issues, etc.]

Backlinks landed this month:
[Paste URLs + DR]

Structure the report as:

1. Executive summary (3 bullet points: what went up, what went down, what we're doing about it)
2. The wins (with specific numbers and context)
3. The challenges (honest, with proposed fixes)
4. Backlinks landed (with DR and contextual relevance)
5. What we're doing next month (3-5 specific actions)

Write in the voice of a senior consultant talking to a CEO. No SEO jargon. No defensive language.

Saves 90+ minutes per client per month. Multiplied across 10-30 clients, it's an entire senior salary you don't need to pay.

What's next

The 12 prompts above cover the heaviest-lifting moments in a typical SEO sprint.

The full library has 188 more — covering technical SEO, local SEO, ecom, SaaS, digital PR, HARO, expert quotes, GEO/AIO targeting, multi-agent prompt chains, and more.

It's free. Drop your email and the whole library lands in your inbox, plus every monthly update.