The Best ChatGPT Prompts for Keyword Research

Keyword research has always been one of the fundamental pillars of any effective SEO strategy. Traditionally, this process required the combined use of tools like Google Keyword Planner, Ahrefs, or SEMrush, along with hours of manual analysis. Today, ChatGPT has introduced a completely new dimension to this workflow: it doesn't replace the quantitative data provided by dedicated tools, but adds a capacity for semantic, contextual, and creative reasoning that no traditional tool can offer. Its strength lies in understanding search intent, generating lexical variants, and the ability to reason around a topic as a domain expert would.

In this article, you'll find a collection of tested and refined prompts to leverage ChatGPT at every phase of keyword research: from initial idea generation to semantic clustering, from user intent analysis to keyword mapping across specific pages. Each prompt is accompanied by an explanation of the prompt engineering principles that make it effective, the expected output, and suggestions for personalizing it to your specific context.

Whether you're working on a content strategy for an e-commerce site, an industry blog, or a corporate website, the prompts presented here have been designed to adapt to real-world scenarios and produce immediately usable outputs.

Ready-to-Use Prompts for Keyword Research

Prompt 1 — Generating Seed Keywords from a Topic

Act as an SEO expert with 10 years of experience in keyword research.
My website covers the topic: [INSERT TOPIC, e.g., "accounting software for SMEs"].
Generate a list of 30 seed keywords divided into the following categories:
1. Informational keywords (user wants to learn something)
2. Commercial keywords (user is evaluating a purchase)
3. Transactional keywords (user is ready to buy)
4. Navigational keywords (user is looking for a specific brand or resource)

For each keyword, indicate in parentheses the probable relative search volume (high/medium/low) based on your knowledge of the industry.

Why it works: This prompt applies the principle of role assignment (assigning an expert role to the model) and structures the output through a consolidated taxonomy — the classification of keywords by search intent. Asking for a relative volume estimate forces the model to reason comparatively rather than generating flat lists.

Expected output: A categorized list of 30 keywords with annotations on intent and a qualitative volume estimate. The output is ideal as a starting point before validating data on tools like Google Search Console or Ahrefs.

Personalization: Replace [INSERT TOPIC] with your project's specific niche. You can add a geographic constraint ("for the Italian market") or audience specification ("aimed at professionals over 40") to further refine the relevance of generated keywords.

Prompt 2 — Semantic Expansion of a Main Keyword

Your task is to semantically expand the following main keyword: "[MAIN KEYWORD]".

Provide:
1. 10 direct synonyms and lexical variants
2. 10 related keywords by topic (not synonyms, but terms that belong to the same semantic universe)
3. 10 long-tail keywords based on frequently asked user questions (format: "how", "what", "why", "which")
4. 5 LSI keywords (Latent Semantic Indexing) that Google typically associates with this term
5. 3 probable "also asked" keywords based on Google's PAA (People Also Ask)

Format the output in a Markdown table with columns: Keyword | Type | Estimated Intent

Why it works: The request for tabular output in Markdown is an output formatting control technique that dramatically improves the readability and reusability of the result. Breaking the request into five distinct categories avoids the model's tendency to generate homogeneous and undifferentiated lists.

Expected output: A Markdown table with approximately 38 keywords organized by type and intent. This output can be directly imported into Google Sheets via copy-paste or parsing.

Personalization: If you work in a highly technical field, add the instruction: "Also include industry-specific terminology and professional jargon used by experts in the field." This unlocks a level of lexical depth that standard tools often overlook.

Prompt 3 — Analyzing Search Intent for a List of Keywords

Below is a list of keywords. For each one, analyze and classify:

KEYWORDS:
- [keyword 1]
- [keyword 2]
- [keyword 3]
- [keyword 4]
- [keyword 5]

For each keyword return:
1. **Primary intent**: Informational / Commercial / Transactional / Navigational
2. **Micro-intent**: one-sentence description of what the user is exactly looking for
3. **Recommended content type**: (e.g., how-to article, product page, comparison, landing page, FAQ)
4. **Funnel stage**: TOFU / MOFU / BOFU
5. **Conversion probability**: High / Medium / Low

Be precise and consider the Italian digital market context.

Why it works: This prompt transforms ChatGPT into an intent analyst — one of the most complex and valuable cognitive tasks in keyword research. The request for "micro-intent" goes beyond simple classification and forces the model to reason about what the user wants to achieve, not just which category the search belongs to. The explicit reference to the "Italian market" reduces the drift toward generalizations based on English-speaking markets.

Expected output: A detailed profile for each keyword with 5 attributes. Particularly useful for deciding what type of page to create and how to position content in the funnel.

Personalization: You can paste up to 20-25 keywords per session without degrading output quality. For keywords with dual possible interpretations, add: "If a keyword is ambiguous, indicate both possible interpretations."

Prompt 4 — Semantic Clustering of Keywords

You are an expert in information architecture and technical SEO.
I'm providing you with a list of 20 keywords. Your task is to group them into semantic clusters,
where each cluster represents a potential page or website section.

KEYWORDS TO CLUSTER:
[paste your 20 keywords here, one per line]

Instructions:
- Create clusters based on intent similarity, NOT just lexical similarity
- Each cluster must have: cluster name, main keyword (head term), secondary keywords, support keywords (LSI/related)
- Indicate whether the cluster is suited to a single page or a content hub with satellite pages
- Suggest an optimized H1 title for each cluster

Output format: use Markdown headings for each cluster.

Why it works: Clustering by intent (not superficial lexical similarity) is the critical distinction that makes this prompt superior to simple keyword grouping. Including the distinction between "single page" and "content hub" adds an information architecture strategy layer that transforms keyword research into a concrete editorial plan.

Expected output: 4-7 well-defined semantic clusters, each with a head term, secondary keywords, recommended page type, and a ready-to-use H1 title. This output is the foundation for a professional content map.

Personalization: Add context about your site: "The website is an e-commerce home products store with an estimated domain authority of 35." This information guides ChatGPT toward realistic clusters achievable for your domain's authority level.

Prompt 5 — Keyword Gap Analysis vs. Competitors

Act as an SEO consultant performing competitive gap analysis.

My website focuses on: [DESCRIBE YOUR INDUSTRY AND PRODUCT/SERVICE]
My main competitors are: [COMPETITOR 1], [COMPETITOR 2], [COMPETITOR 3]

Based on your knowledge of these brands and the industry:
1. Identify 15 thematic areas (topic clusters) that competitors probably cover and I might not have addressed yet
2. For each thematic area suggest 3 specific keywords I could compete for
3. Highlight which areas likely have less editorial competition (opportunities for faster ranking)
4. Suggest 5 "underserved" content angles — topics the market treats minimally but have latent demand

Note: Always specify when you're making an estimate based on general knowledge vs. verifiable analysis.

Why it works: This prompt applies the principle of constraint transparency — by explicitly asking the model to distinguish between verified knowledge and estimates, you increase output reliability and avoid the phenomenon of "confident hallucinations." The request for "underserved angles" is an open prompt that leverages ChatGPT's creative capacity to identify non-obvious opportunities.

Expected output: A strategic document with thematic areas, associated keywords, and content opportunities. This output is ideal as a brief for a content strategist or as input for an editorial planning session.

Personalization: If you know specific competitor URLs, you can cite them. ChatGPT can reason about known brands and sites with reasonable accuracy, but remember its knowledge has a cutoff date: for current data, always integrate with tools like Ahrefs Site Explorer or SEMrush Gap Analysis.

Prompt 6 — Generating Keywords for Featured Snippets and Position Zero

I want to optimize content to win Google Featured Snippets in the industry: [INDUSTRY].

Generate 20 high-potential featured snippet queries following these criteria:
- Direct questions that Google tends to answer with "paragraph" type snippets (who, what, why, how)
- Queries that produce "list" type snippets (step-by-step, element listings)
- Queries that produce "table" type snippets (comparisons, structured data)

For each query indicate:
1. The most probable featured snippet type (paragraph / list / table)
2. The ideal length of the optimized response (in words)
3. An example of how to structure the response to maximize chances of ranking in position zero

Focus on informational intent queries with medium-high volume in the Italian market.

Why it works: This prompt combines keyword generation with optimization strategy. Specifying the three snippet formats (paragraph, list, table) structures the output to cover all patterns Google uses, making the result immediately actionable for an SEO copywriter.

Expected output: A list of 20 queries with snippet type, recommended response length, and an example structure. This prompt is particularly useful for those wanting to scale the production of position-zero-optimized content.

Personalization: Replace "Italian market" with a specific region if you operate in a local market (e.g., "Lombardy market" or "Italian-speaking users in Switzerland") to get queries with more precise geographic nuances.

Prompt 7 — Keyword Research for E-commerce: Transactional Intent and Product Pages

You are an SEO specialist specialized in e-commerce. I'm optimizing the product and category pages of my online store that sells: [PRODUCT CATEGORY].

Conduct keyword research focused on transactional intent:

1. **Keywords for category pages** (medium-high volume, commercial/transactional intent)
   - 10 head term keywords for main categories
   - 10 keywords for subcategories (more specific, long-tail)

2. **Keywords for product pages** (long-tail, high specificity)
   - 10 keyword patterns based on product attributes (color, size, material, brand)
   - 5 keywords based on comparison (e.g., "X vs Y", "best X for [use case]")

3. **Negative keywords to exclude** (informational intent queries that would waste PPC budget)

4. **High-conversion modifiers** to add to base keywords
   (e.g., "affordable", "authentic", "fast shipping", "deal", etc.)

Include a note on which keywords are better suited for Google Ads campaigns vs. organic ranking.

Why it works: The separation between keywords for category pages and product pages reflects real e-commerce architecture and produces output directly mappable to your site structure. The "negative keywords" section is often overlooked but demonstrates SEO depth and provides immediate value even for those managing paid campaigns.

Expected output: A keyword plan structured on three levels (category, subcategory, product) with modifiers and negative keywords. The output is usable for both organic optimization and as a foundation for a Google Shopping campaign structure.

Personalization: Add your average product price range and target demographics. These variables significantly influence intent modifiers (e.g., "premium" vs. "affordable", "professional" vs. "for beginners").

Prompt Engineering Techniques for Keyword Research

1.