The Exact Prompts I Use in ChatGPT, Claude, and Gemini for Keyword Research and Trend Curation

keywords search prompts

Copy-paste prompts for using ChatGPT, Claude, and Gemini together for keyword expansion, search intent analysis, and trend detection.

In the last post, I explained why ChatGPT, Claude, and Gemini each reason differently about search intent — and why using all three together, rather than picking one favorite, gives you a stronger keyword research process. This time, no theory. Just the actual prompts, ready to copy into each tool.

Step 1 — Keyword expansion (ChatGPT)

Use ChatGPT to generate a wide net of related terms and long-tail variations around your seed topic:

“Act as an SEO researcher. Given the seed topic ‘[TOPIC]’, generate 40 related keywords and long-tail variations, grouped by likely search intent (informational, transactional, navigational, commercial investigation). For each group, include 2-3 example searcher questions.”

This gives you a structured starting list instead of a flat pile of keywords — the intent grouping alone saves a huge amount of manual sorting later.

Step 2 — Search intent qualification (Claude)

For any keyword where intent is genuinely unclear, or where the topic is nuanced enough that a wrong call would misdirect your content brief, run it through Claude:

“I’m deciding what type of content to build around the keyword ‘[KEYWORD]’. Walk through the possible search intents behind this query, the type of content that best satisfies each intent, and which intent is most likely dominant based on how the term is typically phrased. Flag any ambiguity rather than forcing a single answer.”

The instruction to “flag ambiguity rather than forcing a single answer” matters — it’s what keeps Claude from just agreeing with whatever framing you gave it.

Step 3 — Trend validation (Gemini)

Finally, check what’s actually moving right now:

“Based on current search trends and recent web activity, is interest in ‘[TOPIC]’ rising, stable, or declining over the past few months? Note any recent news, product launches, or events that might be driving the change, and flag any closely related topics that appear to be rising faster.”

This step is what catches the keywords that looked good on paper but are actually cooling off — or the ones nobody was searching for last quarter that are about to take off.

Putting it together

Run a topic through all three prompts before you commit to a content brief. The combination costs you maybe ten extra minutes per topic — and it consistently catches misdirected briefs before they turn into wasted articles.

Full walkthrough with live examples on the SEO Solver YouTube channel.