How do you research what queries people are asking AI? Traditional keyword research doesn't seem to apply

You’re right that traditional keyword research doesn’t directly apply. Here’s the new research framework:

1. Start with Customer Language

Forget keyword tools initially. Go to:

• Support ticket conversations
• Sales call transcripts
• Customer interview recordings
• Onboarding calls

Listen for how customers describe their problems in natural language. These are the prompts they’ll use with AI.

2. Question Mining Sources

3. Direct AI Testing

Create a list of 50 prompts you THINK users ask. Test each across ChatGPT, Perplexity, Claude, Gemini. Document: Who appears? What’s cited? What’s missing?

4. Monitoring Tools

Am I Cited and similar tools track your visibility across AI platforms. They show which queries trigger your brand and where gaps exist.

This is query research, not keyword research. Different mental model.

Reddit is the best free query research tool for AI.

Why Reddit works:

• People ask questions the way they’d ask AI
• Detailed context included
• Real problems, not marketing language
• AI systems cite Reddit heavily

How to mine Reddit for queries:

1. Find subreddits in your category
2. Search for recommendation threads
3. Look for patterns in how questions are asked
4. Note the context people provide

Example search terms:

• “what do you use for [category]”
• “recommend [category]”
• “alternative to [competitor]”
• “best [category] for [use case]”
• “how do you [task you solve]”

What to extract:

• The exact question phrasing
• The context they provide
• Follow-up questions
• What answers get upvoted

We found 80+ unique query patterns from Reddit in one afternoon. These became our AI optimization targets.

Sales perspective: Your sales team is a query research goldmine.

What prospects ask us:

• “How does this compare to [competitor]?”
• “What’s the typical setup time?”
• “Does it integrate with [specific tool]?”
• “What size companies use this?”
• “What’s the pricing for [specific scenario]?”

These are the exact prompts they’ll ask AI.

We started recording these systematically:

1. After every sales call, rep logs top 3 questions
2. Weekly review for patterns
3. Monthly compilation of top questions
4. Feed to content team for optimization

Discovery: 70% of AI-relevant queries never show up in keyword tools. They’re specific, contextual, use case-driven questions that only surface in real conversations.

Your sales team talks to prospects daily. They know the questions. Ask them.

Systematic prompt testing process:

Step 1: Build Initial Prompt List From all sources (customer language, Reddit, sales, etc.), compile 50-100 prompts.

Step 2: Categorize Prompts

• Awareness: “What is [category]?”
• Consideration: “Best [category] for [use case]”
• Decision: “[Product] vs [Competitor]”
• Technical: “How to [specific task with product]”

Step 3: Test Across Platforms For each prompt, test on:

• ChatGPT
• Perplexity
• Claude
• Google AI Overview

Step 4: Document Results

Step 5: Identify Patterns

• Where are we strong?
• Where are gaps?
• What content would fill gaps?

Step 6: Prioritize Focus on high-intent prompts where you’re not appearing but should be.

This gives you an actionable query list, not just keywords with volume.

Tools specifically for AI query research:

Visibility Tracking:

• Am I Cited - Tracks brand mentions across AI platforms
• SE Ranking AI Toolkit - AI visibility monitoring
• Profound - AI share of voice tracking

Question Research:

• AnswerThePublic - Question visualization
• AlsoAsked - Question relationships
• Semrush “Questions” filter - Keyword questions

Conversation Mining:

• Gong/Chorus - Sales call analytics
• Intercom/Zendesk - Support ticket analysis
• Brand24 - Social mention tracking

The gap: No tool gives you “AI search volume” the way Ahrefs gives search volume. This data simply doesn’t exist publicly.

Our workaround:

1. Use traditional tools for question discovery
2. Use AI visibility tools for tracking presence
3. Use manual testing for gap analysis
4. Combine into prioritized prompt list

It’s more work than traditional keyword research but necessary for AI optimization.

How we categorize and prioritize prompts:

Category 1: Brand Queries

• “[Your Brand] review”
• “Is [Your Brand] good?”
• “[Your Brand] pricing” Priority: Must optimize. You need to own your brand in AI.

Category 2: Comparison Queries

• “[Your Brand] vs [Competitor]”
• “Alternatives to [Competitor]”
• “Best [category] tools” Priority: High value, high intent.

Category 3: Problem Queries

• “How do I [solve problem you address]?”
• “Best way to [task]”
• “What tools help with [challenge]?” Priority: Awareness stage, broader reach.

Category 4: Industry Queries

• “What is [concept in your space]?”
• “[Industry] best practices”
• “[Industry] trends 2026” Priority: Thought leadership opportunity.

We aim to appear in 80% of Brand queries, 50% of Comparison, 30% of Problem, 20% of Industry.

Different categories need different content types.

Query research is iterative, not one-time.

Monthly research rhythm:

Week 1: Fresh Discovery

• New support tickets analyzed
• Recent sales call questions
• Reddit/forum monitoring
• Add new prompts to tracking

Week 2: Performance Review

• Which prompts now mention us?
• Which still don’t?
• Any new competitors appearing?
• Which content is getting cited?

Week 3: Gap Prioritization

• Highest value missing prompts
• What content would fill gaps?
• What existing content needs optimization?

Week 4: Action Planning

• Content assignments for next month
• Optimization priorities
• New tracking additions

AI search is evolving constantly. Query research isn’t a project - it’s a process.

Reverse engineer competitor AI visibility:

1. Identify competitor strengths Test prompts across platforms:

• Where do competitors appear that you don’t?
• What content of theirs is being cited?
• What language triggers their mention?

2. Analyze cited content When competitor appears:

• What page is cited?
• What makes it citable?
• How is it structured?

3. Find their gaps Where are THEY not appearing?

• Can you fill that gap?
• What content would capture that prompt?

Example discovery: Competitor appears for “best [category] for enterprise” but not “best [category] for startups.”

We created startup-focused content. Now we dominate startup-related prompts while they dominate enterprise.

Understanding competitor query coverage reveals opportunities.

This completely reframes how I think about research. New playbook:

Data Sources (replacing keyword tools):

1. Customer support conversations
2. Sales call transcripts
3. Reddit/forum discussions
4. AnswerThePublic/AlsoAsked
5. Direct AI platform testing

Research Process:

1. Mine natural language questions from sources
2. Categorize by intent (brand, comparison, problem, industry)
3. Test across AI platforms
4. Document visibility and gaps
5. Prioritize by business value
6. Create content to fill gaps
7. Track and iterate monthly

Tools:

• Am I Cited for visibility tracking
• Traditional tools for question discovery
• CRM/support tools for customer language

Key mindset shift: Not “what keywords have volume” but “what questions are people asking and how can we be the answer.”

Thank you all - this is the new playbook I needed.