Does keyword optimization still matter for AI search, or is it all about topics now?

The relationship between keywords and AI is nuanced. Let me clarify.

Keywords still matter, but differently.

Old model: Keyword → Page optimized for that keyword → Rank for keyword

New model: Topic → Comprehensive content covering topic → AI recognizes authority → Citation for related queries

Where keywords fit:

1. Topic identification Keywords help you understand what topics to cover. “Best CRM software” tells you there’s demand for CRM evaluation content.

2. Intent understanding Keywords reveal user intent. “What is CRM” is informational. “Best CRM for small business” is commercial. Different content needed.

3. Question mapping AI queries are conversational (10-11 words average). Keywords help identify the questions users ask.

What changed:

Exact match optimization is less relevant. AI understands:

• “best project management software”
• “top PM tools”
• “what’s the best tool for managing projects”

…are all the same intent. You don’t need separate pages.

Topical coverage matters more. Instead of 10 pages targeting 10 keywords, one comprehensive page covering the topic thoroughly performs better for AI.

The synthesis:

Use keywords to understand what to cover. Create content for topics, not individual keywords. Build comprehensive authority, not keyword-optimized pages.

Content strategy perspective on the shift.

What I tell writers:

Old brief: “Write a 1,500-word article targeting ‘best project management software’”

New brief: “Write a comprehensive guide to choosing project management software. Cover what PM software is, evaluation criteria, top options, use cases for different team sizes, and common questions. Make it THE resource someone would want for this decision.”

The keyword role:

Keywords inform the brief, not dictate it. I research keywords to understand:

• What questions people ask
• What angles to cover
• What subtopics matter

But the content targets the topic holistically.

Measurement change:

Old: “Did we rank for the target keyword?” New: “Are we cited when people ask about this topic?”

We use Am I Cited to see which queries lead to citations, not just traditional rank tracking.

The practical shift:

From: Page → Keyword To: Topic → Comprehensive content → Multiple related queries covered

How keyword tools are evolving for AI.

Old keyword tools:

• Search volume
• Keyword difficulty
• CPC
• Exact match focus

What’s needed now:

• Question variants
• Related topics
• Conversational queries
• Topic clusters
• User intent classification

Tools adapting:

Better tools now show:

• “People also ask” questions
• Topic relationships
• Question formats
• Long-tail conversational queries

The research process:

1. Seed keyword: “CRM software”
2. Expand to questions:
   • “What is CRM software used for?”
   • “How much does CRM software cost?”
   • “Is CRM software worth it for small business?”
3. Map to topics:
   • CRM basics (definition, purpose)
   • CRM evaluation (cost, ROI, selection)
   • CRM by segment (small business, enterprise)
4. Create content for each topic

The insight:

AI queries are questions. Your content should answer questions comprehensively, not target individual keywords.

Playing devil’s advocate: keywords still work for traditional search.

The nuance:

AI search and Google search coexist. Different optimization approaches might be needed.

For Google (traditional):

• Keywords in titles still matter
• H1 optimization helps
• URL structure with keywords
• Ranking position still drives traffic

For AI (citations):

• Comprehensive topic coverage
• Question-answer structure
• Authority signals
• Less about specific keywords

The hybrid approach:

Optimize titles and URLs for keywords (Google) Create comprehensive content for topics (AI) Structure with questions for extraction (AI)

My take:

Don’t abandon keyword optimization entirely. Google still drives significant traffic, and traditional ranking signals still work.

But layer on topical authority and AI-friendly structure for the AI search piece.

The balance:

Keyword-informed topic strategy, not keyword-only strategy.

How AI actually processes content.

AI doesn’t keyword-match:

When you ask ChatGPT about CRM software, it doesn’t search for pages with “CRM software” in the title 10 times.

It:

1. Understands the semantic meaning of your query
2. Retrieves content that answers the query
3. Evaluates comprehensiveness and authority
4. Synthesizes and cites

What this means:

A page with “CRM software” stuffed everywhere but shallow content won’t be cited.

A page titled “How to Choose the Right Customer Management System” with comprehensive, expert content WILL be cited for “best CRM software” queries.

The authority factor:

AI considers:

• Topic coverage depth
• Expert signals
• Brand recognition
• Consistency across content

Keywords don’t establish authority. Comprehensive expertise does.

Practical implication:

Write for comprehensive topic coverage, not keyword density. Use natural language that includes relevant terms, but focus on answering all aspects of the topic.

Local SEO keywords still matter more.

The local exception:

For local queries, specific keywords still important:

• “[Service] in [City]”
• “[Business type] near me”
• “[Product] [Location]”

AI for local search still relies on explicit location signals.

Local keyword strategy:

Still optimize for:

• City-specific pages
• Service + location combinations
• NAP consistency

Why local is different:

Local intent is specific. Users want businesses in specific locations. AI needs explicit signals to recommend local options.

The approach:

For local businesses:

• Keep location-keyword optimization
• Add comprehensive local content
• Maintain Google Business Profile
• Build local authority signals

Don’t abandon local keyword strategy based on general AI advice.

E-commerce keyword perspective.

Product pages:

Keywords still matter for product pages:

• Product name variants
• Category terms
• Attribute keywords

Google still shows product pages in results. AI recommends products with specific attributes.

The change:

Product descriptions: Less: Keyword-stuffed feature lists More: Comprehensive product information answering buyer questions

Category pages: Less: “Best [keyword] 2026” repetitive More: Comprehensive buying guides with product selection

What works:

1. Product pages: Clear product information with natural keyword usage
2. Category pages: Comprehensive guides covering topic fully
3. Blog content: Answer buyer questions at each journey stage

The measurement:

Track:

• Product ranking (traditional)
• Product citations in AI shopping queries
• Category topic visibility

Both traditional keywords and topical authority matter for e-commerce.

Data on keyword vs topic approach.

Our experiment:

Tested two approaches on similar topics:

Approach A: Keyword-focused

• 5 pages targeting 5 related keywords
• Each page optimized for specific keyword
• Moderate depth per page

Approach B: Topic-focused

• 1 comprehensive page covering topic
• Natural language, question-based structure
• Deep coverage of all aspects

Results after 4 months:

Google rankings:

• Approach A: Ranked for target keywords
• Approach B: Ranked for MORE keywords (long-tail)

AI citations:

• Approach A: 12% citation rate
• Approach B: 34% citation rate

Traffic:

• Approach A: Higher initial (ranked faster)
• Approach B: Higher after 3 months (more queries)

Conversion:

• Similar conversion rates
• Approach B: Higher absolute conversions (more traffic)

Conclusion:

Topic approach wins for both AI and eventually for traditional search too.

Practical workflow for the hybrid approach.

My process:

Week 1: Research

• Keyword research using traditional tools
• Question research (People Also Ask, forums)
• Competitor content analysis

Week 2: Cluster

• Group keywords into topics
• Identify primary topics to cover
• Map questions to topics

Week 3+: Create

• One comprehensive piece per topic
• Cover all keyword-identified questions
• Structure for question-answer extraction
• Natural keyword inclusion (not forced)

Ongoing: Optimize

• Track traditional rankings
• Monitor AI citations (Am I Cited)
• Identify gaps in topic coverage
• Update based on new questions

The template:

For each topic:

• Comprehensive guide format
• H2s as questions
• Direct answers followed by depth
• Expert signals (author, credentials)
• Internal links to related topics

Keywords inform, topics guide, questions structure.

This thread clarified the keyword question perfectly.

My takeaways:

1. Keywords still matter for research - They reveal topics, questions, and intent
2. Optimization targets topics, not keywords - Comprehensive coverage over keyword stuffing
3. Question structure is key - AI queries are questions, content should answer them
4. Both traditional and AI benefit - Comprehensive content ranks for more keywords too
5. Measure both - Track rankings AND AI citations

My new workflow:

1. Keyword research - Gather related keywords
2. Topic clustering - Group into topics
3. Question mapping - Identify questions to answer
4. Comprehensive content - Cover topic fully
5. Question-based structure - H2s as questions
6. Natural keywords - Include terms naturally
7. Track both - Rankings + AI visibility

The mindset shift:

From: “Rank for this keyword” To: “Be THE authority on this topic”

Thanks for the clear explanations!