The story that changed everything
Yesterday, I was sitting next to my wife while she spent hours switching between browser tabs, trying to find a children's face wash. She spent several hours reading blogs, checking ingredients, and comparing prices, but still couldn't find anything that met her criteria: chemical-free, organic, natural ingredients, and under 1000 rubles.
If I'd known what she was doing, I would have stepped in sooner. I took out my phone and typed a single query into ChatGPT:
"Are there any chemical-free, organic children's face washes with natural ingredients that cost less than 1000 rubles?"
Instantly, the AI narrowed the choices down to one or two options. One query, one clear answer, no detective work.
Here's why this matters to us in SEO. My wife represents most people searching today—they still search the old-fashioned way, using short, traditional keywords like "chemical-free children's face wash".
But AI search is changing everything. Now people can ask complete, conversational questions with all the context in a single query.
The evolution of search research in the AI era
Studies show that more than 60% of all search queries contain question words (who, what, why, when, where, and how). AI assistants such as Gemini, Copilot, Siri, and Alice prioritize direct answers to questions over keyword matching.
This is where keyword research is evolving. Instead of thinking only in terms of short-tail and long-tail keywords, we need to think about natural-language queries, layers of search intent, and how AI conveys context.
Why traditional tools no longer work
Unlike traditional keyword research tools, which provide search volume based on exact keywords, no tool can provide the same data for conversational queries, because ChatGPT, Perplexity, and other AI platforms don't share information about what people search for on their platforms.
Problems with the traditional approach:
- Volume over intent: Prioritizing high-volume terms, even broad ones, can attract visitors whose needs don't match your content
- Outdated data: Relying on historical keyword data can cause you to miss emerging topics and long conversational queries
- Hidden context: Tools "don't reveal the hidden context of searches," causing marketers to focus on popular terms while ignoring search intent
The secret methodology: SEO Mind42's 3-step approach
Step 1: Find conversational queries through common denominators
Instead of traditional keyword brainstorming, we look for conversational, situational queries.
Example: dog food
Dog owners are very sensitive to their pets' health and may search for food that suits their dog's symptoms. The common denominator is the different types of symptoms in dogs.
ChatGPT prompt:
Какие наиболее распространённые симптомы у собак?
Follow-up prompt:
Предоставь мне разговорные и ситуационные запросы, связанные с кормом для собак и перечисленными симптомами, которые владельцы собак, скорее всего, будут искать в LLM.
SEO Mind42's universal prompt
After a thorough analysis, we developed a universal prompt for any business:
Я продвигаю [продукт или услугу]. Пожалуйста:
1. Определи общий знаменатель, который связывает проблемы клиентов, ситуации или потребности с этим продуктом
2. Раздели знаменатель на логические подкатегории
3. Предоставь разговорные и ситуационные запросы, которые реальные люди задавали бы онлайн, сгруппированные по каждой подкатегории, естественно указывающие на мой продукт как решение
Используй этот формат: [указать желаемый формат]
Examples:
- Healthy meal box: "I want to stop ordering takeout so often, but I don't have time to cook. What should I do?"
- Social media marketing agency: "How can I manage all of my business's social media accounts when I can barely keep up with running the business?"
- Budgeting app: "Why do I always spend more than I plan to, and how can I control it?"
Step 2: Find traditional keywords with Perplexity
Large language models don't rely solely on the data they were trained on. They search for web pages related to the query. Since most pages on the internet are optimized for traditional keywords, LLMs are likely to search for several keywords associated with layered search intent.
How to work with Perplexity:
- Enter a conversational query in Perplexity
- Wait while it searches for answers across different web pages
- Click on "Sources" and review the SEO titles of each page
- Extract traditional keywords from the titles
Example query: "I want to stop ordering takeout so often, but I don't have time to cook. What should I do?"
Traditional keywords found:
- Meal Prep for Busy People
- How to Stop Eating Out So Much
- Healthy Meal Prep Ideas
- Meal Prep Shortcuts for Busy People
Step 3: Estimate search volume using layered intent
Since no tools can accurately predict the search volume of a conversational query, we use a logical approach.
Estimation methodology:
- Collect all traditional keywords in Google Keyword Planner
- Identify the layered intents in the query
- Group keywords by intent
- Take the highest search volume from each group
- Add together the search volumes for different intents
Example calculation:
For the query "I want to stop ordering takeout so often, but I don't have time to cook":
- Intent 1: Meal Prep (10K–100K searches/month)
- Intent 2: Stop Eating Out (100–1K searches/month)
Total: 10.1K–101K searches/month
The logic: The need to search remains the same. People don't wake up one day with their problems gone. Before or after AI, search demand remains the same—the way people search is what changes.
Practical case studies and results
Case study 1: B2B SaaS—Predictive Analytics
Conversational queries:
- "How can we predict when our equipment will break down?"
- "Is there a way to avoid unexpected production downtime?"
- "How can we plan maintenance in advance?"
Traditional keywords:
- Predictive maintenance software
- Equipment failure prediction
- Industrial IoT monitoring
Result: Website traffic increased by 28% in three months, and ChatGPT-4 began citing the content as a predictive maintenance resource
Case study 2: E-commerce—Home Goods
Conversational queries:
- "Which humidifier is best for a 15 sq. m. nursery?"
- "How do I choose an air purifier for someone with allergies?"
Traditional keywords:
- Humidifier for nursery
- Air purifier for allergy sufferers
- Home climate control appliances
Result: 40% growth in citations in AI answers, 25% increase in conversions from AI traffic
New tools for the AI era
Perplexity Deep Research
Deep Research performs dozens of searches, reads hundreds of sources, and analyzes materials autonomously, delivering comprehensive reports in 2–4 minutes. This speeds up answering questions, completing in 2–4 minutes what would take a human expert many hours.
AI tools for keyword research
Top platforms in 2025:
- OtterlyAI: The tool streamlines the process by transforming traditional SEO keywords into conversational, natural-language search queries
- Surfer SEO: Uses machine learning to suggest the ideal keyword density and content structure
- SEMrush with AI: AI-powered, intent-based keyword analysis
AI traffic monitoring
AI tools now account for about 10% of website traffic. Perplexity proved to be the most valuable source, driving both the highest volume of traffic and the best conversion rates.
Platform data:
- Perplexity: Highest volume + best conversion rate
- ChatGPT: Second most impactful
- Gemini: Low traffic, no conversions yet
Adapting your content strategy for AI search
Optimizing for conversational search
Voice search queries are often more conversational and longer than traditional text searches.
Optimization principles:
- Use natural language: Write content that sounds like the way people speak
- Answer questions: Create content that answers common questions on the topic
- Local optimization: If the business has a local presence
- Consider voice-enabled devices: Think about how people might search for content using voice commands
Structuring content for AI
Q&A format: This format reflects how Perplexity presents information, increasing the likelihood that your content will be cited as a source
Recommended structure:
**Вопрос:** Как выбрать корм для собаки с аллергией?
**Ответ:** Для собак с аллергией выбирайте гипоаллергенные корма с ограниченным составом ингредиентов. Избегайте распространённых аллергенов: курицы, говядины, пшеницы и кукурузы.
Measuring performance in the AI era
New metrics to track
30% of all marketers already use AI for keyword research and content clustering.
Key KPIs:
- AI citations: How often you're mentioned in AI answers
- Branded searches: Growth following AI exposure
- AI traffic quality: Traffic from AI summaries converts at higher rates and carries stronger intent
- Cross-platform performance: Tracking traditional search, voice search, and AI chatbots
Measurement tools
- Google Search Console: For traditional SEO
- OtterlyAI: For monitoring AI platforms
- Perplexity Analytics: Citation tracking
- Custom dashboards: Combining data from all sources
Predictions for 2025
Trends that will change everything
- The growth of zero-click searches: By the end of 2025, zero-click searches driven by AI-generated answers will dominate
- The semantic revolution: In 2025, Google's algorithms will prioritize comprehensive topical expertise over targeting individual keywords
- Voice search: 43% of searches are now initiated by voice; voice searches are typically 3.7 times longer than typed queries
Preparing for the future
Strategic steps:
- Build expertise: Become a source of knowledge in your niche
- Create content clusters: Group related topics together
- Optimize for conversational search: Use natural language
- Monitor AI platforms: Track citations and mentions
Practical guide: where to start today
Immediate actions (1–2 weeks)
- Audit your current keywords
- Analyze your top 20 keywords
- Turn them into conversational queries
- Test them in ChatGPT and Perplexity
- Create FAQ sections
- Add Q&A sections to key pages
- Use natural language
- Structure content for featured snippets
Medium-term tasks (1–3 months)
- Content clustering
- Identify 3–5 key topics
- Create interconnected content
- Build internal links
- Optimize for voice search
- Add local optimization
- Create content for featured snippets
- Optimize for "near me" queries
Long-term strategy (3–12 months)
- Develop AI visibility
- Monitor mentions in AI answers
- Create authoritative content
- Build topical expertise
- Measure and optimize
- Implement new metrics
- A/B test AI-optimized content
- Analyze performance regularly
Common mistakes when transitioning to AI search
What NOT to do
- Abandon traditional SEO entirely: AI complements classic SEO but doesn't replace it
- Keyword stuffing in conversational content: Keyword stuffing has been outdated for a long time. Just as with traditional SEO—don't do it
- Ignore user experience: Content should be useful to people
- Create only AI-generated content: Expertise and human understanding are essential
The right approach
- Balance AI and traditional methods
- Focus on quality, not quantity
- Create content for people, optimize for machines
- Test and adapt regularly
Conclusion: the future starts today
Keyword research in the age of AI search involves identifying conversational search queries, finding the different traditional keywords AI will search for, determining the search volume of those keywords, and estimating the search volume of a conversational keyword based on layered search intent.
Key principles of the SEO Mind42 methodology:
- From keywords to questions: Think about intent, not phrases
- Use AI to find AI queries: Use ChatGPT and Perplexity for research
- Measure new metrics: Citations matter more than rankings
- Build expertise clusters: Depth matters more than breadth
Search behavior is evolving, and adapting early to AI-driven search is key to succeeding in a fiercely competitive market.
If you’re not considering this shift yet, you’re already falling behind. AI search currently accounts for only a tiny fraction of the market, but widespread adoption will happen when ordinary users naturally start searching this way.
The time to act is now. Implement this methodology, test, adapt, and prepare for the future of search—it has already arrived.