Keyword Research in the AI Search Era: A Practical Guide

Why AI Search Changes SEO

Traditional searches rely on short, keyword-focused queries. In contrast, AI search enables full conversational questions with layered intent, allowing users to ask in natural language and receive concise, targeted answers. This shift means keyword research must evolve from aiming for short-tail and long-tail phrases to understanding conversational queries, the intent beneath them, and how AI passes context across results. Adoption of AI search is currently growing slowly, so early movers who align content with conversational queries can gain a strategic advantage.

Step 1 — Identify Conversational and Situational Queries

The first step in the AI-era approach is to identify conversational and situational queries rather than solely pursuing traditional long-tail keywords. A practical example is the dog-food scenario: dog owners may search for foods suitable for their dog’s symptoms. The goal is to discover the common denominator that connects customer problems, situations, or needs back to the product, and then break that denominator into logical subcategories to generate realistic queries.

Prompt framework commonly used in this process (described for use with ChatGPT):

I'm promoting a product or service, and We classify this as product. Always start with a please. Identify the common denominator that connects customer problems, situations, or needs back to this product. Then I'll give an example right here. Then break the denominator into logical subcategories categories. Then I give examples. Finally, provide conversational and situational queries that real people would ask online. Group under each subcategory that naturally point toward my product as the solution. This is the format that I want. Feel free to use the search tool. Let's say that we are promoting a subscription box for healthy meals. We can add that into the placeholder and hit Enter or Return. In just a while, we will have a list of conversational queries that are related to our business. 

Examples of applying this approach include potential product lines such as a subscription box for healthy meals or other services (e.g., social media management, budgeting apps, or baby strollers). The method is designed to surface the kinds of conversational and situational queries real people would pose online and to organize them under subcategories that point toward the product as the solution.

Step 2 — Connect Conversational Queries to Traditional Keywords

Large language models (LLMs) do not rely solely on their training data; they also consult web pages related to a query. Most pages on the web are optimized for traditional keywords. Therefore, to ensure AI systems can reference relevant pages when answering conversational queries, marketers should create content around traditional keywords that correspond to the layered intents identified in the conversational queries.

A practical way to explore traditional keywords tied to a conversational query is to use an AI search platform such as Perplexity. For example, asking a query like “I want to stop eating out so much, but I don’t have time to cook. What should I do?” prompts Perplexity to search the web and surface traditional keywords from the results. By clicking on Sources, you can see SEO titles such as:

• Meal Prep for Busy People
• Healthy Meal Prep ideas
• Meal Prep Shortcuts for Busy People
• How to Stop eating out

These traditional keywords are the phrases that AI-powered systems may rely on when constructing answers to the original conversational query. The goal is to capture these traditional keywords and prepare content that targets them directly.

Additionally, the Steps tab in Perplexity can reveal what the AI is searching for, helping to identify the traditional keywords to target. The result is a spreadsheet-like list of traditional keywords corresponding to the conversational query.

Step 3 — Estimating Search Volume for Conversational Queries

There are no tools that accurately predict search volume for conversational queries. Even the presented method is not guaranteed to be precise. The approach described involves extracting traditional keywords from the conversational query and then using a traditional keyword tool to estimate volumes, followed by synthesizing those insights into an estimate for the conversational query.

How this estimation works (illustrative examples from the material):

• Example 1: Conversational query related to meal prep for busy people.
  • Traditional keywords identified include: Meal Prep for Busy People, Healthy Meal Prep ideas, Meal Prep Shortcuts for Busy People, How to Stop eating out.
  • Using Google Keyword Planner with the traditional keyword list (set worldwide), the data yields monthly search volumes for each keyword.
  • Intents are grouped and summed by similarity:
    • First intent: meal prep-related keywords (highest volume 10K–100K).
    • Second intent: phrases about avoiding eating out (volume 100–1K).
  • Estimated volume for the conversational query is the sum of these layered intents, approximately 10.1K to 101K searches per month.
• Example 2: Conversational query about foods for senior dogs with morning stiffness.
  • Traditional keywords identified include: dog food for senior dogs, dog food for arthritis, anti-inflammatory foods for dogs, arthritis diet tips, joint supplements for dogs, etc.
  • Three layered intents are identified, and their highest volumes are combined to estimate a conversational query volume of roughly 21K to 210K monthly searches.

The underlying principle is to account for layered intents—group related traditional keywords that would be triggered by the same conversational concept—and estimate the total volume accordingly. The emphasis is on creating content around the most impactful traditional keywords rather than chasing the most or least competitive terms.

Practical takeaway: build a customer avatar to map out conversational queries, then select the group of traditional keywords with the highest predicted impact to target first. The goal is to inform content strategy with intent-driven data rather than focusing solely on competition.

Step 4 — Content Strategy in the AI Era

In the AI search era, the objective is not merely to rank for traditional keywords in search engines. The broader goal is to ensure your content—across written content, video, and other formats—appears in AI-driven answers by being present across the traditional keywords that genuinely help customers. In other words, focus on helping your audience with relevant content that aligns with their problems, so when AI systems assemble answers, your brand is referenced as a trusted solution.

Decision guidance from the material includes:

• Identify conversational queries first, then determine the traditional keywords AI would surface in response.
• Use traditional keyword research tools to estimate volumes for those keywords and to approximate the volume for the conversational query via layered intents.
• Prioritize content that provides clear, helpful answers to customer needs and demonstrates relevance to buyer intent.
• Aim for broad coverage of relevant traditional keywords across multiple content formats to improve the likelihood of brand mention in AI-driven answers.

Practical Prompts and Examples

A practical example used in the discussion involves a product or service and a demonstration of how prompts can structure the output. The prompts are designed to classify the product, identify the common denominator tying customer needs to the product, break that denominator into subcategories, and generate related conversational and situational queries that point toward the product as the solution.

I'm promoting a product or service, and We classify this as product. Always start with a please. Identify the common denominator that connects customer problems, situations, or needs back to this product. Then I'll give an example right here. Then break the denominator into logical subcategories categories. Then I give examples. Finally, provide conversational and situational queries that real people would ask online. Group under each subcategory that naturally point toward my product as the solution. This is the format that I want. Feel free to use the search tool. Let's say that we are promoting a subscription box for healthy meals. We can add that into the placeholder and hit Enter or Return. In just a while, we will have a list of conversational queries that are related to our business. 

A practical takeaway from this approach is that prompts can help generate structured, testable query sets that map to product advantages. The method scales to various industries, including consumer goods, services, and digital products.

Tools, Limitations, and Next Steps

• No existing tool can perfectly predict the search volume for conversational queries. The method described provides a reasoned estimation based on traditional keywords and layered intents.
• Perplexity can help identify traditional keywords associated with a conversational query by examining the sources and the topics those pages target.
• Google Keyword Planner can be used to estimate traditional keyword volumes; combine this with the layered-intent approach to approximate the conversational query volume.
• The overarching aim is to connect with customers by appearing across traditional keywords that address real problems, thereby increasing the likelihood that AI-driven answers reference your brand.

Examples of Application

Conclusion

The AI search era changes how keyword research should be conducted. Start with identifying conversational and situational queries, then map these queries to traditional keywords that AI systems will surface. Estimate the search volume of the conversational query by aggregating layered intents from traditional keywords, and prioritize content that helps customers and reinforces your brand as a trustworthy solution. This approach emphasizes helpful, intent-driven content across formats, with the aim of ensuring your brand is mentioned in AI-driven answers as consumer needs evolve. The method described here is a practical framework for adapting SEO and content strategy to AI-powered search.