**James Dooley:**
Hi. Today I’m joined with Sergey, and today’s topic is how to optimise better for all query fan-out searches. SEO has moved away from pure keyword optimisation and towards covering entire topics, especially now that LLMs perform query fan-out in the background. Sergey, how do you optimise better for all query fan-out queries?

**Sergey Lucktinov:**
You need to start with query intent, not keywords. The first step is understanding the problem you are trying to solve for the end user. Once you understand that, you build your content around the problem. For example, if you are writing a review, you need to explain what the product is, how it helps with specific problems, how to use it, and what outcomes it delivers. That way, you naturally cover all the potential fan-out queries that are relevant to the user.

**James Dooley:**
Are there specific attributes or predicates you should focus on, or should you try to cover every possible attribute tied to the main entity or topic?

**Sergey Lucktinov:**
You should focus on the most important ones. This is a classic semantic SEO approach. You analyse what is most relevant to the product or topic. You can do this by looking at SERPs and seeing what is already ranking in Google. There are tools that help with this, but the key challenge is accuracy. You need to identify which entities and attributes actually matter. Start with what makes sense, then refine and update over time as you see what performs best.

**James Dooley:**
Do query fan-out searches differ between platforms like Perplexity, ChatGPT, Gemini and Claude, or are they broadly the same?

**Sergey Lucktinov:**
They vary slightly. The underlying mechanism is similar, but each model implements it differently. If you optimise for meaning and intent rather than specific keywords, your content will generally work across all LLMs on the market.

**James Dooley:**
Is there a way to estimate how many fan-out queries are being run? For example, do more complex searches trigger more fan-out queries?

**Sergey Lucktinov:**
Yes. The more complex or ambiguous the query, the more fan-out queries are generated. This happens because the LLM is trying to understand what the user actually wants. Context also matters. If the model has prior conversation history or user context, it may generate even more fan-out queries to refine the result.

**James Dooley:**
Are you finding that longer form, semantic content performs better in LLMs because it covers more attributes and questions, or does concise content work better?

**Sergey Lucktinov:**
This is one of the hardest problems. I have seen very long articles fail to rank. The key is balance. You need enough depth to prove you understand the topic, but not so much that you drift into unrelated areas. Cover what is necessary and stay focused. Too little coverage is a negative signal, but too much unrelated content is also harmful. Finding that balance is difficult because we do not fully understand the internal logic of LLMs yet. There are many theories, but no definitive rule.

**James Dooley:**
That makes sense. If you are watching this and trying to optimise for query fan-out terms, leave a comment and share what is working for you right now. Things are changing fast, and shared insight helps everyone stay ahead. Thanks very much, Sergey.