The Volatility of AI Search: Why Small Changes in Queries Matter.

A tip from what i’m seeing on AI-driven search results:

Small tweaks in your queries matter a LOT.

AI search results change more often and are less stable compared to Google's regular search results.

It makes sense since it’s just the beginning of this era.

But how much more volatile are they?

Here’s what I found through some research:

— If you keep your query exactly the same, the results will generally stay consistent.

— But the moment you tweak it—like adding a word or shifting phrasing—everything can change.

Take this example: One word can change a query

A great case study from Tim Soulo and Olga Andrienko

A query change of just one word:

‘’What are THE best tools of SEO to look for professionals?’’ VS ‘’What are SOME good tools to look for SEO for professionals?’’

That one small difference—“THE” vs. “SOME”—is a game-changer.

• Question with “THE”

• Question with “SOME”

The Results: Different answers.

So adding anything to the prompt changes it’s results.

It might seem insignificant, but for LLMs, it's a big deal because of how they analyze relationships between words during crawling.

? LLMs focus heavily on the relationships between words.

Here’s another example:

SearchGPT's map results are inconsistent, often dominated by a single source or failing to trigger maps.

Asking explicitly for a map redirects users to Google Maps, highlighting its strengths in local search.

While it excels in time-sensitive queries like holiday shopping, it struggles with local searches, sometimes misclassifying results or relying on paid directories. Exposing gaps in AI-driven search.

? Location where query was searched? No where near the results.

YMYL queries like "How to lower cholesterol" highlight ChatGPT’s reliance on trusted sources like WebMD, while its ecommerce results excel in fresh content but lack transactional depth compared to traditional engines.

???It’s fascinating—and it also shows just how nuanced (and inconsistent) these AI-driven processes can be.

This volatility reflects the rapid evolution of AI search, with major milestones like:

• Google AI Overviews.
• Bing’s generative search.
• ChatGPT Search all launching in 2024.

Bing’s importance in AI search

? Bing’s importance in AI search is growing as it powers ChatGPT Search. While indexing in Bing is necessary, ranking high isn’t.

SearchGPT often pulls pages from a broader database, including those outside Bing’s top 100. ChatGPT now drives 31% of LLM traffic year-to-date on some sites.

Right now, we’re in the early days. Eventually, optimizing for AI-driven search might become more straightforward.

But for now, AI-driven search feels as volatile as early Google, when you could rank #1 just by stacking PBN links.

As these systems mature, SEO must shift from ranking to building visibility and relevance across ecosystems.

Thanks for reading!

-Dmitry. ??

The original post is on my Linkedin profile

I’d appreciate it if you would share it with anyone who would benefit from it!

? https://www.dhirubhai.net/posts/dmitrydragilev_heres-a-quick-insider-tip-on-ai-driven-activity-7269708654264606722-qM0u?utm_source=share&utm_medium=member_desktop

Need help finding easy to rank keywords?

Head over to TopicRanker to find easy-to-rank keywords based on competitor weak spots in SERPs.

TopicRanker