November 9, 2025
Keywords tell you what people type. Intent tells you what they need.
Search engines today prioritize content that moves users from questions to actions.
It's not enough anymore to be matchy-matchy: a user searches for running shoes, you talk all about running shoes until they can't bear to scroll anymore. That approach worked when keyword matching was enough. When it was ok to have users pogo-sticking back and forth from links to articles and back.
Today's search engines are answer engines. Their goal is to end the search as quickly as possible. As ahrefs wrote, "Traditional search gives you a list of links to click to get answers. AI search gives you direct answers immediately."
They do this by understanding user intent. And the technology that captures intent is vector search.
Brands optimized for keywords because that's what search engines used to rank content. Now that AI platforms use vectors to understand and select content, brands need to adapt their approach to track and increase visibility in AI search.
This is Part 1 of a two-part series on using vector embeddings to increase brand visibility in AI search results.
In this post, we'll explain how vector search works: what vectors are, how they capture intent that keyword analysis misses, and why they're critical for brand visibility in AI search results.
We’ll be using this free template throughout the article.
In Part 2, we'll give you a step-by-step guide to create this template using your data. You’ll export your Google Search Console data, vectorize it using ChatGPT, and cluster it into topics that show where your brand authority intersects with audience demand.
First, let's talk through the technology to get a solid foundation. Then we'll put it to work.
We spent a year debating why the new AI SEO is different from traditional SEO.
While we were doing that, they merged and became the same thing.
The backbone of Google search has been AI-powered since 2019, when Google announced its BERT model:
"BERT models consider the full context of a word by looking at the words that come before and after it, particularly useful for understanding the intent behind search queries. This represents the biggest leap forward in the past five years, and one of the biggest leaps forward in the history of Search."
While Google quietly became an AI engine under the hood, the interface remained the same: the classic 10 Blue Links.
ChatGPT 3.5 launched in November 2022. Using Google's Transformer model (that's the "T" in both BERT and GPT), it used vectors to understand intent and present answers, reducing or eliminating our need to scroll through search results and click links.
But it couldn't do a basic web search. Users often got the message: "My knowledge cutoff is September 2021. Therefore, I cannot provide you with any information about events or developments that have occurred since that time."
But over the next couple years, Google learned to chat and ChatGPT learned to search. Newcomers arrived (Perplexity, Claude, Grok, Copilot, etc). And a single competitive landscape emerged: AI search.
They're all using the same core technology: transformer-based language models that convert queries and content into vector embeddings, then match them by semantic similarity.
And that changes everything for content strategy. To earn visibility, citations, and relevance in AI search, brands need to understand how their content translates into vectors. And how those vectors align with audience intent.
There's a basic problem with using keywords to match search queries with content: computers don't understand words.
As we learned in The Matrix, computers understand numbers. Computers working with words is a lot like how we learned a foreign language in high school: translating words one by one, divorced from context, cheating off the person sitting next to us.
Vectors translate text into numbers in a way that preserves meaning and intent.
This is the breakthrough the keywords struggled to make. Vector search experts Zilliz explained:
Traditional search systems often failed to fully grasp the user’s intent in the query. They struggled to understand the deeper meaning behind the words as their approaches were based on keyword matching, leading to less accurate results.
Vector search improves the accuracy and relevance of search engine results by understanding the semantic meaning of queries. This allows users to find content that matches their intent, even if the exact keywords aren't present.
Every piece of text (a search query, a blog post, a product description) gets converted into a long list of numbers. Like this: “i need a running shoe that doesn’t hurt my knees after three miles”: [0.0555269484870899, 0.0152359994910813]. Intuitive, right? IRL, this list would have thousands of numbers, not just two. For each piece of text. That’s how it captures the nuance beyond the literal words.
This process of converting text into vectors is called embedding. You'll see "vector embeddings" or "text embeddings" used in the industry. They mean the same thing. We'll use "vector" for consistency throughout this post, but know that vector and embedding are used interchangeably in the wild.
These numbers represent that text's position in mathematical space. Think of it like GPS coordinates for meaning.
Two pieces of text that mean similar things end up with similar numbers, even if they use completely different words. "Running shoe with extra cushion" and "maximum support sneaker" would have similar numbers even though they don’t use any of the same words.
When someone searches, their query becomes a vector. The search engine compares that vector to all the content vectors in its database. It finds the closest matches by measuring the mathematical distance between meanings, instead of trying to look up the meaning of words it doesn’t natively understand.
These vectors turned search into a conversation. And the queries got longer and more complicated. SEMRush research reported "the average AI Mode query was almost twice the length of a traditional query (7.22 words vs. 4.0 words)."
Classic searches were single-turn. AI chats are multi-turn, back-and-forth interactions. SEMRush also noted, "The average number of messages in a ChatGPT conversation is eight—indicating the presence of longer interactions, especially in cases where users ask for further clarification or examples about the same topic."
So now the query is something like, "my knees were hurting yesterday starting around mile 2 of my run. i didn't have time to stretch ahead of time like I normally do. and i also didn't take a rest day. but i have been noticing a little pain lately. should i get a running shoe with more cushion?"
And that's only the first of eight turns, all of which might have valuable context for the search response.
If you try to extract keywords from that query, you get a disconnected list: knee pain, stretches before running, rest day workout, cushion running shoe.
You've lost everything that matters.
Vectors preserve all of it: the relationships between concepts, the specific context, the actual intent.
When ChatGPT, Perplexity, or Google AI Overview answers a query, they're not matching keywords anymore. But they are matching vectors. They're finding the content vector that most closely aligns with the query vector.
Your brand becomes visible when your content semantically matches what users want to achieve with their search. Not the words they’re searching for, but the intent and the action their search leads to.
That's brand visibility in AI search results. Not because you matched keywords. Because your content addressed the layered intent that vectors capture.
Understanding this changes how you approach content strategy. Instead of targeting keyword lists with long-form content, you need to identify the intent patterns where your brand expertise aligns with audience demand.
SEO Rank Media explained content strategy needs to align with the way search engines evaluate it: “Semantic relevance is the new table stake. The brands that engineer content for machine comprehension—vector-friendly passages, structured context, demonstrable topical depth—will surface in chatbots, voice assistants, and whatever interface comes next.”
Let's see what that looks like in practice.
Understanding vectors is one thing. Let’s see vectors work with real search data.
We've created an example using synthetic Google Search Console data for our website, Forecast.ing. You can explore it now to see what the end result looks like. Then in Part 2, we’ll show you how to run this analysis on your own GSC data.
Access the Sample Vector Analysis Template
The template has four tabs:
Click through the tabs. See how queries cluster by meaning. Notice how this moves you out of data collection and analysis, straight into topic selection.
You're now starting at the moment of topic selection, having automated the research and analysis.
Topic selection is the most important decision in your content process.
Ryan Law, Director of Content Marketing at Ahrefs, recently wrote, "90% of success comes from good topic selection."
"90% of success comes from good topic selection."
- Ryan Law, Director of Content Marketing at Ahrefs
Once you pick a topic, a large percentage of your expected ROI is already determined. You can write a brilliant article about a topic no one searches for and create zero value for your brand. You can write about a high-volume topic with no relevance to your business and get impressions and clicks but no conversions.
The goal is to thread the needle: create content at the intersection of what your audience needs and where your brand is differentiated.
By vectorizing and clustering your Google Search Console data, you see which topics already connect your brand to audience demand. Not topics you think matter. Not keywords you guessed at. You see the actual semantic patterns in how people find you and where you already have search authority.
Forecast.ing does this automatically, every week, across brand, competitive, news, and search data.
In today’s example, we used 100 rows of synthetic data. The average Forecast.ing project synthesizes 100,000 data points, creating weekly topics and briefs.
We vectorize it all. Cluster it by theme. And prioritize the topics by relevance, frequency, and momentum, with citations to every source.
You get continuous visibility into where your brand expertise meets audience demand, without the manual work.
You now understand how vector search works and why it powers AI visibility.
You've seen the template showing what vectorized and clustered search data looks like.
Next week, in Part 2, we'll walk through the step-by-step process of analyzing your Google Search Console data. You'll export it, vectorize it, and cluster it into topics that reveal where your brand authority meets audience demand.
You'll finish with data-driven content topics that align with actual semantic patterns from your search data. And a repeatable process you can run regularly to track how your brand visibility evolves in AI search.
---
See you in Part 2.
Ready to see how Forecast.ing automates this analysis with weekly topic briefs?
Grab time for a chat or shoot me a note.
---
What is vector search and how does it work?
Vector search converts text into numerical representations called vectors. These vectors capture semantic meaning instead of just matching words. When you search, the search engine converts your query into a vector and finds content with the closest matching vector by measuring the mathematical distance between meanings.
Why did AI search engines replace keyword matching with vector search?
Keyword matching (lexical search) only compares literal words, failing to capture user intent. Vector search uses transformer models to convert queries and content into numerical vectors (embeddings) that encode meaning and context. This allows the AI to match documents based on conceptual similarity, not just word presence. The introduction of vector search (semantic search) didn't replace keyword matching (lexical search); instead, it demoted its primary role in relevancy ranking and created a hybrid architecture where the two methods complement each other.
What's the difference between AI SEO and traditional SEO?
AI SEO and traditional SEO have merged into a single practice. Google has used AI-powered vector search since 2019 with its BERT model. Traditional SEO focused on ranking individual pages for specific keywords. AI SEO focuses on building comprehensive topical authority that AI platforms can confidently cite when answering queries. The goal has shifted from earning clicks to earning citations.
How do conversational search queries break keyword research?
AI chat queries average 7.22 words compared to 4.0 words for traditional search queries. Conversations span an average of eight back-and-forth turns. Intent and context spread across multiple messages. Keyword extraction from these conversations produces disconnected word lists. These lists miss the relationships and layered intent that vectors naturally capture.
How can I use vector search to improve my brand's visibility in AI search results?
You can start by understanding which search intents your content already serves. In Part 2, we will show you how to export your Google Search Console data, vectorize it using ChatGPT, and cluster it into topics. This process reveals where your brand authority aligns with audience demand. This data-driven approach shows you exactly where to focus your content efforts for maximum visibility in AI search results.
© 2025 Forecast.ing, Inc. All Rights Reserved.
