For years, ranking on Google meant optimizing your website and content based on a comprehensive list of SEO ranking factors.
But now, Large Language Models (LLMs) like Gemini, Perplexity, and GPT are making traditional factors lose their strength, fundamentally changing how search engines process and deliver information.
For agencies like yours, this means different opportunities for traffic and rankings. Which begs the question:
How do you help clients stay visible when AI is filtering and repackaging information before users even see search results?
In this guide, we’ll go over what’s changing with AI in SEO, why it matters, and how your agency can stay ahead in this new era of search.
1. The Evolution of Search: From Keywords to AI-Powered Understanding
How Search Engines Used to Work (and Why It Was Relatively Easier Back Then)
Up until Google’s algorithm updates became more sophisticated in the early 2010s, it relied on a fairly straightforward system: link-based algorithms and keyword matching.
Essentially, the idea was to match the keywords in a query to the words on a webpage, and then use backlinks to determine authority. This approach had the major flaw of prioritizing quantity over quality. Low-value pages could start to rank well simply by “gaming the system” through what we know today as black-hat SEO until Google cracked down with updates like Panda and Penguin, which penalized manipulative tactics (like suspicious link-building techniques) and hit many websites hard.
With Google’s PageRank algorithm, the more backlinks you had – especially from high-authority sites – the better your page would rank. It was a step up from spammy, black-hat sites, but barely.
It was no surprise that, by the time AI and LLMs came around, the searchers were ready for something new.
The AI-Driven Shift in Search Engines
Fast-forwarding to May 2023, Google introduced Search Generative Experience (SGE), its first AI-based search engine. However, when Google launched this feature to the public at Google I/O 2024, it gave it a simpler name: AI Overview.
“With new generative AI capabilities in Search, we’re now taking more of the work out of searching, so you’ll be able to understand a topic faster, uncover new viewpoints and insights, and get things done more easily,” explained Elizabeth Reid, Vice President & GM, Search.
Speed. Ease. Information. Mark those words.
This shift has made Google less of a search engine and more of an answer engine, similar to how people now type in queries, ask follow-up questions, and have entire conversations with LLMs like ChatGPT. Instead of leaving it up to you to choose the best results, Google is now doing that work for you.
But which LLMs power these answers? And how do they choose the results?
AI models generate answers based on a much different set of factors, including user behavior, query context, geographical location, and even the AI’s understanding of related concepts – not just ranking pages in a fixed order.
So, instead of the same result showing up for everyone, the answers can change depending on different contexts.
For businesses that rely on SEO for traffic, this shift is huge. As AI overviews become more common, the reliance on traditional blue links in search will shrink, making it harder to depend on Google rankings alone for visibility.
Why This Matters for Agencies
- For agencies, especially small ones, this means keywords and backlinks will now be taking a backseat. Not only those two, but traditional SEO ranking methods aren’t necessarily applicable to Generative Engines.
- AI-driven search engines like Google and Bing are prioritizing content that’s authoritative, well-structured, and actually useful. Agencies need to focus on helping their clients become trusted sources that AI models pull from, rather than being result #1 in the SERPs.
- Agencies should focus on creating multimodal content (including images, videos, and other media) that demonstrates expertise and answers real questions, rather than just going after high-volume keywords or trying to squeeze as many semantically-related keywords in an article.
2. How LLMs Are Changing Google & Search Behavior
How LLMs Process and Retrieve Information
Simply put:
- LLMs crawl and process entire web pages, extracting key insights rather than indexing based on keywords alone.
- They use Natural Language Processing (NLP) to do that, AKA the technology that allows AI to understand and interpret human language through meaning, context, and intent.
- They prioritize content quality, expertise, and coherence over simple ranking factors like backlink count.
Now, you might be thinking, “Doesn’t modern SEO already put quality and relevance over everything else?”
Some shared principles exist, which is why SEO is absolutely still relevant. Both systems aim to give users the most relevant, high-quality content based on the user’s query. That said, SEO still depends on ranking well through traditional signals, while LLMs rely on contextual understanding and content synthesis to give users immediate answers.
As SEO specialist and consultant Rich Sanger explains in his blog:
- You enter a search query. This could be a question or simply a request for information.
- Based on the type of query (whether it’s informational, creative, or related to images) Google uses its system to pick the best LLM for the job. This could involve models like Google's PaLM or LaMDA, which are designed to understand the nuances of your request.
- The LLM looks at the top results that appear for your query and checks for keywords and content relevance. It also considers user context, like location or search history, to fine-tune its response.
- Beyond just the immediate search results, Google taps into related searches and information that other users found useful under similar circumstances. It might also include other forms of content like articles, images, and videos.
- With all this information, the LLM combines insights from various sources to create a response that directly addresses your query. It’s not simply pulling from a single page but synthesizing the best information across multiple sources and rewording it in a helpful way.
- If queries are repeated, the AI adjusts the way it responds based on the context of each submission. It learns from user interactions to continually refine its answers.
- The AI continually tracks how users interact with the search results. For example, if a certain response gets constantly ignored in favor of others, the model adjusts the way it ranks and generates responses in the future to align with user preferences.
- Over time (and in real time), Google’s AI keeps improving based on new interactions and information.
Sanger goes on to explain, “For example, if you search for ‘apple,’ the LLM(s) evaluates the context of your query based on what you searched for recently. If you’ve been searching for tech news or Apple products, the LLM interprets your query as related to Apple Inc. On the other hand, if your recent searches include recipes or health topics, it’s more likely to understand that you’re referring to the fruit.”
In short, rankings aren’t steady anymore. You might search for the same phrase today and get different answers just minutes later, depending on how you phrase it or your past search history.
What This Means for Search Engine Results Pages (SERPs)
The rise of AI-generated summaries like Google’s AI Overviews has a similar pull to featured snippets or “position zero”, which recently dominated organic rankings. This shift is already changing user behavior:
- AI-generated answers are replacing top-ranking organic results for many queries. They take up nearly half the screen space on mobile and desktop.
- Fewer users are clicking traditional search results since AI summaries provide direct answers.
- Informational queries are on the rise.
- SERP layouts are changing, with more space dedicated to AI-generated content and fewer visible blue links.
Take a look at the following AI Overview:
How an AI Overview pushes SERP results down the page
Notice how the concise AI-driven summary pushes organic results further down the page and offers relevant sources on the sidebar links. While the AI doesn’t always attribute specific pieces of information to individual sources, it pulls from what it determines to be the most useful and trustworthy sources to create a well-rounded answer.
In most cases, you wouldn’t need to scroll down to get everything you need from a search. And today, that doesn’t only apply to Google, but to previously “neglected” search engines such as Bing.
🚨 Why You Should Start Caring about Your Bing Rankings
Bing rankings used to be an afterthought for…well, most of us. Google’s dominance made it easy to ignore Microsoft’s search engine, but that’s no longer an option because ChatGPT uses Bing Search as a key input for generating AI-driven responses.
That means:
- If your content ranks well on Bing, it’s more likely to be referenced in AI-generated results (including ChatGPT).
- Agencies need to start tracking and optimizing for Bing, just as they do for Google.
- Bing’s ranking signals, like structured data, page speed, and engagement, now also have a direct impact on AI visibility.
According to Microsoft, “Bing generative search is just the first step in upcoming improvements to define the future of search. We’re continuing to roll this experience out slowly to ensure we deliver a quality experience before making this broadly available.”
Just like AI Overviews, Bing’s AI is experimental – but worth optimizing for ASAP. So, if you haven’t already, sign up for Bing Webmaster Tools and claim your Bing Places listing.
Key Takeaways for Agencies
- Visibility in AI search is about being a trusted source rather than ranking for specific keywords. Keyword stuffing no longer works!
- Create content that’s comprehensive, well-structured, and authoritative to be recognized as a reliable source. Do this by:
- Focusing on depth over length. Let go of the idea that super-long-form content is automatically valuable. What matters is whether it provides clear, thorough, and well-organized insights that fully answer a user’s intent.
-
Earning citations and mentions. LLMs like ChatGPT, are trained on data from trusted sources, including Vox, News Corp, and Reddit. This means content from reputable sources is more likely to be referenced by AI models when generating responses. For agencies, this is proof of how important guest posts, collaborations, and media coverage are.
If you want a deeper dive into content quality and authority for AI search, here’s a thorough explanation that includes a concept you’re familiar with: E-E-A-T.
3. Content Quality and Authority: The New SEO Foundations
Google's E-E-A-T and AI Content Prioritization
Google’s system also evaluates the quality and trustworthiness of the search result documents (SRDs) it gets data from – such as credible websites, industry experts, and authoritative brands and institutions. This is where AI summaries overlap with SEO, as E-E-A-T (Experience, Expertise, Authority, and Trust) now plays a central role in how AI models select content.
For example, an AI model generating a response on financial planning won’t pull from just any finance blog.
Instead, it’ll prioritize insights from certified financial experts, government resources, and well-established institutions, similar to why Google ranks Your Money or Your Life (YMYL) websites higher.
The same applies to health-related topics: a medical AI-generated response is far more likely to pull data from the Mayo Clinic or the CDC than an unverified wellness blog.
For small agencies, this means thin, keyword-stuffed, or formulaic content will be ignored. There’s no more getting away with surface-level optimization that lacks expertise or content that feels engineered for rankings.
Beyond clear authority and expertise, the helpful and widely-surfaced results also have two things in common: they’re structured and context-driven.
Structured, Contextual Content vs. Keyword Optimization
Just to recap: AI-driven search engines no longer prioritize keyword-heavy content. They favor well-structured, deeply informative pages published by known entities. Now, here’s what else AI models focus on when selecting the “top earners”:
Entities, Relationships, and Context
Sticking to the apple example, Imagine a blog post about “Apple.” In traditional SEO, Google would need extra signals (like surrounding text) to determine whether the page is about Apple the tech company or apple the fruit.
LLMs go a step further – they analyze relationships within the content. If the page mentions iPhones, Steve Jobs, and macOS, the AI understands that “Apple” refers to the company, not the fruit.
Structured Data
Structured data or schema markup helps AI models process content more efficiently through a clear information hierarchy.
Schema markup uses machine-readable signals (structured data tags) that help AI understand what a page is about, making it easier for the model to identify and highlight key elements in responses.
In the words of Nadav Nesher, Applied NLP Researcher from GigaSpaces, “Structured data also helps advance decision-making capabilities within GenAI systems. Clear and organized data helps systems generate accurate insights and predictions. Informed decisions are made based on reliable information, limiting the risk of mistakes that result from ambiguity.”
How Agencies Should Adapt
The focus should shift from tactical keyword placement to building content authority and credibility:
- Encourage clients to build authority in their niche. Instead of publishing surface-level content on many topics, help them focus on in-depth, expert-driven content within their field.
- Use schema markup to structure content for AI models. This makes it easier for AI-driven search to extract and present key answers, whether they’re text or media-based.
- Be the strategic partner who shifts your clients’ mindset from “ranking” to “credibility”. Help clients establish themselves as go-to authorities through consistent, high-quality content. This means leveraging expert quotes, citing reliable sources, and showcasing real-world experience – not just publishing content for the sake of SEO.
4. The Decline of Traditional Link-Building & the Rise of Semantic Search
AI-driven search engines are moving beyond simple backlink-based metrics and shifting toward a deeper evaluation of content itself.
Instead of just counting links, AI models assess:
- Content depth and coherence – Is the content well-structured, informative, and engaging, or is it just a substantially thin article designed to attract backlinks?
- Entity recognition and topical authority – Does the content fit within a broader, well-connected topic, or is it an isolated page with no real expertise?
- Engagement signals and social proof – Are users interacting with and sharing the content, or is it getting ignored?
AI models can estimate engagement through indirect signals, like how often links are clicked, content sharing patterns, and even time spent on a page based on interaction history. They also analyze social media shares, comments, and more.
Backlinks still matter, but they’re no longer a shortcut to ranking. AI uses them as an indicator of content engagement, but moving forward, substance matters more than signals.
The key is to focus on creating valuable, engaging content, not just chasing links.
When we mention creating valuable, engaging content, we’re not just talking about individual pages and posts but rather a content ecosystem. That’s where you can put your knowledge of Semantic Search and Topical Authority to good use.
The Rise of Semantic Search & Topic Authority
Google has grown from simple keyword matching to semantic search, where it understands meaning and the relationships between concepts. Instead of chasing individual high-volume keywords, focus on publishing well-organized content to strengthen your topical authority.
AI doesn’t just pull isolated facts. Instead, it tries to understand the relationships between concepts. For instance, it recognizes that sensor size and image quality are often connected, so it brings them together to form a more relevant answer.
So, if you’re telling your content writing team to over-optimize a page because you think you’ll rank higher that way…think again.
For example, instead of optimizing a single page for “email marketing tips,” LLMs and SERPs would be more impressed with a pillar page around email marketing, supported by topic clusters. This pillar page could include content like:
- How to write high-converting email subject lines
- The best email automation tools for small businesses
- How to segment your email list for better engagement
Each of these topic clusters should link back to the pillar page on email marketing, reinforcing topical authority and helping crawlers grasp the relationships between these topics.
How Agencies Can Adapt
- Encourage clients to build content authority in their niche by focusing on depth and expertise. This means going beyond just creating content and building the brand across different platforms, so LLMs have plenty of signals to pull from when determining authority.
- Use schema markup to provide structured information that AI models can process easily.
- Help clients establish credibility through consistent, high-quality, and in-depth content rather than focusing on outdated SEO tricks.
By in-depth content, we also mean including visuals in the mix. Yes, Google’s AI Overview can (and will) surface relevant images and videos in addition to text – a great opportunity for media-heavy clients to get into SEO and LLM-friendly content.
5. Preparing for the Future of AI in SEO
AI-generated search results are only going to increase, pushing traditional rankings further down the page, and AI models will continue to shift the entire search experience.
Personalization is also taking a bigger role. As AI gets smarter, it’ll customize results even more based on individual behaviors and preferences, so that users see what’s most relevant to them based on their search history, location, or interests. For businesses, this means staying agile and aware of who’s searching for what, and when.
We’re also going to see AI-powered local search optimizations that benefit businesses with strong online reputations. Think about how Google Business already helps with local visibility. AI is taking it a step further, using more data points (like reviews and social signals) to boost businesses that show strong community engagement.
Key Takeaways for Agencies
- If chasing keywords based on volume and trying to fit them just right used to be a problem, you can relax! It's now more about becoming a trusted source for AI. Entity-based, authoritative content marketing is the way forward, with the goal of providing real value that's easy for AI to pull from. And your agency already excels at it.
- Be ready for AI and automation to change your SEO workflows, and put tools that can streamline content creation, tracking, and optimization at the top of your list. After all, efficiency will become a major competitive advantage moving forward, especially for smaller agencies trying to stay ahead.
- Adapt, adapt, adapt. Agencies need to stay informed and refine their approach as they go. Google itself is testing its AI overviews in real time because it’s hard to keep up with this shift.
We’re not telling you to stop everything you’re doing but to stay flexible. And even though the things we’ve covered might make SEO seem like it’s on its way out, it’s not dead.
What it does mean is that you need to rethink your approach.
Think of SEO now as part of a new process that focuses on being visible to both LLMs and traditional search engines.
And while you’re at it, make sure you’re positioning yourself (and your clients) as the top industry source!
