I’ve been doing this work for years. And I can tell you with absolute certainty: the panic around “AI killing SEO” is misplaced.
What’s actually happening is way more interesting. And honestly, it’s better for businesses that understand the shift.
Hey, Brandon Leuangpaseuth here…
The real story isn’t that search engine optimization is dead. It’s that what success looks like in search has fundamentally changed. And most people haven’t caught up yet.
Let me explain what I’m seeing on the ground with my clients, and why you need to care about this right now.
The Core Problem Everyone’s Missing
When people say “AI is going to destroy SEO,” they’re usually thinking about one thing: Google’s AI Overviews reducing clicks.
They see the crocodile graph. They see impressions climbing while clicks diverge downward. And they panic.
Here’s the thing though. Users were already moving away from clicking blue links. That trend started way before ChatGPT existed. Mobile changed it. Voice search changed it. Featured snippets changed it.
But people adjusted. And we adjusted. SEO evolved.
This is just the next evolution.
The actual shift is more subtle and more important than most people realize: your goal isn’t to rank first anymore. Your goal is to be cited in the answer by LLMS.
That changes everything about how you should be optimizing. Everything.
What Changed (And Why It Matters)
In traditional search, Google determined which page ranked first for a given query. That was the win. Your job was to get the best position, because that correlated with clicks.
If you compare traditional search with LLM SEO…
Now search engines return synthesized answers. These answers are built from multiple sources. And the sources that appear in those answers get the authority signal, the credibility, the visibility.
If you’re cited as the source inside an AI overview, you win. Your brand gets visibility even if the user never clicks through to your site. And more importantly, the LLM learns to associate your brand with authority on that topic.
This sounds subtle. But it changes the entire strategy.
Think about what happened with featured snippets. Before they existed, everyone obsessed about page one rankings. Featured snippets made rank position five a potential goldmine if you controlled that zero-click real estate.
AI search is similar but bigger. It’s not just about owning one snippet anymore. It’s about being the source the AI system returns across multiple prompts, multiple queries, multiple surfaces.
That’s significantly more valuable. And significantly harder to manufacture.
Training Data vs. Grounding: The Two Paths to Visibility
Here’s what I’m learning from running thousands of prompts across Gemini and ChatGPT:
These models work in two layers. The first layer is the training data. What the model learned during its initial training process. That’s the model’s default opinion about who’s an authority on any given topic.
The second layer is grounding. When the model searches the web in real time to verify answers and pull fresh information, it’s using what we’d call retrieval augmented generation, or RAG.
This is crucial because it creates two completely different strategies.
If you want to influence what a model says about you when it’s not searching, you need to influence the training data. That means building real authority signals across the web. Getting mentioned on authoritative sites. Getting cited by legitimate sources. Building your brand presence at scale. That takes time. Months. Years, even.
But if you want to influence what a model says about you when it is searching, you need to influence the RAG layer. You need to make sure your content is discoverable, extractable, and more credible than competing sources when the model runs a search query relevant to your niche.
That’s faster. That’s more tactical. That’s where I’m seeing wins right now.
Most models search for most queries. Gemini searches roughly 5.7 times per user prompt. ChatGPT searches about 0.7 times. Google’s doing the searching for Gemini answers. So Gemini is weighted toward what it finds in search engine results. ChatGPT leans heavier on training data.
That means if you’re trying to influence Gemini, you need strong search visibility. If you’re trying to influence ChatGPT, you might need to focus more on building ubiquitous brand presence.
Different SEO strategies. Different plays. Same goal.
The Real Metrics That Matter Now
This is where most agencies and consultants are fumbling badly. They’re still tracking CTR. Still obsessing about position.
Those metrics don’t tell you much anymore.
What matters is citation frequency. How often your brand appears as a cited source in AI responses. And more importantly, what the quality of those citations looks like.
Are you cited for the right topics? The right positioning? The right context?
I worked with a client recently where traditional metrics looked terrible. Clicks were down. Impressions were weird. But when we started tracking citations, we saw something different. We were showing up in AI answers constantly. We were building visibility with high-intent users who were consulting AI before making purchase decisions.
Organic traffic from traditional search dropped. But the quality of traffic increased. And most importantly, the early-stage visibility shifted dramatically in our favor.
That’s the real win. Not driving traffic today. Influencing the decision before it’s made.
You don’t measure that with a Google Analytics dashboard. You measure that by running prompts, sampling responses, and seeing where your brand lands in the answer structure.
It’s more work. But it’s also more accurate than trusting that position five rank is helping you.
What Your Content Needs to Do Differently
Here’s the practical part. The stuff you actually need to change.
First: clarity over volume.
LLMs extract snippets from pages. They don’t read the whole thing. The longer your content, the more you’re burying the good stuff.
I’ve started running content through analysis that looks at semantic density. Which sections carry the most information value per word. And consistently, tighter writing wins.
Get to the point. Answer the question in the first paragraph. Use clear headers that signal intent. Break complex ideas into digestible chunks.
Second: specificity over generality.
“How to do SEO” gets crushed in AI summaries because it’s generic. Every tool writes it. Every expert covers it. The model has infinite sources, so pulling from you specifically makes no sense.
But “How to do SEO for construction software companies” is different. That’s specific enough that your expertise shines through. That’s specific enough that when a model is synthesizing for a user in that niche, your content becomes valuable.
Third: entity clarity and structured data.
Models don’t think in keywords the way search engines do. They think in entities. Relationships. What’s connected to what.
If you say “our software helps accounting firms,” that’s good. But if you structure it properly with schema, and you link it to “accounting,” “accounting software,” “tax compliance,” “CPA firms,” you’re teaching the model how to think about you.
Fourth: evidence and attribution.
When you make a claim, cite it. Not just with a link, but with the source in plain text. “According to McKinsey’s 2025 research, 72% of firms are…” That’s better than just linking to McKinsey.
It’s not about manipulating the model. It’s about being the kind of source that generates reliable citations. Because reliable sources get cited more often.
Fifth: multi-modal content.
Video. Diagrams. Infographics. Case studies with screenshots.
Models can process images. And when they’re building an answer for a user, having rich media on your page makes you a more valuable source. You’re not just text anymore. You’re complete information.
The Biggest Misunderstanding (And Why It Costs You)
Most people think the game is: “How do I get into the training data?”
They think if they just get cited enough across the web, eventually the model will learn they’re an authority, and they’ll win.
That’s not wrong. But it’s incomplete.
The actual game is shorter term. It’s: “How do I make sure that when this model searches for information about my niche, my content is the source it returns?”
Training data changes slow. Models retrain maybe twice a year. You can’t wait that long.
But RAG is happening in real time. Gemini searches live for answers. ChatGPT’s training data is stale (January 2025 at latest for most models). So grounding search is actually the fast track to visibility.
That’s where I’m seeing the highest ROI right now for clients. It’s not about influencing the machine learning models over time. It’s about winning the search results that feed into the RAG layer today.
If you’re not in the top search results for the queries models are running, you’re invisible. And if you’re invisible to the models, you don’t get cited.
Why Zero-Click is Actually Good News
I know it sounds insane to be excited about zero-click searches. But hear me out.
A zero-click search where your brand is mentioned and cited is not the same as a zero-click search where it isn’t. The first one is a win. The second one is a loss.
When an AI system cites your brand, even if the user never clicks, you’re building authority. You’re building mind share. You’re influencing the decision before the customer even opens your website.
That’s actually more valuable than a click on a search result that goes to a blog post.
Because it means you’ve moved earlier in the consideration phase. You’re not trying to convince someone to click a link. You’re already positioned as credible before they search your site.
That’s a completely different sales dynamic.
I had a client in B2B SaaS. Their organic traffic dropped about 18% year over year because of AI Overviews in Google (yes, AI overviews can reduce clicks by more than 30%!). They panicked.
But when we looked at AI citations, something interesting happened. They were appearing in Gemini responses for accounting-specific queries at roughly 40% of the time. Not every query. But the high-intent ones where someone was actually comparing solutions.
Traditional organic traffic? Down. But early-stage authority signals? Up significantly. In fact, searches and conversions for brand terms have gone up significantly.
Their sales cycle stayed the same length. But the qualification improved. They were entering conversations later, closer to a buying decision, because they’d already been cited as credible three conversations prior in an AI search.
That’s worth more than ten clicks from irrelevant traffic.
How Models Actually Decide Who to Cite
This is the stuff most people don’t understand because they haven’t tested it.
Models don’t just run a search query and grab the first result. They run multiple queries. They evaluate credibility. They assess freshness. They look at whether the source aligns with what the user is asking.
A model responding to “What’s the best accounting software for construction companies?” might run something like:
- “Best accounting software 2026”
- “Accounting software construction”
- “Construction accounting tools”
- “Best accounting software small business”
- And variations on those
It’s pulling from multiple searches to triangulate the answer. Which means you don’t just need to rank first for the main query. You need to be present and credible across related searches too.
This is actually better than traditional SEO in one way: you can influence this more predictably. Because it’s about being discoverable and credible across a cluster of related queries, not about out-ranking one specific search.
I started running experiments where we deliberately target clusters instead of individual queries. So instead of optimizing for “SEO tools,” we optimize for “Best SEO tools,” “SEO tools for agencies,” “Cheapest SEO tools,” “SEO tools that integrate with Zapier,” etc.
When the model is searching across that cluster, we show up consistently. Which means we get cited more often.
It’s the semantic equivalent of keyword clustering, but applied with AI citation frequency in mind instead of traditional rankings.
The Content Audit You Actually Need to Run
Everyone’s running the same old content audits. “What’s ranking?” “What’s getting traffic?” Useful, but incomplete now.
What you actually need to know is: What’s getting cited?
This requires a different approach. You need to start running prompts. Not one. Hundreds.
Here’s the methodology I use with clients:
First, identify your core topic clusters. What are the main categories you own? What are the sub-categories?
Second, generate synthetic queries that map to those clusters. These are queries that models would likely run when responding to a user question in your niche.
Third, run those queries through the models. ChatGPT, Gemini, Perplexity. Capture what appears in the responses.
Fourth, analyze which URLs are cited most frequently. Which domains? Which specific pages?
Fifth, audit your own content against what’s winning. Are you present? Are you cited? Are you cited for the right positioning?
This audit usually reveals gaps. Sometimes it’s content gaps (you don’t have anything on a topic that’s being cited). Sometimes it’s positioning gaps (you have content, but it’s not structured in a way that makes you citable). Sometimes it’s authority gaps (your content exists but the model trusts other sources more).
Once you know the gaps, you can start filling them strategically.
One client did this and found they were invisible in AI responses for a specific product category they actually dominated. The category was getting cited constantly. Other competitors were showing up. But they weren’t.
Digging deeper, we found the issue: their content on that product was buried three pages deep on their site. It wasn’t prominent. It wasn’t optimized for extraction. It was technically there, but invisible to models.
They reorganized the site architecture, made that content more prominent, restructured it for clarity, and within two weeks they started showing up in citations.
Just from better positioning. Same content. Different structure.
That’s the kind of win that traditional SEO wouldn’t catch. But AI citation tracking catches it immediately.
The Difference Between Visibility and Citations
This matters more than you’d think.
You can be visible in an AI response without being cited as a source. Maybe your content gets referenced in the narrative but not linked or attributed.
That’s visibility. It’s not worthless. The user knows your perspective was in there somewhere.
But citations are different. A citation is when the model says “According to [Your Company], [claim].” That’s credibility transfer. That’s authority building.
Citations are what actually move the needle on brand authority from the model’s perspective.
I started differentiating between the two in tracking. Visibility tracking (does my domain appear anywhere in the response) and citation tracking (is my domain cited as a source).
The citation metric matters more for long-term authority building. But visibility matters for early-stage awareness building.
You want both. But you need to understand which one you’re actually winning on.
Some clients show up constantly in AI responses but rarely get cited. That’s a positioning issue. Your content is relevant but not trustworthy enough to attribute.
Other clients get cited occasionally but not visible often. That’s a scale issue. You’re trusted but not top of mind across queries.
The strategy is different depending on which you’re optimizing for.
How to Think About Competitive Dynamics Now
In traditional SEO, competition is based on ranking position. Everyone’s fighting for position one.
In AI search, competition is based on citation frequency and positioning. It’s less of a zero-sum game and more of a market share game.
If ten sources typically get cited for a query, you want to be in the top three to five. You don’t need to be the only source.
That actually makes it more achievable. Because the bar isn’t “Be better than every competitor.” It’s “Be credible enough to include in the answer.”
I looked at a competitive set recently where the “winning” position showed up in AI citations about 60% of the time across a topic cluster. Second place was around 40%. Third was around 25%.
The difference between first and third? A few key content pieces. Better positioning. Slightly fresher updates. A bit more tactical depth.
Not a massive gap. Just slightly better execution on things they could control.
That’s a different kind of competitive dynamic. It’s not about crushing competitors. It’s about being credible enough to include.
The Measurement Problem (And How to Solve It)
Here’s what nobody wants to admit: most analytics platforms can’t track this well.
Google Analytics sees the traffic (or lack thereof). But it doesn’t see the citations. It doesn’t see the brand mentions. It doesn’t see the early-stage authority signals.
You need supplementary tracking. And most people don’t know how to set it up.
The simplest approach: pick a set of core prompts that map to your business. Run them weekly or monthly through available models. Track which URLs appear in the responses. Track the positioning within those responses. Track trends over time.
It’s manual. It’s not scalable to hundreds of queries. But for your core metrics, it works.
Some newer tools are building better tracking for this. But they’re still catching up. Most of them are sampling, not comprehensive.
I use a hybrid approach. I run my own prompt testing for the top 20-30 queries that matter most. And I use third-party tools to spot check broader trends.
That gives me actionable data without being drowned in noise.
The key insight: you can’t manage what you don’t measure. And most people aren’t measuring this yet. So if you start, you immediately have a competitive advantage.
What’s Going to Happen in the Next 12 Months
Based on what I’m seeing, here’s what’s coming:
- First, models are going to get better at evaluating source credibility. Right now there’s a lot of randomness in citation selection. That’s going to tighten. Which means authority signals will matter more.
- Second, more publishers are going to start explicitly optimizing for AI visibility. Right now most people are still confused. But as the data gets clearer, everyone’s going to pile in. That’s going to make the competition tighter faster.
- Third, advertising is going to move into AI search in a bigger way. Google’s already testing ads in AI Overviews. Once that revenue stream proves out, every platform’s going to lean into it. That changes the dynamics for organic citations.
- Fourth, brand mentions without citations are going to become increasingly valuable. As citation battles get fiercer, the early-stage authority of being mentioned (even without a link) is going to get more competitive attention.
- Fifth, content quality is going to matter more. As more people optimize content for AI visibility, the mediocre stuff is going to get filtered out. Which means your content actually needs to be good. Not just strategically positioned.
If you’re moving now, while everyone’s still figuring this out, you’re ahead of the curve. In 12 months, the tactics you’re using to win today are going to be table stakes.
That’s why urgency actually makes sense here. Not fake urgency. Real “window of competitive advantage” urgency.
The Tactics That Actually Work (And What Doesn’t)
I’ve tested a lot of stuff. Some of it works. Some of it is noise.
What works: Writing specific content for specific use cases and personas.
Not “Best SEO Tools” but “Best SEO Tools for Enterprise Companies.” Not “How to Do Marketing” but “Content Strategy for B2B Tech Companies.” The specificity filters out noise and makes your content more valuable to models searching for answers.
Getting citations on authoritative sites. This is just good PR and digital PR fundamentals. But now you’re doing it with AI visibility in mind. You’re asking “Will this get cited by models?” not just “Will this rank?”
Making your site technically clean. Rendering issues trip up models. Slow servers trip up models. Fix the basics. Make your content crawlable and parseable.
Publishing fresh content regularly. Models prioritize recency. If you published something six months ago, it’s losing to something published last week, all else equal.
Building your brand presence across platforms. LinkedIn, Twitter, YouTube, wherever your audience is. When your content is everywhere, models learn you’re an authority faster.
What doesn’t work (or at least, what I’m skeptical about): Posting on Reddit just to game LLMs. Reddit is important for Perplexity citations — and Reddit content does show up in other LLM citations sometimes. But gaming it with bots or manufactured content is a race to the bottom. However, posting on Reddit the right way could work.
Schema markup as a silver bullet. Schema helps. But it’s not magic. Clean, clear content beats fancy schema on generic, mediocre content.
“LLM-only pages” designed for machines but not humans. Sounds clever in theory. In practice, models are trained on human-readable content. So content written for humans converts better than content written for machines. It still doesn’t hurt to have an LLM.txt page for SEO as it requires practically zero effort.
Playing in overly competitive categories. You don’t win “Best SEO Tools” as a new player. But you can win “Best SEO Tools for Ecommerce Brands” pretty quickly with good content.
The Conversation is Shifting (And You Should Too)
Here’s what’s happening in the industry right now that matters:
Teams are moving away from pure SEO specialists to hybrid teams. They’re bringing in PR. They’re bringing in content strategy. They’re thinking about brand building alongside visibility. They are hiring AI SEO specialists.
Metrics are shifting. CTR is dying as a primary metric. Citation tracking is becoming essential. AI referral traffic is being measured separately.
Competitive dynamics are changing. The traditional link-building industry is struggling because not all links are equal anymore. Some create authority signals, some don’t. You need judgment, not just volume.
Agencies that can’t adapt are getting left behind. The ones that understand both traditional search and AI search are winning big.
This is actually good news if you’re willing to shift. Because it means there’s a window right now where you can move fast. The competition hasn’t fully adjusted. The best tactics aren’t yet saturated.
But that window isn’t going to stay open forever.
What This Means for Your Business
If you’re sitting on organic traffic that’s still healthy, don’t panic. Keep doing the fundamentals. Those still matter.
But start paying attention to AI visibility immediately. Start tracking citations. Start understanding which queries models are running. Start optimizing for that.
If you’re in a vertical where LLM search is taking share fast, you need to be aggressive about positioning now. Because in six months, this will be table stakes.
If you’re building something new or relaunching, think about AI visibility from day one. Think about how you’re positioning relative to what models will search for. Build content with that in mind.
Don’t rebuild everything. But shift the thinking. Shift where you’re investing new effort.
How to Adjust Your Strategy by Vertical
I don’t want to oversimplify this. Different verticals have different dynamics when it comes to AI search.
- B2B SaaS is ahead of the curve. These companies have early adopters who are using LLMs to evaluate solutions. Citation patterns are already clear. The competition is moving fast. If you’re in this space, move faster.
- E-commerce is different. AI shopping agents are emerging, but the citation dynamics are still settling. What works is having clear product specifications, real reviews, and being present on comparison sites that models cite. Less about thought leadership, more about data clarity.
- Agency services (SEO, marketing, design, etc.) are interesting because there’s a lot of personal brand involved. Individual practitioners can build authority faster here because models cite people in addition to companies. If you’re in this space, your personal positioning matters as much as your company’s.
- Professional services (legal, accounting, consulting) are moving slower. But when they do move, citation frequency matters enormously. These are high-intent, high-stakes decisions. Being cited as credible in an AI response is worth serious money.
- For each vertical, the tactical approach shifts. But the fundamental principle doesn’t: understand what models are searching for in your niche, make sure you’re discoverable and credible for those searches, and get cited consistently.
The Specific Steps to Start Winning Today
If you want to actually move on this (not just understand it intellectually), here’s the roadmap:
- Step one: Pick your top 20-30 core queries. These are the searches that matter most to your business. Not vanity metrics. Real business-driving queries.
- Step two: Run those queries through Gemini, ChatGPT, and Claude. Take screenshots. Track which URLs show up. Track the positioning. Track whether you’re cited or just visible.
- Step three: Compare what you find to what’s ranking in traditional Google search. Notice the gaps. Usually there are some.
- Step four: Audit your content against what’s winning in AI responses. Is your content structured for extraction? Is it clear? Is it authoritative? Is it fresh?
- Step five: Make three to five targeted improvements. Could be content reorganization. Could be new content creation. Could be freshness updates. Pick the highest-impact changes.
- Step six: Re-run the prompts two weeks later. Check for changes. Adjust and iterate.
This is the work. Not complicated. But it requires discipline. And it requires actually testing instead of guessing.
Most people won’t do this. Most people will keep doing what they’ve always done. Which is why there’s still a window for competitive advantage.
But the window is closing. Fast.
The Real Risk If You Wait
Here’s the thing that actually keeps me up at night about this:
If you wait another six to twelve months to move on AI visibility, you’re betting that traditional search traffic stays relevant long enough to give you time.
That might be true. Or it might not be.
What I’m seeing is that AI traffic is growing exponentially. Adobe reported a 10x increase in generative AI referral traffic from mid-2024 to early 2025. That’s not slowing down.
Meanwhile, organic click-through rates for informational queries have dropped 60%+ when AI Overviews appear. That’s not a small number.
If you’re in a category where AI traffic is already taking significant share, your window for traditional traffic extraction is closing. The users you’re trying to reach are already consulting AI first.
If you’re not visible in those AI responses, you’re invisible to those users. Period.
And the further behind you get, the harder it is to catch up. Because the authority signals that models use to decide citation frequency compound. If you’re cited consistently, future models learn to cite you. If you’re not, they don’t.
It’s not like traditional SEO where you can suddenly create a viral content piece and rank number one. Authority in AI systems is built through consistency over time.
So the cost of waiting isn’t just “we’ll optimize for this later.” It’s “we’re conceding market share to competitors who are moving now.”
That’s a different calculation. And it’s why this actually does warrant some urgency.
What Success Actually Looks Like
I want to be clear about what you’re optimizing for, because I think a lot of people have the wrong mental model.
Success isn’t:
- “We’re ranked number one for X keyword”
- “We have a ton of backlinks”
- “Our content is the longest article out there”
Success is:
- You’re cited consistently in AI responses for your core queries
- You show up as a trusted source across related search variations
- New users learn about you through AI citations before they ever visit your website
- Your sales team reports that prospects already know who you are when they get on calls
That’s the actual win condition. That’s what translates to business outcomes.
Everything else is theater.
The Conversation You Need to Have Inside Your Organization
If you’re an in-house marketer, you probably need to have this conversation with your leadership. Because this requires a strategic shift. Not just tactics.
It means reallocating resources. It means new tools. It means different metrics on dashboards. It means potentially expanding the team to include people who understand brand building and PR alongside SEO.
If you’re an agency, you probably need to be preparing to help clients understand this shift. Because they’re going to ask. And you need to have a clear answer.
If you’re a founder, this might be one of the most important competitive advantages you can build right now. Because most competitors haven’t figured this out yet. But they will. And soon.
The window to be ahead is now. Not next quarter. Now.
The Bottom Line
SEO isn’t dead. But the definition of SEO success has changed.
It’s not about ranking first anymore. It’s about being cited. Effective SEO strategies is about building authority in the eyes of LLMs. About positioning your content so that when machines search for answers, they find you.
That requires different thinking. Different metrics. Different tactics.
But it’s also way more interesting work than trying to chase position one in a traditional search result.
AI will eventually replace SEO.
The future of search is synthesis, not ranking.
The future of winning that future is being the source the synthesizer returns.
That’s where I’m focused. That’s where you should be focused too.
Start today. Don’t wait for someone else in your industry to figure this out first.
