Every few years, someone declares SEO dead.
I’ve been doing this for 20 years. I’ve watched it “die” when Panda rolled out. When mobile-first indexing landed. When voice search was going to change everything. When featured snippets were going to eat every click. And now, with AI search, the eulogies are back on the front page.
Here’s the thing: SEO isn’t dead. It just got more interesting.
And the businesses that adopted AI search optimization early? They’re already winning. Their content is already getting pulled into ChatGPT answers. Their brand is already showing up in Google’s AI Mode results. While everyone else is still arguing about whether the sky is falling, they’ve quietly moved in.
Let’s talk about why.
What actually changed
Google’s AI Overviews, ChatGPT, Perplexity, Claude — people are asking questions and getting answers without clicking through to a website. That’s real. If your whole strategy was “rank #1 and hope for clicks,” you’ve got a problem.
But the bigger story isn’t “SEO is dead.” It’s that search now has two jobs:
- Get your site ranking in traditional results (still matters)
- Get your brand cited inside AI answers (the new job)
Both depend on the same foundation: clear, useful content written by a real expert, on a site that search engines can actually understand.
How AI search actually works (and why it matters for you)
Before we get into what to do, it helps to know what’s happening under the hood. The two systems most of your customers are using are ChatGPT and Google’s new AI Mode — and they work differently enough that it’s worth a minute.
Here’s how ChatGPT actually works
In plain English:
- Someone asks a question.
- ChatGPT figures out what they actually mean.
- It decides — do I need fresh info, or can I answer from what I already know?
- If it needs fresh info, it runs a Bing search, fetches the top pages, reads them, and summarizes.
- It answers, sometimes with citations linking back to the pages it used.
Translation for marketers: if you want your content pulled into ChatGPT answers, your Bing rankings matter. Yeah — Bing. The thing everyone stopped thinking about a decade ago is now feeding a chunk of AI search, thanks to Microsoft’s partnership with OpenAI.
And here’s how Google AI Mode works
Google’s take is different. Instead of one search, it uses something called query fan-out. Google’s own documentation explains it simply: AI Mode breaks your question into related sub-queries, issues multiple searches at once across different sources, and synthesizes the results into one response.
Here’s the practical version:
- Someone asks a complex question.
- Google breaks it into a bunch of smaller, related sub-questions.
- It runs searches on all of those at once and pulls info from trusted sources.
- Gemini synthesizes everything into one answer.
Translation: your content needs to answer the smaller specific questions underneath the big broad one, not just the headline. If someone asks “what should I look for in an Edmonton web design agency?” — Google isn’t just searching that phrase. It’s searching “typical web design pricing,” “how long does a website take to build,” “what’s included in a WordPress project,” and about a dozen others.
The sites that show up in the final answer are the ones that have clear content on each of those sub-topics. Google rolled AI Mode out across the US in 2025, and the sub-query behaviour is now baked into both AI Overviews and AI Mode.
The content hub: how you actually build for this
Once you understand query fan-out, the content strategy writes itself — but most businesses still don’t do it. The answer is a content hub.
A content hub is one main “pillar” page on your core topic, surrounded by a cluster of supporting pages that each answer a specific sub-question, intent, or objection. Like this (using white-label SEO as the example):
- Main hub page:
/white-label-seo-hub— the big entity page - Definition: What is white-label SEO?
- Process: How does it work for agencies?
- Evaluation: Pros and cons
- Comparison: Us vs. in-house SEO
- Objection: Pricing explained
- Authority: Case studies
- Expert layer: Advanced strategies
Each page is a clean, self-contained answer to one sub-question. Internal links point from the hub to every supporting page, and back again. The whole thing forms a topical map that tells both Google and the LLMs: this site knows this topic inside out.
Translation: when a user asks AI Mode a complex question and query fan-out kicks in, your hub gives the AI a clean, structured answer for every sub-query it generates. You don’t just compete for one keyword. You compete for the whole topic.
Think of it like a resource hub, not a pile of blog islands.
Chunk-level retrieval: how LLMs actually grab your content
Here’s the piece most SEO advice still misses. LLMs don’t read your whole page and then decide to cite you. They pull chunks — short passages that cleanly answer a specific sub-question.
Think about how the retrieval side works:
- The AI has limited token space for any given answer
- It can’t ingest a 3,000-word wall of text and surgically find the one good paragraph
- So the retrieval system pulls clean, self-contained “chunks” from the page — and discards the rest
Translation: shorter, tighter paragraphs = higher chances of being cited.
What this looks like in practice:
- One idea per paragraph. If a paragraph needs to say two things, split it.
- Clear section headings that match the sub-question being answered.
- Pull-quote logic. Could a robot lift this paragraph out of context and have it still make sense? If yes, you’re chunk-friendly.
- Direct answer first, context after. Don’t bury the point three sentences deep.
This isn’t about dumbing content down. It’s about writing in a way that lets humans skim and AI models extract. Notice the structure of this blog post — short paragraphs, clear subheads, one idea at a time. That’s on purpose.
Technical SEO: the foundation for both
Here’s the thing that gets lost in the AI search hype: none of the above matters if the robots can’t read your site.
The same four technical fundamentals that have always driven SEO are now gatekeeping AI visibility too. Let’s walk through them.
Crawlability and indexability
Googlebot isn’t the only crawler knocking on your door anymore. GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot, Google-Extended — they’re all visiting your site to decide whether your content is worth citing. If your robots.txt accidentally blocks them, or your pages are noindexed when they shouldn’t be, you’re invisible by default.
Check: your robots.txt is clean, your sitemap is submitted, and the AI crawlers you want visiting are allowed.
Page speed and Core Web Vitals
A fast site still signals quality — to humans and to the algorithms judging your content. Slow sites get crawled less, ranked lower, and cited less. Core Web Vitals aren’t a vanity metric. They’re table stakes.
Mobile-friendliness
Google’s been mobile-first for years. Most of the web the AI models are trained on is the mobile version. If your site breaks on a phone, you’re invisible to most of the algorithms deciding whether to surface your content.
Log file analysis
This one nobody does, and it matters more than ever. Your server logs tell you exactly which crawlers are visiting, how often, and which pages they’re interested in. With AI crawlers now active on most decent sites, log analysis is the only way to see whether GPTBot, ClaudeBot, PerplexityBot, and Google-Extended are actually indexing your content.
If you’re not looking at your logs, you’re flying blind on AI visibility.
Entities and structured data: teaching AI who you are
One of the biggest shifts in how search engines and LLMs evaluate sites — and it’s bigger than most people realize — is the move from keywords to entities.
An entity is a “thing” the web knows about: your business, your team, your services, your location, your reviews. When Google or an LLM understands that “Holler Digital” is a thing — a business, based in Edmonton, that does WordPress and SEO — it can connect you to relevant searches even when the exact keywords aren’t in the query.
There are three steps to get this right.
Step 1: Identify your core entities
Every business has a handful of key entities that should be clearly defined across the site:
- Business entity — your organization (name, description, address, contact info)
- Key people — founders, experts, authors (especially important for E-E-A-T)
- Services — primary services plus their sub-services
- Locations — offices, service areas, branch locations
Step 2: Mark them up with structured data
This is where structured data (schema markup) comes in. It’s code added to your site in JSON-LD format that tells search engines and LLMs exactly what each entity is.
Standard schema types every local business should have:
- Organization or LocalBusiness — for the business itself
- Person — for key team members
- Service — for each service you offer
- FAQ — for common questions
- Review / AggregateRating — for social proof
For AI visibility, go one step further. Beyond standard schema, include “answer-ready” metadata in your HTML — clean definitions, consistent naming, and descriptive content baked into the markup itself. The retrieval systems feeding LLMs (the tech behind this is called RAG — retrieval-augmented generation) are looking for clean, structured, machine-readable signals they can lift directly into answers.
Step 3: Build entity relationships
Here’s the part that matters most for AI visibility. LLMs don’t just see individual entities — they see relationships between them.
When you write a clear sentence like “Holler Digital provides WordPress web design, SEO, and hosting for small businesses in Edmonton,” the AI learns a whole set of relationships at once:
- Holler Digital → provides → WordPress web design
- Holler Digital → provides → SEO
- Holler Digital → serves → small businesses
- Holler Digital → operates in → Edmonton
Translation: don’t be cute about what you do. Say it clearly, often, and consistently across your site, your schema, your GBP, and your directory listings. The clearer you are about who you are and what you do, the more accurately AI models will cite you when the relevant question comes up.
What didn’t change
A reminder that not everything is new:
Good content still wins. Content that answers a real question, written by someone who actually knows the topic, still outperforms AI-generated filler. That’s the core of Google’s helpful content guidelines, and the LLMs are trained on the same signal.
Local still rules for local businesses. If you’re an Edmonton plumber, a roofer, or a marketing manager looking for local services, people still search “Edmonton [thing]” on Google. That pattern hasn’t changed.
What’s new (and worth your attention)
Three more shifts worth flagging:
1. You need to know what your customer is actually typing
This is the one most agencies skip, and it’s the most important.
SEO used to be about ranking for “web design Edmonton.” AI search is about answering the actual question your customer is asking — which is almost never a short keyword. It’s a full sentence. “How much should a new website cost for a small business?” “What’s the difference between a web designer and an agency?” “Can I edit my own WordPress site after it’s built?”
If you don’t know your ideal customer well enough to predict those questions, you can’t write the content that answers them. And if you can’t answer them, the AI will find someone who does.
Translation: your customer persona work isn’t a fluffy branding exercise anymore. It’s the foundation of your content strategy.
2. Your Google Business Profile is doing more work than ever
For local service businesses, GBP is the new homepage for a lot of searchers. Posts, reviews, photos, service lists, Q&A — all of it feeds the local pack and the AI summaries. If your GBP is half-filled and hasn’t been touched in six months, that’s low-hanging fruit.
3. Brand searches matter more than they used to
People searching your name directly is one of the strongest trust signals out there. It tells Google — and the LLMs — that you’re a real, known entity. PR, partnerships, local events, podcast appearances, actually being part of your community — all of it feeds into your AI visibility, even if you can’t directly measure it.
The early-adopter gap is already here
Here’s the part nobody wants to hear: the businesses that started taking AI visibility seriously a year ago are already pulling ahead. The content they published six months ago is now training the models that answer their customers’ questions today.
And the gap is compounding. Every month a competitor is cited in ChatGPT and you’re not, they’re building the kind of digital authority that used to take years of backlinks to build.
Good news: the bar for catching up is still surprisingly low. Most businesses haven’t done this work yet. The ones who start now — not next quarter, now — will be the default citations for their niche before their competitors realize the game has changed.
So what should you actually do?
If you’re a marketing manager watching your organic traffic and wondering whether to panic, don’t. Here’s the short list:
- Know your ICP cold. What does your customer actually type? Not keywords — full questions. Build content around those.
- Build a content hub, not blog islands. Pick your core topic. Map out the sub-questions, objections, and comparisons. Write a page for each. Link them all back to a pillar page.
- Write for chunks. Shorter paragraphs, clear subheads, one idea at a time. Direct answers up front.
- Nail the technical foundation. Crawlable site, fast load times, mobile-friendly, and check your logs to see which AI bots are actually visiting.
- Map your entities and add schema. Business, people, services, locations. Mark them up with JSON-LD and keep the language consistent across every platform.
- Audit your top 10 pages. Are they genuinely the best answer to their query? Or were they written to hit a word count?
- Clean up your Google Business Profile. Photos, posts, services, reviews — all current.
- Get clear author bios on your content. Real people, real credentials.
- Check where you’re being mentioned outside your own site — directories, news, local coverage. Fix any inconsistencies.
This isn’t different from what good SEO looked like five years ago. The bar is just higher, and the payoff is bigger — the same work now gets you visible in Google, AI Overviews, and ChatGPT answers at the same time.
One thing I didn’t cover here: how to actually measure whether any of this is working. That’s big enough to deserve its own post — read How to Actually Measure AI Search ROI for the four metrics every marketing manager should be tracking in GA4 right now.
The bottom line
SEO didn’t die. It grew up.
The businesses winning in this era aren’t the ones chasing the latest hack. They’re the ones who know their customer inside out, build real authority on their topic, keep their technical house in order, and show up consistently across search, local, and AI results.
Same game. New playing field. And the scoreboard is already lighting up.
Want to know where your site stands in the new search landscape? Book a 15-minute Zoom and I’ll walk you through what’s working, what’s holding you back, and where AI search is already affecting your visibility.
