Quick Summary
Danny Sullivan (Director of Google Search) confirmed that AI Search runs on three mechanisms — general knowledge, traditional search results, and fan-out — and that good SEO is good GEO. The fundamentals didn't die. The penalty for ignoring them got steeper, and dealers who publish unique, specific, authentic content win the new SERP.
What You Should Know
For Dealer Principals
- Google's own Director of Search just confirmed there is no separate AI website to build, no parallel content strategy, no $50K platform you need to license — the fundamentals win.
- The single biggest content shift: 60%+ of new content should be 'non-commodity' (unique, specific, authentic first-hand expertise) instead of generic dealer-channel filler.
- Many vendors selling 'AI-generated city pages' or '1,000 pages a month' content packages violate Google's published spam policy on scaled content abuse — some have already triggered manual actions on dealer sites we've audited.
For GMs
- AI Search runs 'fan-out' on every shopper query — a single question expands into a constellation of related searches pulling from Shopping, Local, Knowledge Graph, Web, Video, and Image surfaces simultaneously.
- Whichever dealer shows up across more fan-out branches wins the answer; competing on the webpage layer alone leaves traffic on the table.
- Stop measuring AI Search by traditional vanity metrics — track time on site from AI-referred traffic, VDP engagement depth, and conversion quality instead of raw sessions.
For Marketing Directors
- Schema coverage on Vehicle, AutoDealer, LocalBusiness, FAQ, Review, and Service is the single highest-ROI technical fix in dealership SEO right now and feeds AI fan-out across multiple Google surfaces.
- Audit your last 50 published pages: how many qualify as non-commodity under Sullivan's three-part test (Unique, Specific, Authentic)? Honest answer for most dealerships is zero to three.
- Expand into Shopping (Merchant Center), Local (GBP completeness), Video (walkarounds with VideoObject schema), and Image (descriptive alt + filenames) — Sullivan called these out specifically as 'new opportunities.'
“When Danny Sullivan walked through fan-out, commodity vs. non-commodity content, structured-data audits, and the three-pillar AI architecture, he was describing the exact framework we built GALAXY around eighteen months before he gave that talk. The fundamentals didn't die — they just became the line between the dealers AI cites and the dealers AI ignores.”
Tim Boyle
Founder & President, A3 Brands
Danny Sullivan, Director of Google Search, recently gave a 22-slide industry talk on how AI is being integrated into Google Search. It's the most direct on-the-record explanation Google has offered on how AI Search actually evaluates content.
If you're a Dealer Principal, General Manager, or Managing Director, the substance of those slides matters more than most of what's circulating in dealership marketing right now.
The last twelve months have buried the dealership channel in acronyms. GEO. AEO. LLM SEO. AI SEO. Vendors are pitching "AI optimization packages." Conference speakers are warning that "traditional SEO is dead." Inboxes are full of platforms promising to make your dealership "AI-visible" for an additional retainer.
Sullivan's premise cut through all of it: good SEO is good GEO. The principles that drove organic visibility for the last decade are the same ones driving AI visibility today. The fundamentals didn't die. They got more important — and the penalty for ignoring them got steeper.
This post breaks down what Sullivan said, what it means for the dealership channel specifically, and what we recommend you do about it before Q3.
Part 1: How AI Search actually works
Sullivan's presentation was structured around a simple thesis: AI Search is not a separate system bolted onto Google. It is Google. And it works through three reinforcing mechanisms.
1. The AI has general knowledge.
The model brings background context to every query — patterns it has learned from massive volumes of content. Think of this as the conversational layer. It's why ChatGPT and Google's AI Overviews can answer general questions without citing a specific source.
For dealers, this is the layer you cannot directly influence. You're not going to retrain a foundation model. Move on.
2. The AI pulls specific knowledge from traditional search results.
This is the layer that matters. When a shopper asks an AI search engine *"what's the best midsize luxury SUV under $70K with adaptive cruise,"* the AI doesn't pull that answer from thin air. It runs a search, identifies relevant pages, and synthesizes an answer from the content it finds.
If your dealership's content isn't in the index, isn't optimized, isn't crawlable, and isn't substantive — it doesn't exist in the AI's answer. Period.
3. Fan-out.
This is the part most dealers don't understand, and it's the most important. When a user submits a single query, the AI doesn't just run that one search. It expands the original question into a constellation of related queries — then pulls sources for each one.
Sullivan's example was bicycles, but here's the dealership translation. A shopper types: *"reliable family SUV for snow with low maintenance costs."*
Behind the scenes, the AI is running queries like:
- ●*"best AWD SUVs for snow"*
- ●*"most reliable midsize SUVs 2026"*
- ●*"cheapest SUVs to maintain"*
- ●*"family SUV third row snow capability"*
- ●*"lowest cost ownership SUV"*
And it's pulling sources across categories Google explicitly named on stage: Shopping, Knowledge Graph, Real World (local), Web, Sport, Weather, Finance.
For your dealership, that means a single shopper question can pull from your VDPs, your model landing pages, your local pages, your Google Business Profile, third-party review aggregators, OEM content, and Edmunds — all simultaneously. Whoever shows up in more of those fan-out branches wins.
How AI Search Resolves a Single Shopper Query
Shopper asks one question
e.g. 'Reliable family SUV for snow with low maintenance costs'
AI fans the query out
Expands into a constellation of related searches: AWD SUVs for snow, reliable midsize SUVs, cheapest to maintain, third-row snow capability, lowest cost of ownership
AI pulls sources across surfaces
Shopping, Knowledge Graph, Local, Web, Video, Image, Sport, Weather, Finance — all in parallel
AI synthesizes the answer
Whichever dealer appears across more fan-out branches wins the citation
Part 2: Three myths Google just debunked
Sullivan dedicated an entire portion of the talk to debunking common SEO myths circulating in the AI era. Three deserve your direct attention.
Myth 1: You need to rewrite your content with "conversational keywords." You don't. Google explicitly stated: *Don't worry if you don't anticipate every variation of how someone might seek your content. Google's language matching systems are sophisticated and can understand how your page relates to many queries, even if you don't explicitly use the exact terms.*
Translation for dealerships: stop letting vendors talk you into rewriting every model page to include eighteen variations of *"best 2026 RX 350 near me."* Google understands intent. Write for humans.
Myth 2: You need to "chunk" your content into AI-friendly micro-blocks.
You don't. Sullivan said it clearly: *No need to chunk content into pieces just for AI; organize and write for a good, human reading experience.* The web in general, he noted, is not even valid HTML — Google rarely depends on hidden semantic meanings.
Translation: that "AI-optimized chunking" upcharge some vendors are quoting? Largely a myth. Write organized, well-structured content for humans. The AI handles the rest.
Myth 3: You need every third-party "AI visibility" score in your dashboard.
This one is the most important. Sullivan said it on stage, on the record: *Google does not evaluate third-party tools. They have no access to our internal metrics. People should think critically about anything they hear and check against the guidance we actually offer.*
That should land hard. The dealership channel is awash in dashboards measuring "AI mentions," "LLM citations," and "GEO score." Google itself just told the world: those scores are being calculated by tools with no access to Google's internal systems. They are estimates. Treat them that way.
Part 3: Commodity vs. non-commodity content
If you take one thing from this entire post, take this.
Sullivan put up a slide that drew a hard line between two types of content: Commodity and Non-Commodity. He used three industry examples — running stores, real estate agents, and interior designers — and the comparison was devastating.
A commodity-content running store publishes *"Top 10 Things to Consider When Buying Running Shoes."* Standard advice on sizing, arch support, and cushioning. The kind of article that already exists on ten thousand other sites.
A non-commodity running store publishes *"Why This Customer's Shoes Collapsed After 400 Miles: A Wear Pattern Analysis."* A specific customer. A specific outcome. First-hand expertise that no aggregator can replicate.
Now apply that lens to dealership content.
Commodity automotive content looks like this:
- ●*"Top 5 Reasons to Lease vs. Buy"*
- ●*"What is AWD?"*
- ●*"2026 [Model] Features and Specs"*
- ●*"How to Get Approved for an Auto Loan"*
You have probably published versions of all of these. So has every other dealer in your DMA. So has every OEM site, every Edmunds article, every CarGurus blog. The AI has thousands of sources for these answers and no reason to choose yours.
Non-commodity automotive content looks like this:
- ●*"Why We Refused to Take a 2019 Range Rover on Trade — and What We Found Under the Carpet"*
- ●*"Three Customers Cross-Shopped Us Against [Competitor] Last Quarter. Here's What Made Them Buy From Us."*
- ●*"The Service Issue We See on Every 2021 [Model] After 60K Miles — And How We Fix It in 90 Minutes"*
- ●*"Why I Stopped Recommending Extended Warranties to Customers Who Lease"*
Sullivan defined good non-commodity content with three words: Unique. Specific. Authentic. It brings a viewpoint others lack. It talks about specific instances, not generic rules. It demonstrates first-hand experience.
This is the content the AI rewards. This is the content that wins fan-out. And almost no dealership is publishing it.
Commodity vs. Non-Commodity Dealer Content
| Feature | Commodity (AI ignores) | Non-Commodity (AI cites) |
|---|---|---|
| Topic | Top 5 Reasons to Lease vs. Buy | Why I Stopped Recommending Extended Warranties to Lease Customers |
| Topic | What is AWD? | Why We Refused to Take a 2019 Range Rover on Trade — What We Found |
| Topic | 2026 Model Features and Specs | The Service Issue We See on Every 2021 [Model] After 60K Miles |
| Topic | How to Get Approved for an Auto Loan | Three Customers Cross-Shopped Us Last Quarter — Here's Why They Bought |
| Voice | Generic, replicable | Unique, specific, authentic |
| AI behavior | Thousands of sources, no reason to cite yours | First-hand expertise no aggregator can replicate |
Part 4: The good news — you don't need to panic
One of Sullivan's final slides was a single sentence: *People don't need to panic or rip apart their sites for AI Search success.*
Read that twice.
Google is telling the industry — directly — that the businesses doing fundamental SEO well, publishing strong content, structuring their data correctly, and providing good page experiences are the businesses winning AI Search. There is no separate "AI website" you need to build. There is no parallel content strategy. There is no $50K platform you need to license.
What there is, is a higher bar for content quality, a stronger emphasis on first-hand expertise, and an expanded set of opportunities — shopping listings, local listings, video, images, and the new Business Agent surface in Google Search — that most dealers are not yet competing for.
Which brings us to the playbook.
Part 5: The dealership playbook before Q3
Sullivan closed his presentation with a "So What to Do?" matrix. Here's the dealership-specific translation, organized by audit area. Walk each checklist across your team.
Content.
Sullivan's three-part test for non-commodity content is Unique, Specific, Authentic. Audit your last 50 published pages against it — honest answer for most dealerships is zero to three. Build a quarterly editorial calendar where at least 60% of new content qualifies. Source ideas:
- ●Service-bay observations
- ●GM perspectives
- ●Specific customer stories (with permission)
- ●Cross-shop breakdowns against named competitors in your DMA
- ●Trade-in pattern analysis
- ●Local market commentary
Anything no other dealer in your DMA can write because they didn't live it qualifies.
Page Experience.
Sullivan said page experience "remains foundational." Run the audit on each surface:
- ●VDPs — Core Web Vitals (LCP, INP, CLS) on mobile
- ●Model landing pages — Core Web Vitals on mobile
- ●Inventory search — page weight and time-to-interactive
- ●Mobile-first parity — desktop and mobile content match
If your VDPs load slowly on mobile, you're invisible to both shoppers and the AI.
SEO Fundamentals & Structured Data.
Most dealership sites run 30-50% of available schema correctly. This is the single highest-ROI technical fix in dealership SEO right now, and it directly feeds AI fan-out across the Knowledge Graph and Shopping branches. Audit each entity:
- ●Vehicle — every VDP marked up with offer, price, availability
- ●AutoDealer — homepage with full address, hours, phone, OEM brands
- ●LocalBusiness — geo-coordinates plus sameAs links to social profiles
- ●FAQPage — at least 4 Q&A pairs on key information pages
- ●Review — aggregate rating and sample reviews on review-rich pages
- ●Service — each fixed-ops service marked up with provider and service area
Shopping, Local, Video, and Image SEO.
Sullivan called these out specifically as "new opportunities" — surfaces beyond the webpage layer where AI Search pulls answers. Run this audit on each one:
- ●Shopping — Are your VDPs in Google Merchant Center?
- ●Local — Is your Google Business Profile fully populated with services, photos, and Q&A?
- ●Video — Do you have model walkarounds with proper VideoObject schema?
- ●Image — Are your inventory photos optimized with descriptive alt text and filenames?
Most dealers compete on the webpage layer alone. The dealers expanding into shopping, local, video, and image surfaces win the new SERP.
Agentic.
Sullivan flagged this as an active development area. Two questions to ask:
- ●Are you eligible for Google's new Business Agent in Merchant Center?
- ●Are you prepared for Universal Commerce Protocol (UCP) integration?
The dealers who adopt the Business Agent surface early will own a conversational shopping channel directly inside Google Search.
Part 6: How to measure what now matters
Sullivan made a quiet but critical point about measurement: *We've seen when people click from AI Overviews, they're more likely to spend more time on the site. AI results may give people more context about a topic overall. This may provide a more engaged audience.*
Translation for dealership reporting: stop measuring AI Search success by the same vanity metrics you used for traditional SEO.
Old framework:
sessions, bounce rate, keyword rankings.
New framework:
time on site from AI-referred traffic, VDP engagement depth, lead form completion rate from AI-Overview traffic, service appointment scheduling from AI surfaces, phone call quality from AI-referred sessions.
If your reporting still emphasizes raw traffic over conversion quality, you are measuring the wrong thing for the era you are operating in. The dealers who reorient their dashboards now will be the ones who can prove ROI on AI Search investment six months from now. The dealers still fixated on session counts will be the ones whose budget gets cut when the GM asks *"what is this actually doing for us?"*
A word on AI-generated content
Sullivan addressed this directly, and dealers need to hear it: *Generative AI can be useful when researching a topic and to add structure to original content. However, using generative AI tools to generate many pages without adding value for users may violate Google's spam policy on scaled content abuse.*
We're going to be blunt. The dealership channel is currently being flooded with vendors offering *"AI-generated city pages," "AI-generated model variants,"* and *"1,000 pages a month"* content packages. Many of them violate Google's published spam policy. Some have already triggered manual actions on dealer sites we've audited.
The only AI content workflow that survives scrutiny in 2026 is this:
- ●Use AI to research a topic
- ●Use AI to structure the page
- ●Use AI to draft the first version
- ●Have a human with first-hand automotive expertise rewrite the result so it qualifies as Unique, Specific, and Authentic
That last step is the entire game. Skip it and you're feeding Google's spam classifier. Do it and you have a non-commodity page the AI will reward.
Spam-policy risk: scaled AI-generated content
Google's published spam policy explicitly flags 'using generative AI tools to generate many pages without adding value for users' as scaled content abuse. Vendors selling 'AI-generated city pages' or '1,000 pages a month' content packages are pushing dealers into manual-action territory. We have seen this trigger penalties on real dealer sites in 2026. Use AI to research, structure, and draft — then have a human with first-hand automotive expertise rewrite the result. That last step is non-negotiable.
Key Takeaways
- ✓AI Search is not a separate system — it is Google. It runs on three mechanisms: general knowledge (uninfluenceable), specific knowledge from traditional search results (the layer that matters), and fan-out (one query → constellation of sub-queries pulling from Shopping, Local, Knowledge Graph, Web, Video, and Image surfaces).
- ✓Three myths Google just debunked: you do NOT need to rewrite content with conversational keywords, you do NOT need to chunk content for AI, and third-party AI visibility scores have NO access to Google's internal metrics.
- ✓The single most important concept: commodity vs. non-commodity content. AI rewards content that is Unique, Specific, and Authentic — first-hand expertise no aggregator can replicate. Almost no dealership currently publishes this.
- ✓Page experience and structured data remain foundational. Most dealer sites run 30-50% of available schema correctly — this is the single highest-ROI technical fix in dealership SEO right now.
- ✓Expand beyond the webpage layer: Google Merchant Center, Google Business Profile, video walkarounds with VideoObject schema, and image optimization are all 'new opportunities' Sullivan called out.
- ✓Stop measuring AI Search by traditional vanity metrics — track time on site from AI-referred traffic, VDP engagement depth, and conversion quality instead of raw sessions.
- ✓Scaled AI-generated content (city pages, model variants, '1,000 pages a month') violates Google's spam policy and has already triggered manual actions on dealer sites in 2026.
Founder & President, A3 Brands
Tim spent a decade distributing products to 3,000+ dealerships, ran the Internet Sales department at Baker Automotive Group, and served as Acura's Field Program Manager and Digital Strategist at Shift Digital before founding A3 Brands — the only SEO agency built exclusively for car dealerships.
Frequently Asked Questions
Who is Danny Sullivan and why does what he says matter?
What is fan-out in AI Search and why should a dealership care?
What is the difference between commodity and non-commodity content?
Do I need to rewrite all my pages for 'conversational AI keywords'?
Are third-party AI visibility scores reliable?
Is AI-generated content safe to use on a dealership website?
Sources & References
- Danny Sullivan keynote on AI Search — Director of Google Search, public presentation on how AI is integrated into Google Search (22-slide industry briefing)
- Google Search Central — AI Search guidance — Official Google guidance on content quality, structured data, and AI Overview eligibility
- Google spam policies — scaled content abuse — Official policy on generative AI tools producing low-value pages at scale
- A3 Brands GALAXY framework — Internal AI visibility framework operationalizing Brand DNA, Audience DNA, and Competitor DNA across the dealership channel
Where Does Your Dealership Actually Stand?
Q3 is when OEM co-op cycles refresh and content programs lock in for the back half. We run a complimentary GALAXY AI Visibility Audit aligned to the framework Google just confirmed on stage. No retainer commitment.