Short answer (TL;DR)
On May 15, 2026, Google made it official: optimizing for AI search (GEO or AEO) is still SEO. AI Overviews reaches 2.5 billion people a month, AI Mode 1 billion. Clicks on informational queries are down by as much as half, but visitors arriving from AI answers convert 4 to 5 times better. The new goal is not just position 1, it is being the cited source inside the AI answer.
What happened: three weeks that changed search
May and early June 2026 brought the biggest changes to Google Search since 2015. In order:
May 15, 2026. Google Search Central published its official guide, "Optimizing your website for generative AI features on Google Search." For the first time Google put its position in writing: AEO and GEO are not new disciplines, "this is still SEO." AI Overviews and AI Mode run on the same core ranking systems as classic search. And the llms.txt file that assorted "AI SEO gurus" have been selling for the past year is not something Google needs.
May 19 and 20, 2026, Google I/O. Google announced that AI Mode had passed 1 billion monthly users and AI Overviews 2.5 billion. It also explained the mechanic called query fan-out: AI Mode breaks a single user question into several subqueries, finds the best sources for each one, and cites the best passages in the final answer. This is the key to everything that follows in this article.
May 21 to June 2, 2026. The May 2026 core update. Early analyses showed the hardest hit was generic content with no E-E-A-T signals: articles with no author, no firsthand experience, no original data. Sites publishing AI content at scale with no editorial work lost the most.
June 3, 2026. Search Console got new "Generative AI performance" reports. For the first time you can see AI impressions, meaning how often your page shows up in AI Overviews and AI Mode. Google also added an opt-out if you do not want to appear in AI answers.
What the numbers mean in practice
Here is the trap most businesses fall into. They look at the drop in clicks and conclude SEO is dead. The real picture is different:
| Metric | Classic search | Search with AI answers |
|---|---|---|
| CTR on informational queries | baseline | drop of up to 50% |
| Visitor conversion (Semrush 2026) | 2.8% | 14.2% |
| Visitor intent | researching | already read the answer, arrives decided |
A visitor who clicks through from an AI answer has already read your summary, has already seen the AI recommend you as a source, and lands on your site with far stronger intent. One hundred of those visitors are worth roughly what 500 classic ones are. Traffic falls, the value of each individual visit rises.
The strategic consequence: you are no longer optimizing only for position, you are optimizing for the citation. Your page has to be the source AI Mode pulls a passage from and credits to you.
Old SEO versus AI SEO: what actually changed
The fundamentals hold, the emphasis moves. The difference is easiest to see in concrete goals:
| Aspect | SEO through 2025 | AI SEO in 2026 |
|---|---|---|
| Main goal | position 1 in results | citation in the AI answer |
| Unit of optimization | the whole page | a single passage (H2 plus paragraph) |
| Main metric | clicks and rankings | AI impressions plus conversion |
| Content that wins | length and coverage | concrete, verifiable claims |
| Technical base | speed, mobile, indexing | same plus schema markup everywhere |
Watch out for a common mistake: in the name of "AI optimization," businesses throw out proven fundamentals. A page that does not rank in classic search will not show up in AI answers either, because both systems run on the same ranking signals. AI SEO is an upgrade, not a replacement.
What this means for US small businesses
Small businesses have three advantages right at this moment:
1. Most niches still have thin, structured coverage. Well organized, genuinely useful sources are scarce in plenty of local and specialized topics. Whoever publishes the best answer to "how much does a roof replacement cost" or "how to choose a bookkeeping service" has a real shot at becoming the default cited source for years. 2. Local competitors are still asleep. Most small business sites do not have even basic schema markup, let alone an answer-first structure. The barrier to entry for GEO optimization is far lower than it is in crowded national verticals. 3. Local queries are protected. "Plumber in Austin" or "hair salon in Denver" cannot be resolved by AI without local sources. Google Business Profile and local SEO became more important, not less.
Concrete tactics: how to become a cited source
1. Answer-first structure
Every important page and every article should answer the main question in the first 40 to 60 words. Exactly the way this article is structured: TL;DR at the top, details below. Query fan-out looks for passages that answer a subquery on their own. An answer buried in the seventh paragraph will not get cited.
2. Numbers and citable claims
AI systems cite concrete, verifiable claims. "SEO services cost between $300 and $3,000 per month" is a citable sentence. "Prices vary depending on your needs" is not. Every page should carry at least 3 to 5 claims with a number, a date, or a clear definition.
3. A clean H2/H3 hierarchy
Each H2 should be a question or a topic that the paragraph below it answers completely. AI splits pages by headings. A page with a logical hierarchy gets more chances at a citation than a wall of text, because each section competes for its own subquery.
4. Schema markup
FAQPage, Article, LocalBusiness, Service, Product. Structured data is the cheapest way to tell Google exactly what is on the page. After the May core update, pages with correct schema markup are visibly more present in AI Overviews.
5. NAP consistency and Google Business Profile
Your name, address, and phone number have to be identical on your site, in your Google Business Profile, on Yelp and the Better Business Bureau, and everywhere else. AI answers for local queries are built on these signals. An incomplete or outdated Business Profile means AI skips you when it makes a local recommendation.
6. E-E-A-T: author, experience, proof
In practice, the May 2026 core update was an E-E-A-T filter. Every article should name its author, every service page should show real references and real photos of the work. "10 years of experience, 200+ projects, a result with an actual number attached" is a signal generic AI content cannot fake.
7. Original data and FAQ content
The strongest magnet for citations is data that exists nowhere else: your pricing, your project statistics, your own measurements. A cabinet maker who publishes "the average custom kitchen cost us $7,500 in 2025, calculated across 34 projects" becomes the only possible source for that claim. The second pillar is FAQ pages, because customer questions overlap almost word for word with query fan-out subqueries. Every service page should end with 4 to 6 real customer questions answered in 40 to 80 words each, marked up with FAQPage schema.
8. Presence beyond Google: ChatGPT and Perplexity
ChatGPT Search and Perplexity cite their own source sets, and more and more people start their research there. Check what those systems say about your industry and your business. If they do not mention you, the reasons are the same as with Google: there are no structured, citable sources. The same tactics solve both problems at once.
What you can check yourself in Search Console
Since June 3, 2026, you have new reports in Search Console. Concrete steps for your first 15 minutes:
1. Open Search Console and, under Performance, find the new Generative AI performance report. 2. Check your AI impressions: how often your pages appear in AI Overviews and AI Mode. If the number is zero, your content is not citable and this article is your to-do list. 3. Compare the CTR of pages that appear in AI answers with the ones that do not. That is where you see what you are losing and what you are gaining. 4. Do not turn on the opt-out. Excluding yourself from AI answers in 2026 means excluding yourself from search. The traffic you "protect" simply disappears.
How we help
At NovusAgency we built an [AI SEO service](https://novusagency.co/en/contact) for small businesses: a review of your site's current AI visibility, an answer-first rewrite of your key pages, schema markup, FAQ content, and monthly tracking of AI impressions in Search Console.
We start with a free review: in 30 minutes we show you whether AI Overviews cites you at all, what ChatGPT and Perplexity currently say about you, and which 3 pages can become cited sources for your industry within 90 days.
[Book a free 30-minute consultation](https://novusagency.co/en/contact) and get a concrete AI SEO plan for your site.
Frequently asked questions
Is SEO dead?
No. On May 15, 2026, Google stated explicitly that AI Overviews and AI Mode run on the same ranking systems as classic search. What changed is the goal: instead of only position 1, you are now competing for a citation in the AI answer. The fundamentals (technical health, quality content, E-E-A-T) stayed the same, they are just measured more strictly.
What are GEO and AEO?
GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) are terms for optimizing content so AI systems cite it in their answers. By Google's official position this is still SEO with a different emphasis: answer-first structure, citable claims, structured data.
Do I need llms.txt?
Not for Google. In its official guide in May 2026, Google confirmed it does not use llms.txt. Put that time into schema markup and content structure instead, both of which demonstrably affect whether you show up in AI answers.
How long until results show up?
Similar to classic SEO. The first AI impressions for long-tail questions appear 4 to 8 weeks after you publish answer-first content, and stable citations for your main queries in 3 to 6 months. The advantage right now is thin competition in most local niches: whoever starts today builds a position that will be hard to catch a year from now.