APREX · EST. 2026
> HERO IMAGE PROMPT: A lone journalist sits at a modern desk in a bright Nordic office, staring at a large monitor showing abstract search analytics graphs, natural window light from the left, soft sensor grain, documentary texture, bright Nordic daylight
The search world exploded — and nobody warned you
While you were busy optimizing for Google rankings, the rules of the entire game changed beneath your feet. Google AI Mode surpassed 1 billion users at I/O 2026, and now nine out of ten searches end without the user ever visiting your website. ChatGPT searches the web directly, backed by 900 million users. Perplexity is crossing 100 million monthly users and growing faster than any search engine before it. The traffic you thought was yours now belongs to the algorithms — and there is no guarantee they will forward it to you.
Platform versus platform: the battle in numbers
| Platform | Users (2026) | Zero-click rate | Clicks to sources | Best for |
|---|---|---|---|---|
| Google Traditional | 8+ bn searches/day | ~65 % | Medium | Broad reach, local SEO |
| Google AI Mode | 1 bn users | ~93 % | Low (growing) | Authority building, visibility |
| ChatGPT Search | 900 mn users | Unknown | Moderate | Deep queries, product search |
| Perplexity | 100 mn/month | Low (source-focused) | High | Research, B2B, journalists |
| Google Gemini | Integrated in Workspace | Varies | Varies | Enterprise, vertical search |
| Perplexity Comet / ChatGPT Atlas | Early stage | Unknown | Unknown | Next gen, e-commerce |
Sources: Google I/O 2026, OpenAI product pages, Perplexity.ai, Smartbrief 2026, Search Engine Land 2026
KEYFIGURE
| 93 % | Zero-click rate in Google AI Mode |
| 33 % | Search traffic stolen by AI Overviews in 2025 |
| 900 mn | ChatGPT users now searching the web directly |
Two games. Same field. Completely different rules.
You thought there was one search game. There are now two, and they require radically different approaches.
Game 1: Traditional SEO still revolves around ranking web pages high in organic results. Backlinks, page speed, mobile optimization, keywords — all of this still applies, according to Moz's analysis of AI search versus traditional SEO.
Game 2: GEO — Generative Engine Optimization is something else entirely. Here the goal is to be used as a source inside AI answers. AI engines do not read your pages the way Googlebot does. They extract facts, verify entities and assess credibility against knowledge graphs.
The two games overlap, but winning one does not automatically mean winning the other.
> PULLQUOTE: "If Google Knowledge Graph, Wikidata and your own website describe you differently, AI engines will doubt who you actually are — and choose someone else."
The GEO guide: how to get cited in AI responses
1. Entity clarity is not optional
AI engines use knowledge graphs — Google Knowledge Graph, Wikidata, successors to Freebase — to understand who and what you are. If your brand name is inconsistent across these sources, or if you lack a Wikidata entry, engines will deprioritize you, according to GSQI's GEO analysis. Make sure company name, founding date, product categories and key people are identical everywhere.
2. Structured data is now a prerequisite
Schema.org markup and JSON-LD are no longer a nice-to-have. Sites without structured data are systematically overlooked by answer engines, as shown in Semrush's GEO guide. Implement Organization, Product, Article, FAQ and HowTo schema wherever relevant. This is the language AI engines prefer.
3. Be the primary source — not the summarizer
AI engines prioritize primary sources over aggregators. If you publish original data, your own studies, exclusive content or your own definitions, you are far more likely to be cited than if you simply summarize others' work. Originality is not just an editorial ideal — it is a technical ranking factor in GEO.
4. FAQ sections and definition boxes
This is one of the most underrated tactics in AI search. Concise, well-structured FAQ sections give AI engines quotable text snippets they can extract directly. The same applies to fact boxes and clear definitions. An AI engine is not looking for poetic prose — it is looking for precise answers to precise questions.
5. Build authority broadly, not just on one platform
Mentions in Wikipedia, Wikidata entries, citations in other authoritative sources, references in academic contexts — all of this builds the entity authority that AI engines reward. Google AI Mode, ChatGPT and Perplexity triangulate credibility from many independent signals.
HIGHLIGHT
E-commerce and AI search: the new product visibility
ChatGPT Shopping and Google AI Mode have introduced product search directly inside AI interfaces. That means products now need to be visible in AI responses, not just in traditional search results. Perplexity Comet and ChatGPT Atlas — next-generation AI browsers — are merging search, commerce and navigation into a single interface. For e-commerce operators not ready with structured product data and AI-optimized descriptions, this could mean lost sales they never even know about.
How to measure AI visibility (even though it is hard)
Measuring AI visibility is still in its infancy. Tools like Profound and LLM-visibility trackers are beginning to emerge, but no standardized KPIs exist yet. That is no reason to wait. Start with these manual methods:
- Search for key products and services in ChatGPT, Perplexity and Google AI Mode
- Check whether you are cited and in what context
- Log changes monthly
- Cross-reference with traffic from the referral source "AI tools" in Google Analytics
Those who start measuring now will have an enormous head start when standardized tools arrive.
> BODY IMAGE PROMPT: Close-up of structured data code on a laptop screen in a modern co-working space, warm amber desk lamp casting soft shadows, clean product-editorial clarity, no text visible, soft morning warmth
What about smaller publishers?
This is the darkest chapter. Smaller publishers without resources for technical optimization risk disappearing entirely from AI responses. They have already lost traffic to Google AI Overviews — which in 2025 took 33 percent of all search traffic from traditional results, according to Smartbrief's analysis. Google is now giving something back with new source references and direct links in AI Mode, but the damage has already been inflicted on many outlets.
The EU Digital Markets Act requires Google to share data with competitors, which could weaken the monopoly over time. But that does little for a niche operator who has not implemented JSON-LD today.
TIMELINE: The rapid collapse and rebirth of search
- 2023: ChatGPT launches with browsing. SEO experts warn. Nobody really listens.
- 2024: Google AI Overviews rolls out globally. Traffic numbers begin to fall.
- 2025: AI Overviews steals 33 % of search traffic from traditional results. GEO is established as a concept.
- May 2026: Google AI Mode passes 1 billion users at Google I/O. ChatGPT has 900 million.
- June 2026: Perplexity Comet and ChatGPT Atlas are announced. The AI browser is real.
BOTTOM LINE
You can no longer choose between SEO and GEO — you need both. But the priorities have shifted. Structured data, entity clarity and original authority are no longer nice-to-have. They are the entry ticket to existing in the search landscape of 2026. Traditional Google SEO lives on, but alone it no longer gives you full coverage. Start with entity clarity today. Implement Schema.org tomorrow. Begin measuring AI visibility this week. Those who wait will become invisible.
Verified against 10 open primary sources.